Metrologia

PAPER
Recent citations
Impact of the latest generation of Josephson - Uncertainty analysis for ac–dc difference
voltage standards in ac and dc electric metrology measurements with the AC Josephson
voltage standard
Jason M Underwood

To cite this article: Alain Rüfenacht et al 2018 Metrologia 55 S152

View the article online for updates and enhancements.

This content was downloaded from IP address 190.96.108.72 on 29/04/2019 at 18:23

 Bureau International des Poids et Mesures Metrologia

Metrologia 55 (2018) S152–S173 https://doi.org/10.1088/1681-7575/aad41a

Impact of the latest generation of

Josephson voltage standards in ac and dc
electric metrology
Alain Rüfenacht , Nathan E Flowers-Jacobs and Samuel P Benz
National Institute of Standards and Technology, Boulder, CO 80305, United States of America

E-mail: alain.rufenacht@nist.gov

Received 8 May 2018, revised 7 July 2018

Accepted for publication 18 July 2018
Published 24 August 2018

Abstract
For decades, the quantum behavior of Josephson junctions has been employed as intrinsic
standards for voltage metrology. Conventional dc Josephson voltage standards have been
the primary standards for voltage, programmable Josephson voltage standards have been
implemented in calibration services and precision measurements, such as the Planck constant,
and Josephson arbitrary waveform synthesizers have been employed in ac voltage calibrations
and precision measurements of the Boltzmann constant. With the anticipated redefinition of
the Système International d’Unités, all types of Josephson voltage standards will become
intrinsic standards and equivalent realizations of the unit volt. Here we review the state-
of-the art performance, best practices, and current impact of these systems for various
applications, with an emphasis on ac voltage metrology. We explain the limitations of each
system, especially regarding the many potential systematic errors that affect their accuracy and
performance for specific applications.

Keywords: digital-to-analog conversion, Josephson junction arrays, measurement standards,

signal synthesis, superconducting integrated circuits, voltage measurement, Josephson voltage
standard

(Some figures may appear in colour only in the online journal)

1. Introduction forward was achieved in the mid 1990s with the successful
design and fabrication of large arrays of Josephson junctions
The discovery of the Josephson effect [1] and the subsequent (JJs) with metallic barriers [4, 5], instead of the insulating bar-
development of the first Josephson voltage standards led to rier used with CJVS. The non-hysteretic electrical properties
dramatic improvements in the accuracy of primary dc voltage of the metallic barrier circuits produced quantized voltages
references. The first steps in the evolution of quantum-based that were intrinsically stable and accurate over a large range of
voltage standards were achieved in the 1970s, first with a bias current, thus enhancing their stability to external current
circuit containing a single Josephson junction and then with perturbations.
array circuits that produced increasingly larger dc output volt­ Two complementary types of Josephson voltage standards
ages by successively adding more junctions in series to the used today are the programmable Josephson voltage standard
arrays. By the late 1980s the output voltage of conventional (PJVS) and the Josephson arbitrary waveform synthesizer
Josephson voltage standard (CJVS) array circuits reached (JAWS, also known as the ac Josephson voltage standard or
the practical value of 10 V, starting a broad international dis- ACJVS). The accuracy of all Josephson voltage standard sys-
semination of Josephson voltage standards among National tems, including the CJVS, PJVS, and JAWS systems, relies
Metrology Institutes (NMIs) and many primary calibration upon the ac Josephson effect, in which the voltage across a
laboratories (see [2, 3] for reviews of CJVS technology and junction is precisely proportional to the rate of change of the
early Josephson based voltage standards). The second leap phase difference between the junction’s two superconducting

1681-7575/18/05S152+22$33.00 S152 Not subject to copyright in the USA. Contribution of NIST 

Printed in the UK
Metrologia 55 (2018) S152 A Rüfenacht et al

electrodes. When a junction is biased with a periodic current of sources. In addition, calibration with JAWS standards could
appropriate amplitude and frequency such as continuous micro- potentially reduce overall measurement uncertainty, measure-
waves or periodic pulses, the superconducting phase difference ment duration, and the extensive overhead of maintaining and
responds synchronously, and the junction produces synchro- calibrating numerous transfer standards in the calibration chain.
nized, voltage pulses with quantized area over a range of dc bias In this review article, we present the best practices for
current. The generated voltage pulses are governed by the ac implementing JAWS and PJVS systems in both dc and ac
Josephson equation, which depends only on two fundamental metrology applications. We discuss the advantages and chal-
constants of nature: the Planck constant h and the elementary lenges associated with the change from a ‘detector-based’ to
charge (equal to the absolute value of the electron charge) e. ‘source-based’ ac voltage metrology. Such implementation
With the upcoming redefinition of the Système International will also require periodic verification of the equivalence of the
d’Unités (SI), the value of these two fundamental constants will disseminated Josephson voltage standards.
be fixed [6, 7]. As a result, in 2019 Josephson voltage stand-
ards will become a direct realization of the unit volt, rather than
2.  Josephson voltage standards
a ‘representation of the volt’ based on the values of e and h
adopted in 1990. After the SI redefinition, the realization of the
A properly operating Josephson voltage standard exploits
unit volt can be independently achieved in every calibration lab-
the perfect frequency-to-voltage conversion of the Josephson
oratory, assuming the frequency standard used to lock the bias
effect in junctions formed by a weak link (or barrier) between
current pulse repetition rate of the Josephson voltage standard
two superconductors. The effect that describes this conversion
is linked to an SI-traceable time base.
was discovered by Josephson [1], and the quantized voltages
The first generation of CJVS-type Josephson voltage
produced by a junction when biased at microwave frequencies
standards, with zero-current crossing steps, is still widely used
were first experimentally observed by Shapiro [15].
by many NMIs and is cur­rently the primary voltage standard
When biased with a current pulse of appropriate magnitude
for their dc voltage calibrations. However, over the past few
and width, a Josephson junction generates a voltage pulse with
years a number of CJVS systems have been replaced by PJVS
a quantized time-integrated area exactly equal to the inverse
systems. Besides their intrinsic stability and rapid program-
Josephson constant KJ−1  =  h/2e (the magnetic flux quantum),
mability, PJVS systems have the additional capability of gen-
erating programmable ac reference waveforms from dc to a that is, the ratio of two fundamental constants. In PJVS sys-
few kilohertz for use in ac voltage and power metrology appli- tems, the combination of an ac current bias with frequency f
cations. Direct dc comparison between PJVS and CJVS sys- (typically  >10 GHz) and a dc current bias results in a signed
tems have demonstrated agreement of 1 part in 1010 or better integer number n of quantized voltage pulses per ac period
and produces a dc voltage V given by
at 10 V [8, 9] and direct comparison of two PJVS systems
has demonstrated agreement better than 1 part in 1011 [10]. V = n × M × KJ−1 × f ,
Thus, the replacement of CJVS systems with PJVS systems
will have no impact on reported calibration accuracy. where M is the number of JJs in series. In JAWS systems, the
Quantum voltage standards are, in principle, intrinsically JJs are biased by a fast pulse generator that controls the pulse
accurate. However, this statement is valid only over a finite pattern to generate a low-frequency waveform with the output
range of bias and environmental parameters. The quantum voltage determined precisely by the known pulse sequence.
locking ranges1 of both PJVS and JAWS systems must be Until May 2019, all Josephson voltage standards will
periodically verified during regular calibration procedures to continue to represent the unit volt by use of the value of
ensure they are performing as quantum standards with their KJ defined in 1990 (KJ-90  =  483 597.9 GHz V−1) [16]. After
expected accuracy. An understanding of, and routine checking the redefinition, the new value will be
for, all systematic errors (e.g. the effect of leakage current KJ = 2e/h = 483 597.848 416 984. . . GHz V−1 ,
and inductive errors for the JAWS), and implementation best
practices for each measurement setup, are critical for suc- based on the following exact values of e and h [17]:
cessful operation of Josephson voltage standards in metrology e = 1.602 176 634 × 10−19 C, and
applications. h = 6.626 070 15 × 10−34 J · s.
Currently, traceability for ac voltage is based exclusively
on ac–dc thermal voltage converters (TVCs) [11]. Unlike dc Note that with the transition to the new SI value of KJ, all
voltage metrology that relies on accurate reference sources, the secondary voltage standards will see a discontinuity of  −1.07
use of TVCs for ac metrology is a ‘detector-based’ approach. parts in 107 in their calibration values.
In general, measurements with TVCs are time-consuming and
require multiple detectors to cover the voltage and frequency 2.1.  Programmable Josephson voltage standards
domain. Recent increases in the output voltage of JAWS stand-
ards [12–14] could greatly simplify the realization of accurate A PJVS system programs the output voltage by biasing various
ac voltages and will enable direct comparisons of commercial segments of the series-connected Josephson array circuit with
1 various predefined dc bias currents (see [18, 19] for details).
The quantum locking range is the range of bias over which every Joseph-
son junction in an array produces a single voltage pulse per input bias pulse To achieve proper operation, this programmable array circuit
(see section 3.1). requires long arrays of uniform JJs with uniformly applied

S153
Metrologia 55 (2018) S152 A Rüfenacht et al

normal metal barrier/superconducting (SNS) junctions made

of Nb/NbxSi1−x/Nb [23]. The National Metrology Institute
of Japan (NMIJ) and the National Institute of Advanced
Industrial Science and Technology (AIST) have developed
NbN/TiNx/NbN SNS junctions for their PJVS circuits, which
can be operated at higher temperatures (up to 11 K) compared
to niobium-based junctions (typically  <5 K). Although opera-
tion at higher temperatures allows use of smaller cryogenic
coolers, the fabrication process for NbN-based junctions,
requiring epitaxial growth, is more challenging.
NMIJ/AIST, PTB, and NIST have all developed PJVS
circuits capable of reaching dc output voltages of at least
10 V. The NIST 10 V PJVS circuit has 265 116 JJs biased at a
Figure 1. Typical V–I curve characteristic of a 16 800 JJ segment
(MSB) measured on a 10 V PJVS circuit. The voltage is accurately selectable frequency between 18.3 GHz and 22 GHz [18, 24].
defined on the n  =  0, n  =  +1, and n  =  −1 constant voltage steps, The junctions are distributed in 32 parallel coplanar wave-
but only over the bias current range shown in the grayed areas. The guides (CPWs), and uniform microwave power is delivered
optimum bias value of each step (0 mA, +9.3 mA and  −9.3 mA for to each of the 32 CPWs by use of Wilkinson dividers made
n  =  0, +1, and  −1, respectively) is indicated by a blue vertical line. with lumped-element superconducting circuits [25]. The
The n  =  ±2 constant voltage steps are visible, but the bias current
range is too small to be reliably operated (<1 mA). Wilkinson dividers suppress microwave reflections at each
branch that can negatively affect microwave power uniformity
microwave bias along the whole array. Compared to CJVS along the array and circuit performance. The NIST circuits
systems, the main advantages of PJVS systems are the large also use impedance tapering of the CPWs, typically from 50 Ω
milliampere operating current range, the rapid programma- to ~30 Ω, which allows a more uniform distribution of micro-
bility, and the intrinsic stability of the selected output voltage. wave current across all JJs in a given CPW [26]. To make the
A disadvantage of PJVS systems is that the electronics pro- circuits more compact, the junctions are stacked vertically in
viding the dc bias currents are always connected to the PJVS groups of three [27], as shown in the scanning electron micro-
circuit, resulting in multiple leakage current paths to ground scope (SEM) image in figure  2. Such vertical integration is
(Earth). The PJVS bias electronics and circuit wiring must be required to reach a density of more than a quarter million JJs
properly designed to minimize leakage current contributions over a 12 mm  ×  17 mm surface area.
so they do not cause significant systematic errors. The PTB circuit design is based on 128 parallel micro-strip
To achieve the best voltage resolution, a PJVS circuit is microwave transmission lines and is optimized for frequencies
subdivided into segments or subarrays, each containing a defi- of around 70 GHz. Its main advantage is that it requires only
nite number of JJs. Initially the number of JJs for the shorter 69 632 single-stacked JJs to reach 10 V [28, 29]. The same
segments were distributed in a binary sequence (multiple of: circuit design is implemented with the 10 V PJVS system
1, 2, 4, 8, 16, …) [20, 21]. However, to take full advantage of commercially available from Supracon AG2. PTB has realized
the three bias states (n  =  0, n  =  +1 and n  =  −1) of the PJVS a PJVS circuit with an output voltage of 20 V by adding an
voltage–current (V–I) characteristic (figure 1), a distribution additional JJ in a vertical stack (double stack) to their single-
of the JJs in a ternary sequence (multiple of: 1, 3, 9, 27, 81, stacked 10 V circuit design, effectively doubling the number
…) is a much more efficient option that minimizes the number of JJs in the circuit [30].
of bias channels without compromising the resolution [22]. With 524 288 JJs, the NMIJ/AIST circuit is at present
This method can easily be extended to a quinary sequence of the most complex PJVS device to reach a voltage of at least
JJs (multiple of: 1, 5, 25, 125, 625, …) if future performance 10 V (figure 3). Operated at 16 GHz, it can achieve an output
of PJVS circuits allows simultaneous operation of the five voltage of 17 V when all segments are biased in the n  =  +1
steps n  =  +2, +1, 0, −1 and  −2. state. By containing redundant JJs, this design can accom-
The segment with the smallest number of JJs is called the modate circuits with minor defects, thereby increasing the
least significant bit (LSB). Sometimes more than one large number of NbN-based circuits capable of reaching 10 V [31,
segment or most significant bit (MSB) is required in the 32]. However, achieving high-yield fabrication processes for
PJVS circuit to maximize circuit performance and to attain PJVS circuits having several hundred thousand JJs is a chal-
the largest output voltages. Because their values depend on lenging task [33, 34]. Double-stacked JJs are incorporated in
temper­ature, applied microwave power and frequency, the bias the 17 V design, and the microwave distribution uses a 64-way
current range of all subarrays must be experimentally meas- splitter. NMIJ/AIST is currently developing a new design of
ured with a digital voltmeter (DVM). The circuit performance ‘serial-parallel’ power dividers [35] for future Josephson
is optimized by selecting the midpoint of the range for the voltage standard circuits.
bias values for each voltage of each segment (these values are
2
typically collected in a current-bias or current-margins table). Certain commercial equipment, instruments, or materials are identified
The present JJ technology used at the National Institute in this paper to facilitate understanding. Such identification does not imply
recommendation or endorsement by NIST, nor does it imply that the mat­
of Standards and Technology (NIST) and the Physikalisch– erials or equipment that are identified are necessarily the best available for
Technische Bundesanstalt (PTB) is based on superconducting/ the purpose.

S154
Metrologia 55 (2018) S152 A Rüfenacht et al

Figure 2.  (Left) SEM image of a NIST 10 V PJVS circuit in cross section along propagation direction of the CPW center conductor. Two
separate triple-JJ stacks can be seen. (Right) photograph of crypopackage developed for the NIST 10 V PJVS circuit. The chip size is
12 mm  ×  17 mm. Images courtesy of NIST Boulder precision imaging facility (SEM) and Dan Schmidt (cryopackage).

A summary of the different types of PJVS circuits cur­

rently being fabricated and their main characteristics are pre-
sented in table 1.
The initial motivation behind the PJVS was to develop
a quantum-accurate rms source [41], based on the calcu-
lated voltage levels of a stepwise-approximated waveform.
Unfortunately, the ~1 µs switching duration (transients
between steps) of the bias electronics developed in the mid
1990s limited the relative accuracy to 1 part in 105 at 1 kHz.
Newer, faster electronics reduced the switching time and dura-
tion of the transients by more than two orders of magnitude, but
realization of an accurate rms waveform synthesis with PJVS
circuits was prevented by intrinsic systematic errors associ-
ated with the transients (see next section). However, develop-
ment of measurement methods that eliminate the transients,
namely those that exploit only the quantum-accurate portion
of the steps in the PJVS stepwise waveforms, have opened
numerous applications in ac voltage metrology for frequen-
cies up to a few kilohertz. With programmable dc voltage and
ac waveform capabilities, PJVS systems now largely surpass
Figure 3.  Photograph of 17 V NMIJ/AIST PJVS circuit including CJVS systems in both performance (i.e. intrinsic stability) and
524 288 NbN/TiN/NbN JJs. The chip size is 15.28 mm  ×  14.70 mm.
Image courtesy of Hirotake Yamamori [31, 32]. range of applications (see section 4). NMIJ/AIST, NIST, and
Supracon, have successfully operated PJVS circuits on liquid-
Operation of a 17 V or 20 V PJVS circuit requires bias helium-free dry cryocoolers [42–44]. Fully automated, liquid-
electronics designed with twice the output voltage compliance cryogen-free PJVS systems are available from NIST through
compared to 10 V circuits. The main secondary standards that the NIST standard reference instrument program3 and from
are typically calibrated are the Zener dc references, which are Supracon4. As turn-key systems, these cryocooled PJVS sys-
calibrated at 10 V [36, 37]. So, a Josephson standard with an tems do not require an expert operator or cryogenic handling
output voltage greater than 10 V [30, 32] has limited applica- skills to be successfully and reliably operated. The ease of
tion for dc voltage calibrations. use of these systems extends the range of dissemination of
Two other laboratories have developed PJVS circuits that PJVS from primarily NMIs to a larger number of calibration
have demonstrated lower voltages with alternative fabrication. laboratories.
The National Institute of Metrological Research (INRIM) in
Italy, in collaboration with PTB, has developed a 1 V PJVS
2.2.  Josephson arbitrary waveform synthesizer
circuit based on superconducting-normal-insulating-super-
conducting (SNIS) junctions [38]. The Institute of Electronic A JAWS system generates accurate output voltages by control-
Measurements (KVARZ) in collaboration with the Institute ling the timing, presence, and polarity of the quantized pulses
for Physics of Microstructure from the Russian Academy of created by the JJs [45]. Like the PJVS system, the JAWS
Science was the first laboratory to successfully implement a circuit requires long homogeneous arrays of JJs to generate
voltage standard made of a high-temperature superconductor useful voltages. Unlike the PJVS approach, in this case the JJ
(HTS). The KVARZ HTS circuit has 161 JJs made of grain array is biased by a pulse generator so that during each clock
boundaries between two YBa2Cu3O7 layers [39, 40]. The
output voltage of this HTS-based circuit reaches about 25 mV 3
NIST standard reference instruments: www.nist.gov/sri/standard-reference-
and has the main advantage that it can operate at 77 K (i.e. instruments/sri-6000-series-programmable-josephson-voltage-standard-pjvs.
liquid nitrogen instead of liquid helium temperatures). 4
Supracon AG: www.supracon.com/en/standards.html.

S155
Metrologia 55 (2018) S152 A Rüfenacht et al

Table 1.  Summary of PJVS circuits currently in fabrication. The circuits (columns) are sorted by bias frequency.

NMIJ/ PTB/Supracon
AIST [32] NIST [24] [29] PTB [30] INRIM/PTB [38] KVARZ [40]
Voltage 17.3 V 10 V 10 V 20 V 1.19 V 25 mV
# of JJs 524 288 265 116 69 632 139 264 8 192 161
Bias Frequency 16 GHz 18.3 GHz 70 GHz 70 GHz 70 GHz 75 GHz
Junction SNS SNS SNS SNS SNIS HTS grain boundary
Material NbN/ Nb/NbxSi1−x/Nb Nb/NbxSi1−x/Nb Nb/NbxSi1−x/Nb Nb/Al-AlOx/Nb YBa2Cu3O7 bicrystal
TiNx/NbN
Temperature <11 K <5 K <5 K <5 K <5 K <77 K
# of JJs/Stack 2 3 1 2 1 —
# of JJs in LSB 128 6 1 2 — —
LSB resolution 4.235 mV 227 µV 145 µV 270 µV — —

Figure 4.  Calculated spectrum from dc to 50 MHz of a 1 kHz sine wave with rms magnitude of 1 V, based on 12 810 JJs and a clock rate of
14.4  ×  109 pulses s−1.

cycle, each JJ in the array generates either a positive pulse, a magnitude and phase of all frequency components are deter-
negative pulse, or no pulse. A repeating sequence of JJ voltage mined from a Fourier transform of the pulse pattern. Typically,
pulses has a voltage spectrum with a calculable, quantum- both the pulse pattern and the calculated value of the relevant
accurate magnitude and phase. The pulse sequence is typically frequency components are stored for later use.
determined using a delta-sigma modulation algorithm [46, Two significant differences between other typical uses of
47]. In practice, a pure, 1 kHz ac voltage waveform (figure 4) the sigma-delta algorithm and how it is used in JAWS sys-
can be created with a calculable spurious-free dynamic range tems are: the need for high pulse density, and the use of pre-
(SFDR) greater than 200 dBc for frequencies up to 100 kHz generated patterns. The maximum output voltage of the JAWS
[14] due to the large oversampling ratio between the clock system is proportional to the maximum pulse density, which is
rate (typically 14.4  ×  109 pulses s−1) and the frequency of the typically taken to be between 90% and 95% (with 100% corre­
desired waveform (typically in the audio range). sponding to a pulse at every clock cycle). Operation at high
As applied to JAWS circuits, the idea behind the sigma- density reduces the stability of the delta-sigma feedback loop
delta algorithm is to step through the desired waveform as and can also result in excess digitization noise. In general, a
a function of time at the pulse rate and decide at each step pulse density above 50% would be considered ‘overloaded’
(sampling period) whether the agreement between the desired by the standard definition for a second-order algorithm [46,
waveform and the programmed waveform is better with a pos- 47]. However, standard delta-sigma algorithms are often
itive pulse, negative pulse, or no pulse. The decision is made intended to be run in real time and accept or generate arbitrary
using the output of a feedback loop whose input is the dif- waveforms as part of an analog-to-digital converter (ADC) or
ference between the desired waveform and the pulse pattern digital-to-analog converter (DAC), respectively. Because the
at that time step, that is, the ‘delta’ in the algorithm name. JAWS waveforms are typically pre-generated, the delta-sigma
The structure of the feedback loop can be more sophisticated algorithm settings can be tuned for a given output voltage,
with multiple loops but typically involves the summation of making the overload condition less important. This must be
past errors, that is, the ‘sigma’ in the algorithm name [46, 47]. confirmed by checking that the frequency content of the pat-
Typically, feedback loops used to calculate JAWS patterns for tern after generation produces no distortions.
output voltages at frequencies  <1 MHz contain two stages of The JAWS systems developed by both NIST and PTB use
integration and a low-pass filter. The net effect of this filter the same basic type of JJs as they use in their PJVS systems
and algorithm is to move the digitization noise caused by the (described above), though the JJ properties are optimized for
fast pulses away from the frequencies of interest (in this case, operation at around 20 GHz [23, 48]. The maximum output
to  >10 MHz), leaving a pure tone with a small background voltage and the number of JJs used in the JAWS systems is
(see figure 4). After the entire pulse pattern is generated, the smaller than in the PJVS systems. PTB has recently created
S156
Metrologia 55 (2018) S152 A Rüfenacht et al

Figure 5.  (Left) NIST JAWS circuit diagram and (right) photograph of a cryopackaged JAWS chip. The circuit can generate an output
voltage of 1 V rms. The circuit includes two pulse generator channels (green, labeled ‘D+’ and ‘D−’), a single layer of Wilkinson dividers
(pink), inside-outside dc blocks (yellow), and JJ arrays (red ‘X’s). The JJ arrays are connected in series through inductive taps, as is one
floating low-frequency current compensation per JJ array (brown, ‘IB’). The JAWS chip size is 10 mm  ×  10 mm.

waveforms with an rms output voltage of 1 V by summing resistive structures for microwave termination of the JJ arrays
the signals from eight JJ arrays operated on four separate to avoid unwanted reflection.
cryopackages with a combined total of 63 000 JJs [12, 49]. Current PTB [12, 49] and early NIST JAWS circuit designs
NIST has recently summed two cryopackages with a total of [64, 65] did not use explicit on-chip microwave power dis-
102 480 JJs to create waveforms with an rms output voltage of tribution elements (figure 6); instead, each JJ array was con-
2 V [13]. Many of the circuits fabricated by NIST [50–52] and nected to the pulse generator through separate high-frequency,
PTB [53, 54] are used by other NMIs around the world. semi-rigid cables, connectors, and room temperature comp­
In comparison to PJVS circuits, the smaller output voltage onents. This direct connection was necessary to maximize the
and number of junctions used in JAWS systems are directly input bandwidth for the pulse waveform bias and minimize
related to the added complexity of the microwave pulse bias. reflections in the pulse transmission path. This direct connec-
Instead of a single-frequency bias, the JAWS pulse-pattern tion means that each JJ array was biased by a separate channel
bias typically contains significant power at frequencies from of the pulse generator. This approach had the advantage that
1 GHz to  >30 GHz [14]. Therefore, the JAWS system is the bias settings of each channel could be optimized to match
even more sensitive to the non-ideal behavior (e.g. non-linear the properties of each JJ array, but had the disadvantages that
response) of the circuit elements and the quality of the high- pulse generator channels are expensive. Every microwave
speed pulse bias source and amplifiers. To combat this sen- cable also increases system complexity and heat dissipation at
sitivity, JAWS circuits have adapted many of the elements cryogenic temperatures.
developed for PJVS circuits, as shown in figure 5, and have Recent NIST JAWS circuits have adopted the on-chip
also added many parameters and pulse generation methods to power division approach used in PJVS and CJVS circuits by
optimize the pulse bias [14, 55–62]. adding on-chip Wilkinson dividers, so that each pulse gen-
Like PJVS systems, JAWS circuits use JJs embedded in erator channel and microwave line can bias multiple JJ arrays
tapered CPW waveguides, and special emphasis is placed in [13, 14]. This drastically reduces system expense and com-
the fabrication process on uniformity of electrical properties plexity but also places more stringent constraints on the uni-
of the JJs [26]. These design and process features compensate formity across the entire chip for both the frequency response
for loss in the CPWs, so that the optimal input current pulse is of the on-chip microwave elements and the electrical proper-
similar for JJs at the beginning and end of each array. JAWS ties of the JJs. That is, by biasing multiple JAWS arrays with
circuits use the same low-pass superconducting inductive taps a common pulse generator channel, one cannot independently
as used in PJVS circuits to extract the output voltage across optimize the pulse bias to each array. The removal of optim­
each JJ array [63]; the inductance prevents the taps from dis- ization parameters also typically reduces the range of bias
turbing the high-frequency bias, while also having minimal parameters over which the system is ‘quantum locked,’ where
effect at low frequencies. Finally, both the PJVS and JAWS one JJ output pulse is generated for every input pulse (see sec-
systems use similar, though not always identical, on-chip tion 3.1 for details). This reduction in QLR makes the system

S157
Metrologia 55 (2018) S152 A Rüfenacht et al

mode signals. The dc blocks are high-pass filters that shift the
offset of each pulse based on the recent pulse-density history.
This changing offset narrows the bias range over which the
JJ arrays remain quantum-locked and the JAWS circuit pro-
duces accurate voltage waveforms. The current source ‘com-
pensates’ for the shift by reintroducing the components of the
spectrum removed by the filter. However, the compensation
current is also a source of systematic error [62, 67, 68], as
described below in section 3.2.2.
A separate solution to the presence of a high-pass filter is to
change the shape of each pulse by removing the low-frequency
content [61, 62, 66]. This is often called a ‘zero-compensation’
Figure 6.  Photograph of JAWS circuit with two arrays of 6000 JJs method. In the simplest version, a half-pulse of opposite sign
fabricated by PTB. The chip size is 10 mm  ×  10 mm. Four sample
is applied both before and after the conventional pulse. The
holders and eight arrays are combined in series to reach 1 V rms
(see [12] for details). Image courtesy of Johannes Kohlmann and spectrum of this ‘pulse-like’ object has very little power at the
Oliver Kieler. frequencies affected by the high-pass filter (i.e. inner–outer dc
block), and therefore the filter has minimal effect on the pulse
more difficult to operate correctly and can affect stability. shape. After optimization, the non-linearity of the JJ response
NIST’s recent use of a finite-impulse-response (FIR) filter on to the pulse-like object results in a single voltage pulse. Using
the pulse generator outputs allows an improvement in pulse this zero-compensation pulse-bias method both reduces the
shape uniformity, which partially compensates for the added complexity of the electronics and removes sources of error
circuit components and circuit complexity [14, 66]. in the output voltage. The main disadvantage of the zero-
The use of Wilkinson dividers on the NIST PJVS and JAWS compensation method is that it reduces the maximum output
circuits requires the implementation of an additional passive voltage [66]. Each pulse-like group is spread over a longer
on-chip microwave component: an inside-outside dc block time interval, thus reducing the maximum pulse repetition rate
between each JJ array input and Wilkinson divider outputs. and the output voltage. In practice, there is usually a trade-
This takes the form of capacitive breaks on the grounds and off (that depends on synthesis frequency) between the need
inner conductor of the CPW. This low-frequency isolation of for higher voltage or improved signal-to-noise ratio and the
the individual JJ arrays from the pulse generator is required so need to remove frequency-dependent sources of error. Larger
that the JJ arrays can be summed in series to generate larger, synthesized amplitudes at lower frequencies are typically syn-
quantum-accurate voltages. The PTB JAWS system and ear- thesized with compensated biases, whereas at higher frequen-
lier NIST JAWS systems use room temperature inside-outside cies (>100 kHz) where systematic errors from compensation
dc blocks and maintain that isolation from room temperature become significant, zero-compensated bias methods are used
to each JJ array. to produce lower-amplitude signals. The zero-compensation
The typical room-temperature equipment used by a JAWS waveforms can also be directly compared to compensated
system consists of pulse generators, broadband microwave waveforms as a function of frequency at lower voltages to
amplifiers, and low-frequency, isolated current sources [58– better understand and characterize the systematic errors pre-
61]. The generator’s pulse repetition rate f (clock) limits the sent in the higher-voltage compensated waveforms.
maximum output voltage V  =  M × f × h/2e for M JJs, though Unlike the CJVS and PJVS systems, JAWS systems have
typically other circuit components and the JJ properties also only recently begun to make the transition from being research
limit the maximum repetition rate. The generator’s memory tools towards becoming disseminated standards. This change
size, which determines the maximum duration of the synthe- is motivated by having reached an rms output voltage of 1 V,
sized waveform, places a lower bound on the output voltage where the JAWS systems can begin to play a larger role in
frequency. In some cases, the output amplitude of a pulse thermal convertor calibrations. An automated, liquid-cryogen-
generator channel is sufficient to bias a JJ array. However, in free JAWS system is now available from NIST through the
other cases, especially when dividers are used to bias multiple NIST standard reference instrument program5.
arrays, the pulses require additional amplification. The ampli- Over the past 30 years there has also been progress in
fier must meet stringent requirements of large output current, making digital circuits using single flux quanta (SFQ): using
wide bandwidth, and small gain ripple. This is particularly true JJs as active logical elements, instead of transistors [69]. One
when driving arrays through Wilkinson dividers; at least 3 dB application of these circuits is to create a JAWS system in
more power is required for each layer of Wilkinson dividers. which the JJ array and the room-temperature pulse generator
Finally, achieving the highest output voltage typically are replaced by an array of superconducting quantum inter-
requires biasing each JJ array by a low-frequency, isolated, ference devices (SQUIDs) biased by on-chip logic [70, 71].
current source with the same shape as the desired JAWS output To date, these systems have only found use in applications
voltage waveform, often called a ‘compensation’ current. involving very small voltages [72, 73].
This requirement is a side effect of the need for inner–outer
dc blocks to isolate the JJ arrays from the pulse generators 5
NIST standard reference instruments: www.nist.gov/sri/standard-reference-
in order to connect multiple arrays in series without common instruments/sri-6011-josephson-arbitrary-waveform-synthesizer.

S158
Metrologia 55 (2018) S152 A Rüfenacht et al

Figure 7.  Two-dimensional quantum-locking range of a 10 V PJVS

array with the subarrays biased such that exactly the same number Figure 8.  Graphical visualization of the quantum locking range
of junctions are at  +5 V and  −5 V, producing a null voltage. The of a JAWS system generating a 1 kHz waveform with an rms
horizontal (x) axis is the dither current applied through all the magnitude of 2 V (data from Flowers-Jacobs et al [13]). In the
subarrays, while the vertical ( y ) axis shows microwave power offset graph, the color at each point indicates the voltage residual of a
in dB relative to initial (optimum) power. The white area indicates the sine-wave fit as a function of dither offset current (y axis) and
QLR where the output voltage remains constant. The measurement waveform period (x axis). The data was digitized over 80 ms with
was performed on the 1 mV range of the digital nanovoltmeter. a National Instruments PXI-5922 digitizer (range 10 V, input
impedance 1 MΩ, sampling rate 1 MHz). The blue lines show the
3.  Josephson quantum locking range and 1.6 mA quantum locking range of this waveform and system, that
is, where the residuals are limited by digitizer distortion and do
limitations not depend on offset current over the entire waveform period. This
visualization technique highlights the pulse sequences (e.g. at 0.25,
3.1.  Quantum locking range 0.6, and 0.9 parts of a period) that limit the QLR.
A Josephson voltage standard (JVS)6 is operated within its its remains constant. The bias parameter most commonly tested
quantum locking range (QLR)7 if every JJ produces a single is a dc offset current (also called dither current or trim cur­
voltage pulse (n  =  ±1) per input bias pulse. When biased on rent), which is applied to every subarray. In figure 7, two PJVS
the QLR, the output voltage remains constant over a significant parameters, current offset and microwave power, have been
adjustable range around the step-centered bias parameters. A varied from their initial (and optimal) bias setpoint while the
JVS can be in one of two states: either it is operated within a output voltage, which was set to 5 V  +  (−5 V)  =  0 V, is meas-
well-characterized QLR, and the output voltage generated by the ured. The QLR is defined as the area or the range in which the
JJ array is precisely known; or it does not have a QLR, and the voltage remains constant (white region at 0 V in figure 7). In
output voltage is undefined. In the second case, the JVS cannot this example, the microwave power can be reduced by more
serve as a metrology reference. Measuring only the stability of than 4 dB while still maintaining perfect voltage quantization,
the generated voltage over a long period of time, possibly with thus validating the operating parameters initially selected
low uncertainty, is not a sufficient criterion to test the quantum from the margin table of bias setpoints for all subarrays.
accuracy of voltage standard; it simply tests the stability of the Because the JAWS waveform is periodic in time, the QLR
system biases and environment. If biased slightly outside its is generally verified with a digitizer or a high-resolution oscil-
QLR the output voltage may remain stable (and probably very loscope. One method to determine the QLR is to add a low-
close to the ideal value) but will provide inaccurate voltages. frequency triangular sweep to the bias current compensation
For maximum confidence in output voltage accuracy, the JVS signal and measure the time-dependent distortion of the JAWS
should be operated with the largest QLR achievable; ideally, the waveform. NIST has developed a graphical method (figure 8)
output should be confirmed during calibration measurements. for displaying QLR [13], where the ideal synthesized wave-
In principle, a JVS system should have a built-in procedure form is subtracted from the measured signal, and only the
to self-check its QLR either manually or automatically. With voltage residuals (distortion) are shown in a surface density
the NIST PJVS system, the QLR is confirmed routinely with plot. The vertical axis shows the dc offset current magnitude,
a commercial digital nanovoltmeter. To reach the maximum while the horizontal axis represents one period of the JAWS
sensitivity of the nanovoltmeter, the subarrays are biased in signal. The central speckled band (white, light blue and light
series and with opposite sign to obtain zero net dc output red) shows the QLR where the residuals are minimal. By
voltage [18]. In this configuration, one or more parameters monitoring the residual measurement while varying other bias
are dithered to determine the limits of the operating range parameter, the QLR area can be seen to shrink, thus deter-
(also known as the operating margin) and whether the voltage mining the QLR for that parameter8.
6
‘JVS’ refers to both PJVS and JAWS systems described in this manuscript. 8
For a video example of the QLR display method see: www.nist.gov/pml/
7
‘QLR’ was previously referred to as ‘flat spot’ or ‘margins range’ in the quantum-electromagnetics/superconductive-electronics/quantum-locking-
literature. ranges.

S159
Metrologia 55 (2018) S152 A Rüfenacht et al

QLR verifications are primarily performed on JVS systems

through automated software, where the voltage output leads
are connected only to the nanovoltmeter (for PJVS) or digi-
tizer (for JAWS). When a JVS is imbedded in a calibration or
a more complex measurement circuit, then the initial QLR
verification may not be easily accessible and cannot be per-
formed automatically. In more complex measurements, addi-
tional factors such as the presence of current noise induced by
undesired ground loops can reduce the QLR. In a worst-case
scenario, these unaccounted perturbations could jeopardize
the quantum accuracy of the measurement. One way to verify
the quantum accuracy of the JVS reference without disturbing
the measurement circuit is to repeat the measurement multiple Figure 9.  Relative change of rms voltage measured with a thermal
times while successively applying an offset to one of the bias transfer standard for a 0.4 V stepwise-approximated sine wave
parameters, with the bias offset magnitude not exceeding the versus applied dither current. The voltage value at 0 mA is used
QLR. For example, the automated measurement sequence of a as the reference for the voltage difference ΔV. The slope observed
within the  ±0.7 mA dither range is due to transients present in the
direct (dc) comparison between two PJVS systems can be pro- waveform; the QLR of the PJVS array when generating dc voltages
grammed with three levels of offset current: 0 mA, +0.2 mA is  ±0.7 mA.
and  −0.2 mA [10, 74]. During data analysis, if the voltage
difference between the two PJVS systems is not correlated the voltage output leads and the temperature gradient between
with the bias parameter change, then the quantum accuracy the PJVS circuit at ~4 K and instruments connected at room
(or QLR over the tested range) is confirmed for the entire temperature. Thermal EMF errors can easily be eliminated by
duration of the measurement. A similar verification approach performing polarity reversal measurements. Leakage current
is built into the NIST PJVS-Zener dc reference calibration errors can be limited by properly engineering the JVS bias
measurement. Application of a similar verification method is electronics and by using a material with high isolation resist­
strongly encouraged for all types of measurement involving ance for all leads connected to the PJVS circuit.
any JVS, and especially with the implementation of a PJVS Direct synthesis of PJVS ac waveforms is undesirable for
with a Kibble balance or Joule balance [75–80]. rms calibration of devices under test (DUT) such as ac volt-
meters, TVCs and thermal transfer standards (TTSs). The
finite rise time of the current bias-electronics and the stepwise
3.2. Limitations
nature of the waveforms introduce multiple non-negligible
Obtaining a well-defined QLR is the first validation test that systematic errors. Development of faster switching bias elec-
should be completed before performing any calibration mea- tronics [82] or techniques to drive low impedances [83, 84]
surement with a JVS. However, other factors may still limit do not eliminate this inherent problem. The rms value of a
the accuracy of the JVS. For every JVS, the microwave or PJVS step-wise approximated output waveform is never fun-
pulse-bias electronics must be locked to a reference frequency damentally accurate; a small change in one of the PJVS bias
of high accuracy and traceable to the SI. As discussed in sec- signals, such as the bias current to one subarray or the applied
tion 2, a JVS is a direct frequency-to-voltage converter. This microwave power, has an influence on the rms output voltage.
means that the relative uncertainty in the frequency reference For a review of this error mechanism, see Burroughs et al [85,
will be transferred with the same magnitude to the voltage 86]. Many measurements in the literature have been subject
uncertainty. Today, compact, commercially-available fre- to such transient problems [83, 84, 87–98]. As an example,
quency standards can routinely achieve an absolute frequency figure 9 shows the absence of accurate voltage over the range
accuracy better than one part in 1011. Frequency standards can we expect to observe a QLR (data from Burroughs et al [85]).
easily be made traceable to the SI anywhere on Earth with the The plot shows the effect of current bias offset on the rms
use of a global positioning system disciplined oscillator [81]. voltage measured by a thermal transfer standard operated with
The second limitation, also common to all JVSs, is the a high input impedance. Due to transient effects, the magni-
potential for the JJs to source current when the output of the tude of the slope varies almost linearly with frequency [85].
JVS circuit is connected to a low-impedance load. Any cur­ PJVS stepwise-approximated waveforms contain large num-
rent flowing in the voltage output leads will generate a voltage bers of harmonics [89], with the magnitude of the harmonics
error due to the finite resistance of the wires. Driving excessive (including the fundamental) affected by the transients [99].
current out of the JVS device may also exceed the operating The use of a two-level PJVS waveform is one way to slightly
range of the JJs, such that the resulting loss of quantization minimize the effect of transients on the fundamental [100],
will destroy the accuracy of the JVS voltage. but it has the undesirable consequence of generating a large
quantity of large-amplitude digitization harmonics. The use
3.2.1.  PJVS limitations.  When a PJVS is operated to generate of two-level PJVS waveforms has been implemented for fast
dc voltages, the two main error contributors are the thermal reverse dc measurements [101, 102] and was suggested for
electromotive forces (EMF) and voltages generated from leak- impedance ratio measurements based on two PJVS systems
age currents. These errors are due to the finite resistance of [90, 91, 93, 95–98].

S160
Metrologia 55 (2018) S152 A Rüfenacht et al

Figure 10.  In-phase (green) and quadrature (blue) magnitudes of the residual difference between two NIST JAWS systems, each generating
a 1 kHz output with an rms magnitude of 1 V, versus different bias parameters. The standard operating point (vertical red dotted line) is
the bias setting that maximizes the QLR for dc current offset. (A) The residual magnitude is independent of the pulse magnitude (within
the QLR from about 1100 to 1550) but in depends strongly on (B) compensation phase and (C) compensation magnitude. The phase is
expressed in terms of the time delay of the compensation relative to the pulse generator output, while the compensation magnitude is
expressed as a change ΔIB from the standard operating point of about 10 mA.

As shown in figure  9, tuning just one of the bias signals par­ticularly affect small amplitude measurements, although a
will always modify the transients in some fashion and results stable thermal EMF can be removed by performing a polarity
in a change in rms voltage. If the rms value of the transients reversal measurement as is done with the PJVS. On the other
matches exactly the rms value of the PJVS waveform, then the hand, AC measurements, which separate lower-frequency
rms output is independent of the waveform frequency selected. noise from higher-frequency signals of interest will not be
This special operating bias point (the intersection of curves at affected by thermal EMF. Nevertheless, a system design that
different waveform frequencies) was first determined exper­ minimizes the thermal EMF is still worthwhile because it sim-
imentally in TTSs and TVCs [83, 85, 103] and later described plifies data analysis and improves system stability.
in detail by Burroughs et al [104]. However, the independent As mentioned at the end of section  2.2, the compensa-
tuning method (without relying on a TVC standard) is rather tion current applied using isolated current sources is a source
fastidious, and the accuracy of the method remains affected of systematic error [61, 62, 67, 68] but can be avoided with
by any external perturbations including current noise, stability use of zero-compensation waveforms if the reduction in
of the bias electronics and changes in microwave power due output voltage is acceptable. Because the JJ array is induc-
to changes in the liquid helium level in the dewar [104]. Due tive, compensation currents create an error voltage with fre-
to transient-related errors and the resulting lack of quantum quency-dependent amplitude that is approximately 90° out
accuracy in rms amplitude for stepwise-synthesized wave- of phase with the desired output voltage. In a typical NIST
forms, most rms measurements with PJVS systems have JAWS system generating an rms output of 1 V at 1 kHz with
been abandoned. Sampling or differential-sampling methods, a total inductance of about 50 nH and compensation current
which avoid the effects of the transients, are recommended for amplitude of about 10 mA, the resulting error voltage has an
performing ac voltage calibrations with stepwise PJVS wave- rms magnitude of about 3 µV. The magnitude of this error
forms (see section 4). voltage increases linearly with frequency and compensation
amplitude.
3.2.2.  JAWS limitations.  The JAWS system shares some of the The impact of this systematic error can be calculated by
same basic limitations as the PJVS, but the dominant source effectively measuring the JJ array inductance, the magnitude
of systematic errors changes as the waveform frequency of the compensation current and the phase of the compensa-
increases above about 10 kHz. Thermal EMFs play a similar tion current relative to the synthesized waveform from the JJ
role as in the PJVS system. However, current flowing in the array. Figure 10 shows the dependence of output voltage on
circuit can have a much more pervasive effect as a function of compensation current. As in the earlier PJVS example (figure
frequency. Specifically, systematic errors will arise from both 7), we obtain the maximum voltage resolution from a null
the compensation current through the JJ arrays mentioned ear- measurement resulting from the sum of two waveforms with
lier [62, 67, 68] and the current flowing in the output leads the same, 1 V rms magnitude and opposite phase. The residual
between the JJ arrays and the DUT [51, 67, 105–107]. This voltage difference is measured by a digitizer and expressed in
second source of error is typically called the ‘voltage lead’ figure 10 in terms of the in-phase and quadrature components.
error or correction. A bias parameter for one of the waveforms is then slightly
Thermal EMFs will affect the JAWS output voltage detuned from those for ‘standard operation’ where the QLR in
and cause offsets in rms or power measurements. This will response to a dc current offset is maximized.

S161
Metrologia 55 (2018) S152 A Rüfenacht et al

Another possible source of error is a low-frequency cur­ parameters when making comparisons between systems. In
rent from the pulse generator. Although this current should this case, we see that reducing the compensation magnitude
be removed by the dc blocks, an error voltage with a magni- by about 0.4 mA gives a measured difference of approxi-
tude and phase that depends on the details of the system will mately zero between the two systems. Further averaging and
occur at higher output waveform frequencies [68, 108]. The fine-tuning of the compensation magnitude and phase could
magnitude of this error can be determined by changing the allow the systems to agree with a type A uncertainty limited
amplitude of the bias from the pulse generator. In figure 10(A) only by the stability of the system. However, this would not
we observe the desired behavior, where the residual voltage is represent the true agreement; in practice, the effect of the
not dependent on detuning of the pulse magnitude while the compensation current and all the other bias parameters must
JJs are in the QLR. be measured and accounted for, either in the reported value or
In figures 10(B) and (C) we detune the compensation phase in the type B uncertainty.
and magnitude, respectively, and observe significant changes in
the residual voltage even while the system is quantum locked.
4.  Present applications of PJVS and JAWS systems
These changes are consistent with the above model of an error
voltage generated by compensation current passing through 4.1.  dc applications of JVS
the JJ arrays. Because the error voltage is approximately in
quadrature with the generated voltage, small changes in phase In principle, dc voltage calibrations can be performed by
will linearly affect the in-phase residual voltage while having either PJVS or JAWS systems. If the voltage output of a
minimal effect on the quadrature voltage. Similarly, small JAWS system reaches 10 V in the future, one might expect
changes in magnitude will linearly affect both the quadrature that JAWS systems will supplant PJVS systems for dc appli-
and in-phase residual voltages, but the scale of the effect will cations. However, it will be difficult to decrease the magni-
be significantly smaller for the in-phase residual voltage. tude of JAWS leakage currents to that of PJVS systems and
A more fundamental source of error is current flowing in difficult to reduce the instrument cost of the necessary multi-
the output leads [51, 67, 105, 106]. This current can have two channel pulse generator to match that of a single microwave
sources: leakage paths and the impedance of both the voltage source. Until now, PJVS circuits show the largest practical
output leads and the device under test. The first case is iden- output voltages, and dc voltage calibrations that require pro-
tical to that of the PJVS, except that stray capacitances will grammable voltages are presently performed almost exclu-
also result in leakage currents with frequency-dependent sively with PJVS systems.
magnitudes. In the second case, the fact that the JAWS is The main dc application for PJVS systems is the direct
an ac voltage source causes more significant errors due to calibration of secondary voltage standards [111]. With the
the impedance of the connecting leads and the inputs of the near total elimination of the use of chemical cells (Weston
DUTs. The PJVS system operating at dc typically uses preci- cells) from NMI laboratory benches, the secondary standard
sion dc voltmeters that have a large input resistance  >1 GΩ for dissemination of the unit volt is the temperature-controlled
and leads with resistance  <  1Ω. Inductances and capaci- Zener dc reference at 10 V [36, 37]. Although these devices
tances have no effect, so the output leads and DUT cause a have been well engineered to minimize sensitivity to environ­
simple voltage division at the DUT and an error of  <1 nV mental conditions and their inherent drift has been reduced
at 1 V dc. On the other hand, ac instruments typically have to a few µV/V per year [112], their output voltage remains
input resistances  <10 MΩ, and the input shunt capacitances   dependent on variations in temperature, atmospheric pressure
>10 pF cannot be neglected. The combination of the DUT and humidity. To build a history and establish confidence in
input impedance, on-chip inductance of the JJ arrays and its predicted value and uncertainty, a Zener dc reference must
inductive filters, and the cable inductance and capacitance typ- be regularly calibrated [113]. The principal advantage of a
ically results in a resonance between 10 MHz and 100 MHz PJVS system over a CJVS system is that its output voltage
[109]. The tail of this resonance has a significant effect on ac can be adjusted to perfectly match the Zener dc reference
measurements above 100 kHz that is proportional to the fre- voltage with use of the bridge measurement configuration. A
quency squared [51, 67, 105–107]. digital nanovoltmeter can then be used to measure very small
The effect of the output leads is exacerbated by the long residual voltages (typically  <1 µV, corresponding to the mag-
leads needed to reach cryogenic temperatures. If the JJ arrays nitude of thermal EMFs in the circuit). With no voltage step
are cooled with liquid helium, then the output leads are typi- jumps, unlike CJVS, the calibration duration may also be
cally about 1.5 m in length, whereas cryogen-free systems shorter. Another application of PJVS systems is calibration of
typically have leads that are longer than 0.5 m. Shorter leads gain and linearity of voltmeters [114, 115] and ADCs, which
are possible and would reduce the error voltage [53], but can requires the rapid programmability of PJVS systems.
result in a significant heat load on the cold stage. An alterna- With a greater immunity to external noise than CJVS sys-
tive approach is to attempt to flatten the frequency response of tems, PJVS systems can also be implemented in applications
the leads by adding additional impedance [108, 110]. such as the direct calibration of the dc range of commercial
Figure 10(C) also directly highlights the importance of calibrators and DACs [116]. Another application is the cali-
accounting for sources of systematic error and detuning bias bration of voltage ratios; for instance, DVM ratio calibrations

S162
Metrologia 55 (2018) S152 A Rüfenacht et al

Figure 11.  Diagram of various sampling methods: (a) and (b) sampling and switching and (c) and (d) differential sampling. The DUT can
be an ac source, as shown in (a) and (c), or the source output can be connected to a DUT, as shown in (b) and (d). Synchronization signals
to the various instruments are not shown in these diagrams for simplicity.

are required for applications such as measuring load cells used of dc resistance ratios with a commercial calibrator to supply
in mass metrology [117]. NIST is currently working to extend the current [129]. This ‘quantum calibrator’ approach still
the dual-voltage output option on a single PJVS system from requires an independent calibration of the current shunt or the
the 2 V circuit [118] to the 10 V circuit, eliminating the need reference resistances, and the overall accuracy of the measure-
for a Zener dc reference for the ratio calibration. ment relies on the stability and performance of the commer-
The dc calibration of nanovoltmeters cannot be performed cial calibrator used as a current source.
with a single PJVS system, because the voltage resolution is
limited by its LSB (see table 1) or, at best, would be defined
4.2.  AC application of JVS
by the voltage of a single JJ (typically many microvolts). To
generate voltages with amplitude less than a few hundred 4.2.1. PJVS stepwise approximated waveforms.  The only
microvolts, one method has been to use two independent accurate way to exploit a stepwise-approximated PJVS wave-
PJVS standards in a differential configuration [119–122]. The form for ac metrology is to implement a measurement method
same differential configuration is obtained when performing that avoids the transients. Only those portions of the PJVS
a direct comparison of two PJVS standards. By slightly waveform where the voltage is fully settled can be used as a
detuning the microwave bias frequency of one PJVS circuit, voltage reference for ac signals. With this constraint, a logi-
the gain calibration of the nanovoltmeter can easily be meas- cal measurement approach is to use sampling methods. Two
ured [8, 10, 74, 123]. Other methods to generate low-ampl­ types of sampling methods have been implemented with PJVS
itude dc voltages (and ac voltages with the same principle) waveforms: sampling plus switching (or multiplexing) and
involve reducing the pulse density with JAWS standards [124] differential sampling.
or developing a PJVS circuit with dual microwave bias signals In either case, the goal of sampling is to transfer the acc­
with independently tunable frequencies. uracy of the PJVS reference waveform to a custom or com-
In addition, other metrology applications rely on the acc­ mercial ac source. The output of the ac source can then be
uracy of PJVS systems. The electronic kilogram experiments connected directly to a detector (e.g. ac voltmeter, TTS, TVC)
with Kibble balances [75–79] and the Joule balance [80] were as long as the source can provide the necessary loading cur­
the first metrology applications of 1 V PJVS systems. Kibble rent to drive the detector. In this case, the sampler and the
and Joule balances will become a practical way to realize the PJVS, which acts as a sense input, should be placed as close as
unit of mass after the proposed SI redefinition [125]. Another possible to the detector reference plane (figure 11). Both sam-
example is the quantum metrology triangle, where electrical pling methods require a stable ac source with high spectral
quantum-based measurements are expected to verify the con- purity. Undesired harmonics in the frequency bandwidth of
sistency of three electrical quantum phenomena: the Josephson the sampler could create aliasing effects on the fundamental
effect for voltage, the quantum Hall effect for resistance of the waveform and potentially induce systematic error in the
(QHR), and single electron tunneling for current [126, 127]. rms voltage reconstruction.
More recently, an elegant method to realize the unit ampere The ‘sampling plus switching’ PJVS method was first
was proposed by coupling two electrical quantum standards demonstrated by PTB for the electric power standard [130].
(PJVS and QHR) with a cryogenic current comparator [128]. A slightly different version was implemented at METAS
New research efforts are focused on extending the PJVS based on comparing the fundamental harmonic of ac source
calibration range beyond ‘core’ voltage reference applica- waveforms to the PJVS [99, 131, 132]. Switching methods
tions. Recent work presented the concept of dc current meas- rely on the measurement capabilities of the sampler over the
urements with a calibrated current shunt and the measurement full waveform amplitude range (peak-to-peak). To minimize

S163
Metrologia 55 (2018) S152 A Rüfenacht et al

errors due to sampler non-linearity, the amplitude of the PJVS

waveform is adjusted to match that of the ac source wave-
forms. Synchronization between the ac source, PJVS circuit,
sampler and switching unit (multiplexer) is recommended to
remove the transient contributions from the PJVS waveform
and to facilitate rms amplitude reconstruction. However, the
switching approach does not require phase alignment between
the two waveforms. One factor to consider in the measurement
uncertainty is the resistive divider effect between the resist­
ance of the leads connecting the PJVS array to the sampler
input terminal and the input impedance of the sampler. The
magnitude of this effect is typically less when the multiplexer
is connected to the ac source (assuming an ideal output source
impedance) due to a shorter cable and hence lower resistance
on the ac source path.
The concept of PJVS differential sampling with an ac
source was first proposed by PTB [133] and was first demon- Figure 12.  Example of two types of PJVS reference waveforms
strated experimentally at NIST [134]. Further developments to track the ac source over multiple periods. (Top) alternating
and applications of the differential sampling method followed sampling windows (A) and (B) are shifted by T/2N to cover the full
at various NMIs [135–139], including the expansion of the range of the ac source period in the same measurement. (Bottom)
the PJVS reference is adjusted to match the ac source over m  =  3
rms waveform amplitude up to 7 V [140–142] and 10 V [143, periods.
144]. With the differential sampling method, the sampler is
used in a null-detector configuration, thus reducing errors due Due to the presence of the transients, the sampling win-
to the gain, non-linearity and input impedance of the sampler. dows do not usually cover a full period T  =  1/f of the ac source
To minimize the differential voltage measured by the sampler, signal. The rms amplitude calculation uses a sinusoidal model
the PJVS waveform shape, amplitude and phase are adjusted with an applied fit or fast Fourier transform, which may be
to match the ac source waveform. This can be achieved only if inaccurate if the ac source has spurious harmonics [147]. One
the three instruments (PJVS, ac source and sampler) are locked method to recover the full rms content is to combine two dis-
to a common frequency reference. Presence of phase jitter, tinct measurements with shifted sampling windows A and B
produced by either the ac source or the sampler, will auto- [148]. This principle can be applied to a single measurement
matically translate to an apparent increase in voltage noise in by programming a PJVS reference that covers multiple periods
the differential voltage measurement, an effect inherent to the of the waveform (figure 12, top). In this model, the sampling
differential sampling method. As a result, the rms output of duration corresponds to 50% of the PJVS step duration, or
the ac source is attributed with a larger type-A uncertainty and T/2N. When the switching occurs between the PJVS reference
does not reflect the real amplitude stability of the source. The A and PJVS reference B, no precision measurement can be
jitter effect can be observed with ac sources (ac voltage cali- performed (Gray zones containing PJVS transitions separated
brators) phase-locked at the same frequency as the generated by T/2N instead of T/N standard duration). Programming a
waveform. This inherent effect is only masked and hidden— PJVS waveform with multiple periods for each reference A
not eliminated—by performing an average of the differential and B, interleaved with the switching period, may improve
voltage signal over a large number of periods. the overall efficiency of the method. Compared to the two dis-
The results of the sampling method should be independent tinct measurements [148], a single measurement may more
of the selected number of steps in the PJVS waveform. Adding accurately track the short time stability of the source and may
more voltage steps per period (N) for the PJVS waveform has reduce overall uncertainty.
the benefit of reducing the differential voltage at the input of Another approach is to generate a PJVS reference wave-
the sampler. This is a definite advantage if the amplitude of the form that tracks the ac source over m multiple periods (figure
ac source waveform is large and the input range of the sampler 12, bottom). In this case, either N or m is often chosen to be
is limited [140]. However, increasing N reduces the duration an odd number. Over m periods of the ac source, the sampling
of the well-defined voltage for each step of the waveform. windows will cover the entire ac source waveform. One limita-
Generally, shorter sampling duration (or aperture duration) tion of this method is that the frequency f  =  (m/N)  ×  fPJVS-Clk
increases the noise in the voltage measurement [120, 145]. of the ac source waveform may not always have a fixed dec-
The same reduction effect of the sampling window arises if imal value. To facilitate the removal of transients, fPJVS-Clk
the frequency of the waveform increases for a given value of must be an integer sub-multiple k of the sampler digitizing
N, because the transient duration remains the same. Typically, frequency fs (i.e. fPJVS-Clk  =  fs/k). This technique can also
differential sampling methods are limited to waveform fre- be used to match some ac source frequencies that cannot
quencies below a few kilohertz. Subsampling methods have be commensurate with either fPJVS-Clk or fs over a single
been proposed to extend the frequency of the ac source while period (for instance, at f  =  60 Hz with m  =  3, N  =  100 and
keeping the reference PJVS waveform frequency f below fs  =  
10 MHz). Ultimately, for both methods shown in
1 kHz [146]. figure 12, the memory size of the PJVS bias electronics limits

S164
Metrologia 55 (2018) S152 A Rüfenacht et al

the number of PJVS samples N and indirectly the number of

periods that can be programmed.
Examples of applications in which PJVS sampling or
differ­ential sampling methods have been implemented include
calibration of ac sources (called ‘ac quantum voltmeter’ by
PTB and Supracon) [138, 140–144], electric power standards
[130, 149], calibration of TTSs and TVCs up to ~1 kHz [131,
137, 141], and dynamic linearity measurement of ADCs [150].

4.2.2. JAWS applications.  The main focus of the initial

JAWS systems was the calibration of TTSs using single-
frequency and DC waveforms [110, 151, 152]. A desire to
expand the calibration range has pushed the development of
higher output voltages [13], to the point where it is now pos- Figure 13.  Voltage versus frequency diagram showing the
sible to directly calibrate the high-impedance ranges of typi- application range of sampling methods for PJVS and JAWS. The
cal TTSs. Progress has also been made on using dividers to blue dashed line shows the future extension enabled by ongoing
calibrate standards’ higher voltage ranges [52, 153]. Further research at NIST to develop JAWS circuits with 4 V rms amplitude.
The red dash-dotted line represents the rms amplitude limit of 10 V
improvements in direct calibration will require accommo- PJVS circuits.
dating lower input impedances, perhaps by using buffer or
transconductance amplifiers [72, 153–155] similar to the one allow much to be learned about systematic errors. In par­
developed for implementation with a PJVS [83, 84]. Reaching ticular, pure JAWS ac waveforms can enable measurements
higher frequencies will require improved methods for limiting of systematic errors present in PJVS systems operated away
the systematic errors from the output leads, as discussed ear- from dc, and the effort taken to remove sources of error at dc
lier [51, 67, 105–107]. in PJVS systems will help improve the performance of JAWS
JAWS waveforms have been used to calibrate and char- systems at dc.
acterize the non-linearity of electrical components using a
variety of waveforms: dc offsets and single tones [124], two- 5.  AC voltage metrology within the new SI
tone waveforms in a general setting [156], and many tones as
a key component in Johnson noise thermometry [157, 158]. With the forthcoming redefinition of the SI, both PJVS and
Johnson noise thermometry extracts temperature by meas- JAWS systems will become key components for the direct
uring the Johnson noise of a resistor and these systems have realization of the unit volt. This has implications not only for
recently been used to measure the value of the Boltzmann dc voltage metrology, but could also have a huge impact for
constant [159–161]. JAWS waveforms are used to calibrate the future dissemination of ac voltages. For more than three
the entire measurement chain with μV/V precision at frequen- decades, dc voltage metrology has relied on quantum-accurate
cies up to 1 MHz, directly enabling μK/K temperature meas- sources (CJVS and PJVS) to represent the unit volt (KJ-90) and
urements. In this context, the JAWS source is often called a to calibrate secondary voltage standards (Zener voltage refer-
quantum voltage noise source (QVNS) because a ‘pseudo- ences). Currently, ac voltage metrology is uniquely based on
random’ JAWS waveform is used, that is, a comb in the fre- rms detectors (TVCs or TTSs) that typically require long cali-
quency domain, with a magnitude approximately matched to bration durations to ‘fingerprint’ each artifact’s behavior as a
the expected Johnson noise. function of both voltage and frequency. In principle, a switch
JAWS systems have also begun to be used as the dual from detector-based to source-based references can be applied
voltage sources in both two-terminal [162] and four-terminal to ac voltage metrology as well. Sampling or differential sam-
[163] automated impedance bridges. The advantage of JAWS pling methods are applicable to both PJVS waveforms (see
sources is that they can generate ac voltages (with arbitrary section 4.2.1) and JAWS waveforms [164, 167], overlapping
relative magnitude and phase) on command so that a single most of the frequency and voltage domain currently covered
bridge can compare any two impedances in an automated by TVCs and TTSs (figure 13). The accuracy of JAWS and
fashion. Typical high-precision bridges are manually operated PJVS reference waveforms must be transferred to low-noise
and use transformers as dual sources which severely limits the and stable ac sources capable of driving low-impedance loads.
type and value of the impedances that can be compared using Such ac sources would fulfill the role of secondary ac voltage
any single transformer. This shift from transformers to JAWS standards in the same way Zener standards are used to dissem-
systems has the potential for massively simplifying and auto- inate dc voltages. To significantly impact dissemination of ac
mating ac impedance metrology. voltage, the amplitude stability of the ac source must be of the
Finally, as will be discussed in more detail in the next sec- order of 1 part in 107, a value comparable to the performance
tion, an important use of JAWS systems is for intercomparions of the best TVCs. Unfortunately, today’s commercially avail-
with other JAWS [50, 164] and PJVS systems [49, 50, 165, able calibrators, with a short-term stability around 1 parts in
166]. While both JAWS and PJVS systems use JJ arrays, the 106, are not sufficient to perform this role. Additional require-
bias circuitry is sufficiently different that such comparisons ments on the ac source include low harmonic dist­ortion and

S165
Metrologia 55 (2018) S152 A Rüfenacht et al

the ability to synchronize to an external time base to enable

sampling methods. From an instrument design point of view,
achieving sub µV/V stability and excellent spectral purity
are not easily engineered. To reach these higher performance
levels, the thermal stability of the electronic components
inside the instrument plays a significant role, as does the DAC
architecture.
When an ac source is used in direct combination with JVS
standards, its absolute accuracy is not important. In this case,
the key consideration is the source’s short-term stability, so that
the ac voltage can be accurately transferred from the JVS to the
DUT (see figures 11(b) and (d) above). However, if the internal
voltage reference of the ac source can be monitored and the
ac transfer function calibrated at regular intervals with a JVS
and a sampling method, such an ac source becomes a ‘true’
standalone secondary ac voltage standard. The accuracy of the
Figure 14.  Schematic of direct and indirect comparison
ac amplitude is inferred from the measured dc reference and
possibilities between two PJVS, two JAWS, or a JAWS and a PJVS
the calibration factors based on the amplitude, frequency and voltage standards. The diagonal arrows with dashed outlines show
phase of the waveform generated. A source prototype based on indirect comparison links with an ac source as a transfer standard.
this approach has been developed by Nissilä et al [168].
The successful dissemination and excellent performance of tem based on it constitute an intrinsic voltage stand-
existing detector-based ac voltage metrology has not provided ard. Compariso­ns between a Josephson standard and
incentive for precision instrument manufacturers to develop any other primary standard, including another Joseph-
ac sources with better accuracy and stability. Hopefully, the son standard, never lead to a new calibration value for
potential for improved ac metrology with readily-available, the Josephson standard, but rather to a level of confi-
SI-realizable ac quantum standards will promote a paradigm dence that the Josephson standard itself is functioning
shift in ac metrology and spur innovation in the instrument correctly (no unaccounted-for errors exist). Periodic
market, in term of both sources and samplers. This paradigm compariso­ns between the JVS and other primary stand-
shift would take full advantage of the order of magnitude ards are required to ensure proper operation”.
improvement in the stability and accuracy provided by the Even after the expected 2019 SI redefinition, when JVS
JVS standards. systems will provide a direct realization of the SI unit volt and
An alternative approach to a new ac source is to develop not merely a representation, this statement will continue to
methods for JAWS systems to directly drive low-imped- remain valid by replacing ‘representation’ with ‘realization.’
ance loads without compromising waveform accuracy. This Determining the level of confidence in comparisons of JVS
approach would certainly not revolutionize the existing systems will become even more relevant in the presence of
detector-based ac metrology but would offer a faster and many disseminated ‘nominally equivalent’ Josephson systems
intrinsic method to calibrate the ac–dc difference of TVCs. that are all intrinsic standards. To verify the equivalence of
Because of the systematic errors due to transients, this two PJVS, two JAWS, or a JAWS and a PJVS, new compar-
approach cannot be successfully applied to PJVS waveforms ison protocols must be implemented for both dc and ac volt­
(see section 3.2.1). ages. As the schematic in figure 14 shows, some comparison
possibilities use an ac source as the transfer standard (indirect
6.  Comparison of JAWS and PJVS comparison). The different types of comparison measure-
ments can be grouped into four categories:
The proposed redefinition of the SI also provides new incentive (1) dc to dc comparison
to reconsider the way traceability is currently implemented at (2) ac stepwise to ac stepwise comparison
NMIs. The dissemination of the unit volt within a nation is (3) ac sinewave to ac stepwise comparison (direct and indi-
often traceable to a unique Josephson (‘gold’) standard. This rect)
‘legal’ approach is inconsistent with the universality of the (4) ac sinewave to ac sinewave comparison (direct and indi-
Josephson effect and its ability to realize the volt anywhere. rect).
The issue of traceability and equivalence of JVS systems was
already raised two decades ago [2] as formulated in the fol-
lowing statement from Hamilton [169]: 6.1.  DC to dc comparisons

“Because of its realization of a quantum physics Presently, key comparisons of JVS systems for dc voltages
phenom­enon, and the adoption of that phenomenon follow a well-established BIPM protocol [170] designed to
as the basis of the SI Volt Representation, a properly be implemented with the BIPM (CJVS) traveling standard
realized Josephson standard is correct by definition. [171]. Multiple direct comparisons between PJVS and CJVS
The Josephson array device and the metrological sys- systems [8, 9, 172–174], and between two PJVS systems

S166
Metrologia 55 (2018) S152 A Rüfenacht et al

[10, 44, 74, 123, 130, 141] have been performed at NMIs the two types of JVS systems (see figure 13) is limited by the
and calibration laboratories. But, the comparison of dc volt­ JAWS output voltage (presently  ⩽2 V rms) and the range of
ages should also include JAWS systems [175]. The different the PJVS stepwise waveform frequency (⩽1 kHz).
microwave bias used by JAWS and PJVS systems means that
performing a dc JAWS-to-PJVS comparison would provide
6.4.  Direct ac sinewave to ac sinewave comparisons
an intrinsic verification of JAWS voltage accuracy without
the influence of inductive errors that arise with ac waveforms. Direct comparison of two sinewaves can be achieved by con-
Such dc comparisons can also provide important information necting two JAWS arrays in series and applying a relative
about the magnitude of the dc leakage current error within the phase shift of 180° between the two waveforms. This type
measurement circuit and are an easy way to demonstrate the of comparison requires synchronization and fine adjustment
fundamental equivalence of JAWS and PJVS systems. of the relative phase between the two waveforms to null the
differential voltage. Fine phase adjustment is achieved by
rotating the pulse pattern of one waveform by one or multiple
6.2.  AC stepwise to ac stepwise comparisons
clock cycles of the pulse generator. The residual voltage can
Direct comparisons between two stepwise approximated be either measured with a lock-in amplifier at the frequency
PJVS waveforms [120, 133] provide useful information of the sinewaves [167] or with a digitizer [164]. Like the ac
regarding the accuracy of voltage steps and verify that the sinewave to ac stepwise comparison, JAWS-to-JAWS com-
proper method to remove transients has been applied. If the parisons provide the high resolution needed to perform QLR
two PJVS arrays to be compared have different designs or are characterization of bias parameters. As an example, such QLR
biased at different microwave frequencies, then the residual measurements would be useful to characterize systematic
voltage δVi for each step of the waveform will be in the range   errors that scale with the frequency of JAWS sinewave and
−VLSB/2  <  δVi  <  VLSB/2, with VLSB being the largest LSB errors that are due to the ac leakage current.
voltage of the two PJVS arrays [120]. Therefore, reaching a
perfect cancellation between two waveforms is not guaran-
6.5.  Indirect comparison measurements
teed. A 10 nV V−1 relative agreement between two stepwise
waveforms was measured for frequencies  <60 Hz [120]. The Another way to link two JVS systems is through indirect com-
accuracy of the comparison is ultimately limited by the sta- parison measurements. Such comparisons require using an ac
bility, noise, non-linearity and gain error of the digitizer used source with a good short-term amplitude stability as a transfer
to measure the differential voltage. Because JAWS systems standard (see section  5). The indirect comparisons that can
can generate stepwise-approximated waveforms, this compar- be applied to JAWS and PJVS systems are represented by the
ison method can also be applied to JAWS-PJVS comparisons. four diagonal arrows in figure 14. At least two measurements
Even if this method does not exploit the lack of inherently performed in quick succession are required to establish equiv-
non-quantized transients in JAWS, this type of comparison alence between two JVS systems. The main advantage of this
may provide a useful tool to test the measurement setup and method is its ability to verify that proper measurement and
the digitizer. analysis methods have been applied to transfer the accuracy of
The measured agreement between two stepwise approxi- the JVS to the ac source. The measured uncertainty will not be
mated waveforms remains an abstract concept. It is mis- as small as with a direct JVS comparison, because the trans-
leading to quote the result of this comparison measurement as ferring ac source is not an intrinsic standard. However, if the
a specification for a PJVS system, because the measurement amplitude of the ac source can be predicted with a long-term
is not applicable to any ac voltage metrology application. For uncertainty of 1 µV V−1 or less, then it can be implemented
example, the measurement does not account for any of the as a traveling standard and can be used in ‘round-robin’ inter-
errors associated with rms values of an ac source, as discussed laboratory comparisons (ILC). Establishing verification of ac
in previous sections. voltage measurement capability through a transfer-standard
ILC would simplify logistics compared to performing on-site
JVS comparisons. dc voltage ILCs with a Zener traveling ref-
6.3.  Direct ac sinewave to ac stepwise comparisons
erence are periodically conducted in North America [176].
Comparing JAWS sinewaves to PJVS stepwise-approximated
waveforms is probably the most interesting and challenging 6.6.  SI redefinition impact
method to verify the equivalence of these systems. The first
direct JAWS-to-PJVS comparison was performed at 100 mV The proposed redefinition of the SI will impact the voltage
and 500 Hz with a sampling and multiplexing method and metrology community far beyond a simple change in termi-
reported a relative agreement of (−0.18  ±  0.26) µV V−1 nology from ‘representation’ to ‘realization’ and a shift in the
[165]. A lower relative uncertainty of (+3.5  ±  11.7) nV V−1 numerical values of the fundamental constants e and h. After
was achieved at 1 V and 250 Hz with a differential sampling the redefinition, traceability will be replaced with verification
method [49]. This type of comparison is an ideal tool to test because every JVS system will realize the unit volt and not just
the limits of the differential sampling method and to quantify a handful of ‘gold standard’ JVS systems at NMIs. The role of
the magnitude of potential systematic errors associated with the NMIs will shift from providing a traceable JVS reference
the JAWS or the PJVS. Unfortunately, the overlap between to verifying that the disseminated JVS systems are functioning
S167
Metrologia 55 (2018) S152 A Rüfenacht et al

correctly. The biggest impact of the SI redefinition may be References

on ac voltage metrology. JAWS and PJVS waveforms, with
a direct link to the SI, provide a strong motivation to switch [1] Josephson B D 1962 Possible new effects in superconductive
from detector-based to source-based ac voltage metrology. tunnelling Phys. Lett. 1 251–3
The success of this paradigm shift depends on development [2] Hamilton C A 2000 Josephson voltage standards Rev. Sci.
of new commercial ac sources that can perform as secondary Instrum. 71 3611
[3] Jeanneret B and Benz S P 2009 Application of the Josephson
references. Additional research on direct comparison of JAWS
effect in electrical metrology Eur. Phys. J. Spec. Top.
and PJVS waveforms is needed to characterize all potential 172 181–206
sources of error, followed by establishing the measurement [4] Benz S P 1995 Superconductor-normal-superconductor
protocol for ac voltage comparisons. junctions for programmable voltage standards Appl. Phys.
Lett. 6711 2714
[5] Hamilton C A, Burroughs C J and Kautz R L 1995 Josephson
7. Conclusion D/A converter with fundamental accuracy IEEE Trans.
Instrum. Meas. 44 223–5
This review article describes in detail the presently fabricated [6] CGPM 2014 Resolution 1 of the 25th Meeting of the CGPM
General Conf. on Weights and Measures, 2014 (Versailles)
PJVS and JAWS devices, systems and measurement capabili- (www.bipm.org/utils/common/pdf/CGPM-2014/25th-
ties, as well as their performance and the systematic errors CGPM-Resolutions.pdf)
associated with their different voltage generation methods. [7] Consultative Committees of the CIPM 2017 Information for
PJVS and JAWS are increasingly implemented in numerous users about the proposed revision of the SI (www.bipm.org/
applications and calibrations, for both dc and ac voltage utils/common/pdf/SI-statement.pdf)
[8] Djordjevic S, Séron O, Solve S and Chayramy R 2008 Direct
metrology. By employing full automation and cryocooled comparison between a programmable and a conventional
refrigeration, the dissemination of quantum voltage standards Josephson voltage standard at the level of 10 V Metrologia
is expanding beyond NMI laboratories and into secondary 45 429–35
and industrial calibration laboratories. After the SI redefi- [9] Tang Y, Ojha V N, Schlamminger S, Rüfenacht A,
nition, PJVS and JAWS systems will, in practice, directly Burroughs C J, Dresselhaus P D and Benz S P 2012 A
10 V programmable Josephson voltage standard and its
realize the unit volt in all laboratories. JAWS systems have applications for voltage metrology Metrologia 49 635–43
huge potential for continued development, including further [10] Rüfenacht A, Tang Y, Solve S, Fox A E, Dresselhaus P D,
increases in output voltage through design of more complex Burroughs C J, Schwall R E, Chayramy R and Benz S P
circuits, increasing the output waveform frequency above 2018 Automated direct comparison of two cryocooled 10
1 MHz, and by applying new bias techniques. In the future, V programmable Josephson voltage standards Metrologia
55 585–96
electricity, mass, and temperature metrology and commercial [11] Inglis B D 1992 Standards for AC–DC transfer Metrologia
instruments will largely benefit from the order of magnitude 29 191–9
improvement in the stability, programmability, linearity and [12] Kieler O F, Behr R, Wendisch R, Bauer S, Palafox L and
accuracy provided by ac JVS sources. Kohlmann J 2015 Towards a 1 V Josephson arbitrary
waveform synthesizer IEEE Trans. Appl. Supercond. 25 1–5
[13] Flowers-Jacobs N E, Fox A E, Dresselhaus P D, Schwall R E
Acknowledgment and Benz S P 2016 Two-Volt Josephson arbitrary waveform
synthesizer using wilkinson dividers IEEE Trans. Appl.
Supercond. 26 1–7
The authors thank J Brevik, C Burroughs, G Butler, P Dres-
[14] Flowers-Jacobs N E, Waltman S B, Fox A E, Dresselhaus P D
selhaus, A Fox, D Olaya, R Schwall, and S Waltman in the and Benz S P 2016 Josephson arbitrary waveform
NIST Quantum Voltage Project for their contributions to the synthesizer with two layers of wilkinson dividers and an
development, fabrication and evaluation of the NIST JVS cir- FIR filter IEEE Trans. Appl. Supercond. 26 1–7
cuits as well as our collaborators S Cular, D Haddad, T Lipe, T [15] Shapiro S 1963 Josephson currents in superconducting
tunneling: the effect of microwaves and other observations
Nelson, F Seiffert, Y Tang, W Tew, J Underwood, and B Wal-
Phys. Rev. Lett. 11 80–2
trip for the implementation of JVS metrology applications at [16] Quinn T J 1989 News from the BIPM Metrologia 26 69–74
the NIST Gaithersburg campus. They would also like to thank [17] Newell D B et al 2018 The CODATA 2017 values of h, e, k,
Hirotake Yamamori, Oliver Kieler, and Johannes Kohlmann and NA for the revision of the SI Metrologia 55 L13–6
for their kind permission to present the picture of their devices [18] Burroughs C J, Dresselhaus P D, Rüfenacht A, Olaya D,
Elsbury M M, Tang Y-H and Benz S P 2011 NIST 10 V
in the figures.
Programmable Josephson voltage standard system IEEE
Trans. Instrum. Meas. 60 2482–8
[19] Behr R, Kieler O, Kohlmann J, Müller F and Palafox L 2012
ORCID iDs Development and metrological applications of Josephson
arrays at PTB Meas. Sci. Technol. 23 124002
Alain Rüfenacht https://orcid.org/0000-0002-1579-9738 [20] Burroughs C J, Benz S P, Hamilton C A and Harvey T E
1999 Programmable 1 V dc voltage standard IEEE Trans.
Nathan E Flowers-Jacobs https://orcid.org/0000-0002- Instrum. Meas. 48 279–81
9081-5187 [21] Behr R, Grimm L, Funck T, Kohlmann J, Schulze H,
Samuel P Benz https://orcid.org/0000-0002-8679-0765 Muller F, Schumacher B, Warnecke P and Niemeyer J 2001

S168
Metrologia 55 (2018) S152 A Rüfenacht et al

Application of Josephson series arrays to a DC quantum [38] Lacquaniti V, De Leo N, Fretto M, Sosso A, Müller F and
voltmeter IEEE Trans. Instrum. Meas. 50 185–7 Kohlmann J 2011 1 V programmable voltage standards
[22] Chong Y, Burroughs C J, Dresselhaus P D, Hadacek N, based on SNIS Josephson junction series arrays Supercond.
Yamamori H and Benz S P 2005 2.6 V high-resolution Sci. Technol. 24 045004
programmable Josephson voltage standard circuits using [39] Khorshev S K, Pashkovsky A I, Rogozhkina N V,
double-stacked MoSi2-barrier junctions IEEE Trans. Levichev M Y, Pestov E E, Katkov A S, Behr R,
Instrum. Meas. 54 616–9 Kohlmann J and Klushin A M 2016 Accuracy of the new
[23] Baek B, Dresselhaus P D and Benz S P 2006 Co-sputtered voltage standard using Josephson junctions cooled to 77 K
amorphous NbxSi1−x barriers for Josephson-junction 2016 Conf. on Precision Electromagnetic Measurements
circuits IEEE Trans. Appl. Supercond. 16 1966–70 (CPEM 2016) (IEEE) (https://doi.org/10.1109/
[24] Dresselhaus P D, Elsbury M M, Olaya D, Burroughs C J and CPEM.2016.7540701)
Benz S P 2011 10 V programmable Josephson voltage [40] Klushin A M, Pestov E E, Galin M A and Levichev M Y 2016
standard circuits using NbSi-barrier junctions IEEE Trans. High-temperature superconductor Josephson junctions for
Appl. Supercond. 21 693–6 voltage standards Phys. Solid State 58 2196–202
[25] Elsbury M M, Dresselhaus P D, Bergren N F, Burroughs C J, [41] Hamilton C A, Burroughs C J, Benz S P and Kinard J R 1997
Benz S P and Popovic Z 2009 Broadband lumped-element AC Josephson voltage standard: progress report IEEE
integrated N-way power dividers for voltage standards IEEE Trans. Instrum. Meas. 46 224–8
Trans. Microw. Theory Tech. 57 2055–63 [42] Yamada T, Urano C, Nishinaka H, Murayama Y, Iwasa A,
[26] Dresselhaus P D, Elsbury M M and Benz S P 2009 Tapered Yamamori H, Sasaki H, Shoji A and Nakamura Y 2010
Transmission Lines With Dissipative Junctions IEEE Trans. Single-chip 10 V programmable Josephson voltage standard
Appl. Supercond. 19 993–8 system based on a refrigerator and its precision evaluation
[27] Chong Y, Burroughs C J, Dresselhaus P D, Hadacek N, IEEE Trans. Appl. Supercond. 20 21–5
Yamamori H and Benz S P 2005 Practical high-resolution [43] Rüfenacht A, Howe L A, Fox A E, Schwall R E,
programmable Josephson voltage standards using double- Dresselhaus P D, Burroughs C J and Benz S P 2015
and triple- stacked MoSi2-barrier junctions IEEE Trans. Cryocooled 10 V programmable Josephson voltage standard
Appl. Supercond. 15 461–4 IEEE Trans. Instrum. Meas. 64 1477–82
[28] Mueller F, Behr R, Palafox L, Kohlmann J, Wendisch R and [44] Schubert M et al 2016 A dry-cooled AC quantum voltmeter
Krasnopolin I 2007 Improved 10 V SINIS series arrays for Supercond. Sci. Technol. 29 105014
applications in AC voltage metrology IEEE Trans. Appl. [45] Benz S P and Hamilton C A 1996 A pulse-driven
Supercond. 17 649–52 programmable Josephson voltage standard Appl. Phys. Lett.
[29] Mueller F, Behr R, Weimann T, Palafox L, Olaya D, 68 3171
Dresselhaus P D and Benz S P 2009 1 V and 10 V SNS [46] Candy J C 1997 An overview of basic concepts Delta-Sigma
programmable voltage standards for 70 GHz IEEE Trans. Data Converters: Theory, Design, and Simulation ed
Appl. Supercond. 19 981–6 S R Norsworthy et al (Piscataway, NJ: IEEE Press)
[30] Mueller F, Scheller T, Wendisch R, Behr R, Kieler O, [47] Pavan S, Schreier R and Temes G C 2017 Understanding
Palafox L and Kohlmann J 2013 NbSi barrier junctions Delta-Sigma Data Converters (New York: Wiley)
tuned for metrological applications up to 70 GHz: 20 V [48] Kohlmann J, Kieler O, Scheller T, Egeling B, Wendisch R
arrays for programmable Josephson voltage standards IEEE and Behr R 2016 Series arrays of NbSi barrier Josephson
Trans. Appl. Supercond. 23 1101005 junctions for AC voltage standards 2016 Conf. on Precision
[31] Yamamori H, Ishizaki M, Shoji A, Dresselhaus P D and Electromagnetic Measurements (CPEM 2016) (IEEE)
Benz S P 2006 10 V programmable Josephson voltage (https://doi.org/10.1109/CPEM.2016.7540562)
standard circuits using NbN/TiNx/NbN/TiNx/NbN double- [49] Behr R, Kieler O, Lee J, Bauer S, Palafox L and Kohlmann J
junction stacks Appl. Phys. Lett. 88 042503 2015 Direct comparison of a 1 V Josephson arbitrary
[32] Yamamori H, Yamada T, Sasaki H and Shoji A 2008 A 10 V waveform synthesizer and an ac quantum voltmeter
programmable Josephson voltage standard circuit with a Metrologia 52 528–37
maximum output voltage of 20 V Supercond. Sci. Technol. [50] Filipski P S, van den Brom H E and Houtzager E 2012
21 105007 International comparison of quantum AC voltage standards
[33] Yamamori H, Yamada T, Sasaki H and Shoji A 2010 Improved for frequencies up to 100 kHz Measurement 45 2218–25
Fabrication Yield for 10 V programmable Josephson voltage [51] van den Brom H E and Houtzager E 2012 Voltage lead
standard circuit including 524288 NbN/TiN/NbN Josephson corrections for a pulse-driven ac Josephson voltage standard
junctions IEEE Trans. Appl. Supercond. 20 71–5 Meas. Sci. Technol. 23 124007
[34] Fox A E, Dresselhaus P D, Rüfenacht A, Sanders A [52] Hagen T, Budovsky I, Benz S P and Burroughs C J 2012
and Benz S P 2015 Junction yield analysis for 10 V Calibration system for AC measurement standards using a
programmable Josephson voltage standard devices IEEE pulse-driven Josephson voltage standard and an inductive
Trans. Appl. Supercond. 25 1–5 voltage divider 2012 Conf. on Precision Electromagnetic
[35] Yamamori H and Kohjiro S 2016 Fabrication of voltage Measurements (CPEM 2012) (IEEE) pp 672–3
standard circuits utilizing a serial-parallel power divider [53] van den Brom H E, Kieler O F O, Bauer S and Houtzager E
IEEE Trans. Appl. Supercond. 26 1–4 2017 AC–DC calibrations with a pulse-driven AC
[36] Fluke 2013 Fractional ppm traceability using your Josephson voltage standard operated in a small cryostat
734A/732B series DC reference standards Fluke IEEE Trans. Instrum. Meas. 66 1391–6
Calibration Application Note (http://download.flukecal. [54] Durandetto P, Monticone E, Trinchera B, Serazio D and
com/pub/literature/1260304D_734A_732B_Fract_Trace_ Sosso A 2017 Cryocooled programmable and pulse-
AN_w.pdf) driven Josephson voltage standards at INRiM 2017
[37] Maruyama M, Urano C, Kaneko N, Sannomaru E, IEEE Int. Instrumentation and Measurement Technology
Yonezawa T, Kanai T, Yoshida H and Yoshino Y 2016 Conf. (I2MTC) (IEEE) (https://doi.org/10.1109/
Development of a compact Zener DC voltage standard I2MTC.2017.7969790)
with detachable module system 2016 Conf. on Precision [55] Benz S P, Hamilton C A, Burroughs C J and Harvey T E 1999
Electromagnetic Measurements (CPEM 2016) (IEEE) AC and DC bipolar voltage source using quantized pulses
(https://doi.org/10.1109/CPEM.2016.7540703) IEEE Trans. Instrum. Meas. 48 266–9

S169
Metrologia 55 (2018) S152 A Rüfenacht et al

[56] Benz S P, Burroughs C J, Harvey T E and Hamilton C A 1999 [75] Eichenberger A, Baumann H, Jeanneret B, Jeckelmann B,
Operating conditions for a pulse-quantized AC and DC bipolar Richard P and Beer W 2011 Determination of the Planck
voltage source IEEE Trans. Appl. Supercond. 9 3306–9 constant with the METAS watt balance Metrologia
[57] Benz S P, Burroughs C J and Dresselhaus P D 2001 AC 48 133–41
coupling technique for Josephson waveform synthesis IEEE [76] Fang H, Kiss A, Picard A and Stock M 2014 A watt balance
Trans. Appl. Supercond. 11 612–6 based on a simultaneous measurement scheme Metrologia
[58] van den Brom H E, Houtzager E, Brinkmeier B E R and 51 S80–7
Chevtchenko O A 2008 Bipolar pulse-drive electronics for [77] Haddad D, Seifert F, Chao L S, Possolo A, Newell D B,
a Josephson arbitrary waveform synthesizer IEEE Trans. Pratt J R, Williams C J and Schlamminger S 2017
Instrum. Meas. 57 428–31 Measurement of the Planck constant at the National
[59] Houtzager E, Benz S P and van den Brom H E 2009 Operating Institute of Standards and Technology from 2015 to 2017
margins for a pulse-driven Josephson arbitrary waveform Metrologia 54 633–41
synthesizer using a ternary bit-stream generator IEEE [78] Wood B M, Sanchez C A, Green R G and Liard J O 2017 A
Trans. Instrum. Meas. 58 775–80 summary of the Planck constant determinations using the
[60] Houtzager E, van den Brom H E and van Woerkom D 2010 NRC Kibble balance Metrologia 54 399–409
Automatic tuning of the pulse-driven AC Josephson [79] Thomas M, Ziane D, Pinot P, Karcher R, Imanaliev A, Dos
voltage standard 2010 Conf. on Precision Electromagnetic Santos F P, Merlet S, Piquemal F and Espel P 2017 A
Measurements (CPEM 2010) (IEEE) pp 185–6 determination of the Planck constant using the LNE Kibble
[61] Benz S P and Waltman S B 2014 Pulse-bias electronics and balance in air Metrologia 54 468–80
techniques for a Josephson arbitrary waveform synthesizer [80] Li Z et al 2017 The first determination of the Planck constant
IEEE Trans. Appl. Supercond. 24 1–7 with the joule balance NIM-2 Metrologia 54 763–74
[62] Zhou K, Qu J and Benz S P 2015 Zero-compensation method [81] Lombardi M A 2016 Evaluating the frequency and time
and reduced inductive voltage error for the AC Josephson uncertainty of GPS disciplined oscillators and clocks
voltage standard IEEE Trans. Appl. Supercond. 25 1400806 NCSLI Meas. 11 30–44
[63] Watanabe M, Dresselhaus P D and Benz S P 2006 Resonance- [82] Williams J M, Henderson D, Patel P, Behr R and Palafox L
free low-pass filters for the AC Josephson voltage standard 2007 Achieving Sub-100 ns switching of programmable
IEEE Trans. Appl. Supercond. 16 49–53 Josephson arrays IEEE Trans. Instrum. Meas. 56 651–4
[64] Benz S P, Waltman S B, Fox A E, Dresselhaus P D, [83] Seron O, Djordjevic S, Budovsky I, Hagen T, Behr R and
Rüfenacht A, Underwood J M, Howe L A, Schwall R E Palafox L 2012 Precision AC–DC transfer measurements
and Burroughs C J 2015 1 V Josephson arbitrary waveform with a Josephson waveform synthesizer and a buffer
synthesizer IEEE Trans. Appl. Supercond. 25 1–8 amplifier IEEE Trans. Instrum. Meas. 61 198–204
[65] Benz S P, Waltman S B, Fox A E, Dresselhaus P D, [84] Budovsky I, Behr R, Palafox L, Djordjevic S and Hagen T
Rüfenacht A, Howe L, Schwall R E and Flowers- 2012 Technique for the calibration of thermal voltage
Jacobs N E 2015 Performance improvements for the NIST converters using a Josephson waveform synthesizer and a
1 V Josephson arbitrary waveform synthesizer IEEE Trans. transconductance amplifier Meas. Sci. Technol. 23 124005
Appl. Supercond. 25 1–5 [85] Burroughs C J, Rüfenacht A, Benz S P, Dresselhaus P D,
[66] Brevik J A, Flowers-Jacobs N E, Fox A E, Golden E B, Waltrip B C and Nelson T L 2008 Error and transient
Dresselhaus P D and Benz S P 2017 Josephson arbitrary analysis of stepwise-approximated sine waves generated by
waveform synthesis with multilevel pulse biasing IEEE programmable Josephson voltage standards IEEE Trans.
Trans. Appl. Supercond. 27 1–7 Instrum. Meas. 57 1322–9
[67] Burroughs C J, Benz S P and Dresselhaus P D 2003 AC [86] Burroughs C J, Rüfenacht A, Benz S P and Dresselhaus P D
Josephson voltage standard error measurements and 2009 Systematic error analysis of stepwise-approximated
analysis IEEE Trans. Instrum. Meas. 52 542–4 AC waveforms generated by programmable Josephson
[68] Landim R P, Benz S P, Member S, Dresselhaus P D and voltage standards IEEE Trans. Instrum. Meas. 58 761–7
Burroughs C J 2008 Systematic-error signals in the AC [87] Helisto P, Nissila J, Ojasalo K, Penttila J S and Seppa H 2003
Josephson voltage standard: measurement and reduction AC voltage standard based on a programmable SIS array
IEEE Trans. Instrum. Meas. 57 1215–20 IEEE Trans. Instrum. Meas. 52 533–7
[69] Likharev K K and Semenov V K 1991 RSFQ logic/memory [88] Behr R, Williams J M, Patel P, Janssen T J B M, Funck T and
family: a new Josephson-junction technology for sub- Klonz M 2005 Synthesis of precision waveforms using
terahertz-clock-frequency digital systems IEEE Trans. Appl. a SINIS Josephson junction array IEEE Trans. Instrum.
Supercond. 1 3–28 Meas. 54 612–5
[70] Sasaki H, Bubanja V, Kiryu S, Hirayama F, Maezawa M and [89] Burroughs C J et al 2007 Development of a 60 Hz power
Shoji A 2001 Evaluation of AC–DC difference of thermal standard using SNS programmable Josephson voltage
converters using an SFQ-based D/A converter IEEE Trans. standards IEEE Trans. Instrum. Meas. 56 289–94
Instrum. Meas. 50 318–21 [90] Lee J, Schurr J, Nissilä J, Palafox L and Behr R 2010 The
[71] Maezawa M, Mizugaki Y, Takahashi Y and Shimada H 2014 9 Josephson two-terminal-pair impedance bridge Metrologia
bit superconductive single-flux-quantum digital-to-analogue 47 453–9
converter Electron. Lett. 50 1637–9 [91] Lee J, Schurr J, Nissilä J, Palafox L, Behr R and Kibble B P
[72] Budovsky I, Sasaki H and Coogan P 2003 AC–DC transfer 2011 Programmable Josephson arrays for impedance
comparator for the calibration of thermal voltage converters measurements IEEE Trans. Instrum. Meas. 60 2596–601
against Josephson alternating voltage standards IEEE Trans. [92] Budovsky I, Georgakopoulos D, Hagen T, Sasaki H
Instrum. Meas. 52 538–41 and Yamamori H 2011 Precision AC–DC difference
[73] Maezawa M, Yamada T and Urano C 2014 Integrated quantum measurement system based on a programmable Josephson
voltage noise source for Johnson noise thermometry J. voltage standard IEEE Trans. Instrum. Meas. 60 2439–44
Phys.: Conf. Ser. 507 42023 [93] Palafox L, Behr R, Nissila J, Schurr J and Kibble B P 2012
[74] Solve S, Rüfenacht A, Burroughs C J and Benz S P 2013 Josephson impedance bridges as universal impedance
Direct comparison of two NIST PJVS systems at 10 V comparators 2012 Conf. on Precision Electromagnetic
Metrologia 50 441–51 Measurements (IEEE) pp 464–5

S170
Metrologia 55 (2018) S152 A Rüfenacht et al

[94] Behr R, Kieler O F O, Schleubner D, Palafox L and Ahlers F- standards using a 10 V programmable Josephson voltage
J 2013 Combining Josephson systems for spectrally standard at NMIJ IEEE Trans. Instrum. Meas. 64 1606–12
pure AC waveforms with large amplitudes IEEE Trans. [112] Witt T J 2002 Maintenance and dissemination of voltage
Instrum. Meas. 62 1634–9 standards by Zener-diode-based instruments IEE Proc.,
[95] Palafox L, Behr R, Schurr J and Kibble B P 2014 Precision Sci. Meas. Technol. 149 305–12
10:1 capacitance ratio measurement using a Josephson [113] Hamilton C A and Tarr L W 2003 Projecting Zener dc
impedance bridge 2014 Conf. on Precision Electromagnetic reference performance between calibrations IEEE Trans.
Measurements (CPEM 2014) (IEEE) pp 232–3 Instrum. Meas. 52 454–6
[96] Eklund G, Bergsten T, Hagen T, Palafox L and Behr R 2016 [114] van den Brom H E, Houtzager E, Rietveld G, van
A comparison of the Josephson impedance bridges of Bemmelen R and Chevtchenko O 2007 Voltage linearity
PTB and SP 2016 Conf. on Precision Electromagnetic measurements using a binary Josephson system Meas. Sci.
Measurements (CPEM 2016) (IEEE) (https://doi. Technol. 18 3316–20
org/10.1109/CPEM.2016.7540584) [115] Maruyama M, Takahashi H, Katayama K, Yonezawa T,
[97] Hagen T, Thevenot O, Seron O, Khan S, Palafox L and Kanai T, Iwasa A, Urano C, Kiryu S and Kaneko N 2015
Behr R 2016 Comparison of the frequency dependence of Evaluation of linearity characteristics in digital voltmeters
capacitance ratios between LNE and PTB 2016 Conf. on using a PJVS system With a 10 K cooler IEEE Trans.
Precision Electromagnetic Measurements (CPEM 2016) Instrum. Meas. 64 1613–9
(IEEE) (https://doi.org/10.1109/CPEM.2016.7540585) [116] Rüfenacht A, Burroughs C J, Benz S P and Dresselhaus P D
[98] Hagen T, Palafox L and Behr R 2017 A Josephson impedance 2012 A digital-to-analog converter with a voltage standard
bridge based on programmable Josephson voltage reference 2012 Conf. on Precision Electromagnetic
standards IEEE Trans. Instrum. Meas. 66 1539–45 Measurements (CPEM 2012) (IEEE) pp 436–7
[99] Jeanneret B, Overney F, Callegaro L, Mortara A and [117] Tang Y, Bartel T W and Sims J E 2008 Ratio calibration
Rüfenacht A 2009 Josephson-voltage-standard-locked sine of a digital voltmeter for force measurement using the
wave synthesizer: margin evaluation and stability IEEE programmable Josephson voltage standard NCSLI Meas.
Trans. Instrum. Meas. 58 791–6 3 70–5
[100] Jeanneret B, Overney F, Rüfenacht A and Nissila J 2010 [118] Rüfenacht A, Fox A E, Dresselhaus P D, Burroughs C J,
Strong attenuation of the transients’ effect in square waves Benz S P, Waltrip B C and Nelson T L 2016 Simultaneous
synthesized with a programmable Josephson voltage double waveform synthesis with a single programmable
standard IEEE Trans. Instrum. Meas. 59 1894–9 Josephson voltage standard 2016 Conf. on Precision
[101] Burroughs C J, Benz S P, Hamilton C A, Harvey T E, Electromagnetic Measurements (CPEM 2016) (IEEE)
Kinard J R, Lipe T E and Sasaki H 1999 Thermoelectric (https://doi.org/10.1109/CPEM.2016.7540471)
transfer difference of thermal converters measured with a [119] Chong Y, Kim M S and Kim K T 2006 Fast and almost
Josephson source IEEE Trans. Instrum. Meas. 48 282–4 continuously programmable Josephson voltage standard
[102] Funck T, Behr R and Klonz M 2001 Fast reversed DC system with multiple microwave drive 2006 Conf. on
measurements on thermal converters using a SINIS Precision Electromagnetic Measurements (CPEM 2006)
Josephson junction array IEEE Trans. Instrum. Meas. Conf. Digest pp 382–3
50 322–5 [120] Rüfenacht A, Burroughs C J and Benz S P 2008 Precision
[103] Eklund G, Bergsten T, Tarasso V and Rydler K-E 2011 sampling measurements using ac programmable Josephson
Determination of transition error corrections for low voltage standards Rev. Sci. Instrum. 79 044704
frequency stepwise-approximated Josephson sine waves [121] Georgakopoulos D, Budovsky I, Hagen T, Sasaki H and
IEEE Trans. Instrum. Meas. 60 2399–403 Yamamori H 2012 Dual radiofrequency drive quantum
[104] Burroughs C J, Rüfenacht A, Benz S P and Dresselhaus P D voltage standard with nanovolt resolution based on
2013 Method for ensuring accurate AC waveforms with a closed-loop refrigeration cycle Meas. Sci. Technol.
programmable Josephson voltage standards IEEE Trans. 23 124003
Instrum. Meas. 62 1627–33 [122] Li H, Gao Y and Wang Z 2016 A differential programmable
[105] Filipski P S, Kinard J R, Lipe T E and Benz S P 2009 Josephson voltage standard for low-measurement 2016
Correction of systematic errors due to the voltage leads in Conf. on Precision Electromagnetic Measurements
an AC Josephson voltage standard IEEE Trans. Instrum. (CPEM 2016) (IEEE) (https://doi.org/10.1109/
Meas. 58 853–8 CPEM.2016.7540473)
[106] Filipski P S, Boecker M, Benz S P and Burroughs C J 2011 [123] Solve S, Chayramy R, Maruyama M, Urano C, Kaneko N-
Experimental determination of the voltage lead error in H and Rüfenacht A 2018 Direct DC 10 V comparison
an AC Josephson voltage standard IEEE Trans. Instrum. between two programmable Josephson voltage standards
Meas. 60 2387–92 made of niobium nitride (NbN)-based and niobium (Nb)-
[107] van den Brom H E, Zhao D and Houtzager E 2016 Voltage lead based Josephson junctions Metrologia 55 302–13
errors in an AC Josephson voltage standard: explanation [124] Behr R, Kieler O and Schumacher B 2017 A precision
in terms of standing waves 2016 Conf. on Precision microvolt-synthesizer based on a pulse-driven
Electromagnetic Measurements (CPEM 2016) (IEEE) Josephson voltage standard IEEE Trans. Instrum. Meas.
(https://doi.org/10.1109/CPEM.2016.7540666) 66 1385–90
[108] Zhao D, van den Brom H E and Houtzager E 2017 Mitigating [125] Robinson I A and Schlamminger S 2016 The watt or
voltage lead errors of an AC Josephson voltage standard Kibble balance: a technique for implementing the new SI
by impedance matching Meas. Sci. Technol. 28 95004 definition of the unit of mass Metrologia 53 A46–74
[109] Underwood J M 2018 Uncertainty analysis for ac–dc [126] Devoille L, Feltin N, Steck B, Chenaud B, Sassine S,
difference measurements with the AC Josephson voltage Djordevic S, Séron O and Piquemal F 2012 Quantum
standard in preparation metrological triangle experiment at LNE: measurements on a
[110] Lipe T E, Kinard J R, Tang Y, Benz S P, Burroughs C J three-junction R-pump using a 20 000:1 winding
and Dresselhaus P D 2008 Thermal voltage converter ratio cryogenic current comparator Meas. Sci. Technol.
calibrations using a quantum ac standard Metrologia 23 124011
45 275–80 [127] Scherer H and Camarota B 2012 Quantum metrology
[111] Maruyama M, Iwasa A, Yamamori H, Chen S-F, Urano C triangle experiments: a status review Meas. Sci. Technol. 23
and Kaneko N 2015 Calibration system for zener voltage 124010
S171
Metrologia 55 (2018) S152 A Rüfenacht et al

[128] Brun-Picard J, Djordjevic S, Leprat D, Schopfer F and Electromagnetic Measurements (CPEM 2016) (IEEE)
Poirier W 2016 Practical quantum realization of the (https://doi.org/10.1109/CPEM.2016.7540720)
ampere from the elementary charge Phys. Rev. X 6 041051 [145] Kurten Ihlenfeld W G, Mohns E, Behr R, Williams J M,
[129] Lee J, Behr R, Schumacher B, Palafox L, Schubert M, Patel P, Ramm G and Bachmair H 2005 Characterization
Starkloff M, Bock A C and Fleischmann P M 2016 of a high-resolution analog-to-digital converter with a
From AC quantum voltmeter to quantum calibrator Josephson AC voltage source IEEE Trans. Instrum. Meas.
2016 Conf. on Precision Electromagnetic Measurements 54 649–52
(CPEM 2016) (IEEE) (https://doi.org/10.1109/ [146] Ihlenfeld W G K and Landim R P 2016 Investigations on
CPEM.2016.7540470) extending the frequency range of PJVS based AC voltage
[130] Palafox L, Ramm G, Behr R, Kurten Ihlenfeld W G and calibrations by coherent subsampling 2016 Conf. on Precision
Moser H 2007 Primary AC power standard based on Electromagnetic Measurements (CPEM 2016) (IEEE)
programmable Josephson junction arrays IEEE Trans. (https://doi.org/10.1109/CPEM.2016.7539732)
Instrum. Meas. 56 534–7 [147] Katkov A, Gubler G, Lee J, Behr R and Nissila J
[131] Rüfenacht A, Overney F, Mortara A and Jeanneret B 2011 2014 Influence of harmonics on AC measurements
Thermal-transfer standard validation of the Josephson- using a quantum voltmeter 2014 Conf. on Precision
voltage-standard-locked sine-wave synthesizer IEEE Electromagnetic Measurements (CPEM 2014) (IEEE) pp
Trans. Instrum. Meas. 60 2372–7 526–7
[132] Jeanneret B, Overney F and Rüfenacht A 2012 The [148] Lee J, Nissila J, Katkov A and Behr R 2014 A quantum
Josephson locked synthesizer Meas. Sci. Technol. voltmeter for precision AC measurements 2014 Conf. on
23 124004 Precision Electromagnetic Measurements (CPEM 2014)
[133] Behr R, Palafox L, Ramm G, Moser H and Melcher J 2007 (IEEE) pp 732–3
Direct comparison of Josephson waveforms using an AC [149] Waltrip B C, Gong B, Nelson T L, Wang Y, Burroughs C J,
quantum voltmeter IEEE Trans. Instrum. Meas. 56 235–8 Rüfenacht A, Benz S P and Dresselhaus P D 2009 AC
[134] Rüfenacht A, Burroughs C J, Benz S P, Dresselhaus P D, power standard using a programmable Josephson voltage
Waltrip B C and Nelson T L 2009 Precision differential standard IEEE Trans. Instrum. Meas. 58 1041–8
sampling measurements of low-frequency synthesized [150] Overney F, Rüfenacht A, Braun J-P, Jeanneret B and
sine waves with an AC programmable Josephson voltage Wright P S 2011 Characterization of metrological grade
standard IEEE Trans. Instrum. Meas. 58 809–15 analog-to-digital converters using a programmable
[135] Kim M-S, Kim K-T, Kim W-S, Chong Y and Kwon S-W Josephson voltage standard IEEE Trans. Instrum. Meas.
2010 Analog-to-digital conversion for low-frequency 60 2172–7
waveforms based on the Josephson voltage standard Meas. [151] Benz S P, Burroughs C J, Dresselhaus P D and Christian L A
Sci. Technol. 21 115102 2001 AC and DC voltages from a Josephson arbitrary
[136] Williams J M, Henderson D, Pickering J, Behr R, Müller F waveform synthesizer IEEE Trans. Instrum. Meas. 50 181–4
and Scheibenreiter P 2011 Quantum-referenced voltage [152] Benz S P, Burroughs C J, Dresselhaus P D, Lipe T E and
waveform synthesiser IET Sci. Meas. Technol. 5 163–74 Kinard J R 2006 100 mV AC–DC transfer standard
[137] Amagai Y, Maruyama M and Fujiki H 2013 Low- measurements using an AC Josephson voltage standard
frequency characterization in thermal converters using NCSLI Meas. J. Meas. Sci. 1 50–5
AC-programmable Josephson voltage standard system [153] Karlsen B, Lind K, Malmbekk H and Ohlckers P 2016
IEEE Trans. Instrum. Meas. 62 1621–6 Development of high precision voltage dividers and buffer for
[138] Kurten Ihlenfeld W G and Pinheiro Landim R 2015 An AC voltage metrology up to 1 MHz 2016 Conf. on Precision
automated Josephson-based AC-voltage calibration system Electromagnetic Measurements (CPEM 2016) (IEEE)
IEEE Trans. Instrum. Meas. 64 1779–84 (https://doi.org/10.1109/CPEM.2016.7540456)
[139] Wang Z, Li H, Yang Y and Gao Y 2016 Research on [154] Budovsky I and Palafox L 2016 10 V transconductance
differential sampling with a Josephson voltage standard amplifier for the comparison of Josephson standards
2016 Conf. on Precision Electromagnetic Measurements and thermal converters 2016 Conf. on Precision
(CPEM 2016) (IEEE) pp 1–2 Electromagnetic Measurements (CPEM 2016) (IEEE)
[140] Rüfenacht A, Burroughs C J, Dresselhaus P D and Benz S P (https://doi.org/10.1109/CPEM.2016.7540632)
2013 Differential sampling measurement of a 7 V RMS [155] Palafox L, Behr R, Kieler O, Lee J, Budovsky I, Bauer S
sine wave with a programmable Josephson voltage and Hagen T 2016 First metrological applications of
standard IEEE Trans. Instrum. Meas. 62 1587–93 the PTB 1 V Josephson arbitrary waveform synthesizer
[141] Lee J, Behr R, Palafox L, Katkov A, Schubert M, Starkloff M 2016 Conf. on Precision Electromagnetic Measurements
and Böck A C 2013 An ac quantum voltmeter based (CPEM 2016) (IEEE) (https://doi.org/10.1109/
on a 10 V programmable Josephson array Metrologia CPEM.2016.7540602)
50 612–22 [156] Toonen R C and Benz S P 2009 Non-linear behavior of
[142] Schubert M, Starkloff M, Lee J, Behr R, Palafox L, electronic components characterized with precision
Wintermeier A, Boeck A C, Fleischmann P M and May T multitones from a Josephson arbitrary waveform
2015 An AC josephson voltage standard up to the kilohertz synthesizer IEEE Trans. Appl. Supercond. 19 715–8
range tested in a calibration laboratory IEEE Trans. [157] Benz S P, Dresselhaus P D, Tew W L, White D R and
Instrum. Meas. 64 1620–6 Martinis J M 2003 Johnson noise thermometry
[143] Chen S-F, Amagai Y, Maruyama M and Kaneko N 2015 measurements using a quantized voltage noise source for
Uncertainty evaluation of a 10 V RMS sampling calibration IEEE Trans. Instrum. Meas. 52 550–4
measurement system using the AC programmable [158] White D R and Benz S P 2008 Constraints on a synthetic-
josephson voltage standard IEEE Trans. Instrum. Meas. noise source for Johnson noise thermometry Metrologia
64 3308–14 45 93–101
[144] Amagai Y, Maruyama M, Shimazaki T, Yamamori H, [159] Qu J, Benz S P, Coakley K, Rogalla H, Tew W L, White R,
Fujiki H and Kaneko N 2016 Characterization of high- Zhou K and Zhou Z 2017 An improved electronic
stability AC source using AC-programmable Josephson determination of the Boltzmann constant by Johnson noise
voltage standard system 2016 Conf. on Precision thermometry Metrologia 54 549–58

S172
Metrologia 55 (2018) S152 A Rüfenacht et al

[160] Urano C, Yamazawa K and Kaneko N-H 2017 Measurement Measurements (CPEM 2016) (IEEE) (https://doi.
of the Boltzmann constant by Johnson noise thermometry org/10.1109/CPEM.2016.7540454)
using a superconducting integrated circuit Metrologia [169] Hamilton C A 2001 NCSLI Recommended Intrinsic Derived
54 847–55 Standards Practices—1 (RISP-1): Josephson Voltage
[161] Flowers-Jacobs N E, Pollarolo A, Coakley K J, Fox A E, Standard 4th edn (Boulder, CO : NCSLI)
Rogalla H, Tew W L and Benz S P 2017 A Boltzmann [170] Solve S 2012 Protocol of direct on-site Josephson voltage
constant determination based on Johnson noise standard comparisons: BIPM.EM-K10.b (http://kcdb.
thermometry Metrologia 54 730–7 bipm.org/appendixB/appbresults/BIPM.EM-K10/BIPM.
[162] Bauer S, Behr R, Hagen T, Kieler O, Lee J, Palafox L and EM-K10_Technical_Protocol_option_A.pdf)
Schurr J 2017 A novel two-terminal-pair pulse-driven [171] Solve S and Stock M 2012 BIPM direct on-site Josephson
Josephson impedance bridge linking a 10 nF capacitance voltage standard comparisons: 20 years of results Meas.
standard to the quantized Hall resistance Metrologia Sci. Technol. 23 124001
54 152–60
[172] Yamada T, Urano C, Nishinaka H, Murayama Y, Iwasa A,
[163] Overney F, Flowers-Jacobs N E, Jeanneret B, Rüfenacht A,
Fox A E, Underwood J M, Koffman A D and Benz S P Yamamori H, Sasaki H, Shoji A and Nakamura Y 2009 A
2016 Josephson-based full digital bridge for high-accuracy direct comparison of a 10 V Josephson voltage standard
impedance comparisons Metrologia 53 1045–53 between a refrigerator-based multi-chip programmable
[164] Flowers-Jacobs N E, Rüfenacht A, Fox A E, Dresselhaus P D system and a conventional system Supercond. Sci. Technol.
and Benz S P 2016 2 V pulse-driven Josephson 22 095010
arbitrary waveform synthesizer 2016 Conf. on Precision [173] Honghui L, Yuan G and Zengmin W 2015 Comparison of
Electromagnetic Measurements (CPEM 2016) (IEEE) the 10 V josephson voltage standards of the conventional
(https://doi.org/10.1109/CPEM.2016.7540601) and the programmable at NIM 2015 12th Int. Conf. on
[165] Jeanneret B, Rüfenacht A, Overney F, Van Den Brom H and Electronic Measurement & Instruments (ICEMI) (IEEE)
Houtzager E 2011 High precision comparison between pp 713–7
a programmable and a pulse-driven Josephson voltage [174] Tang Y-H, Wachter J, Rüfenacht A, FitzPatrick G J and
standard Metrologia 48 311–6 Benz S P 2015 Application of a 10 V programmable
[166] Rüfenacht A, Flowers-Jacobs N E, Fox A E, Burroughs C J, Josephson voltage standard in direct comparison with
Dresselhaus P D and Benz S P 2016 Direct comparison of a conventional Josephson voltage standards IEEE Trans.
pulse-driven Josephson arbitrary waveform synthesizer and a Instrum. Meas. 64 3458–66
programmable Josephson voltage standard at 1 V 2016 Conf.
on Precision Electromagnetic Measurements (CPEM 2016) [175] Rüfenacht A, Flowers-Jacobs N, Fox A E, Waltman S B,
(IEEE) (https://doi.org/10.1109/CPEM.2016.7540603) Schwall R E, Burroughs C J, Dresselhaus P D and
[167] Kieler O F, Behr R, Schleussner D, Palafox L and Benz S P 2018 DC comparison of a programmable
Kohlmann J 2013 Precision comparison of sine waveforms Josephson voltage standard and a Josephson arbitrary
with pulse-driven Josephson arrays IEEE Trans. Appl. waveform synthesizer 2018 Conf. on Precision
Supercond. 23 1301404 Electromagnetic Measurements (CPEM 2018) (IEEE)
[168] Nissila J, Sira M, Lee J, Ozturk T, Arifovic M, de submitted
Aguilar J D, Lapuh R and Behr R 2016 Stable arbitrary [176] Parks H V, Tang Y, Reese P, Gust J and Novak J J 2013
waveform generator as a transfer standard for ADC The North American Josephson voltage interlaboratory
calibration 2016 Conf. on Precision Electromagnetic comparison IEEE Trans. Instrum. Meas. 62 1608–14

S173