Technical Note

Solutions from HP EEsof

Mixer Simulation with HP Advanced
Design System
Introduction
Mi xers are nonl i near ci rcui ts used to transl ate si gnal s from one frequency to another.
They are characteri zed by how much gai n or l oss they i ntroduce i n thi s conversi on, as wel l
as how much di storti on and noi se they contri bute. For thi s reason, accurate nonl i near
devi ce model s are essenti al to getti ng good resul ts, but obtai ni ng these model s i s not
di scussed here. Thi s techni cal note covers some of the detai l s requi red to si mul ate mi xers
wi th HP Advanced Desi gn System. Unl ess otherwi se noted, al l of the si mul ati on set-ups
and data di spl ays are i ncl uded i n the HP ADS 1.0 exampl e project examples/RFIC/
Mixers_prj. A desi gner coul d easi l y copy and modi fy parts of thi s project i n order to run
the same si mul ati ons on hi s or her own desi gn. Note that some of these si mul ati ons are
appl i cabl e to recei vers as wel l as mi xers. Si mul ati ons di scussed i ncl ude:
DC anal ysi s and back-annotati on to the schemati c (p. 2)
Conversi on gai n (p. 3)
Noi se gure (p. 3, 11)
Thi rd-order i ntercept (TOI ) poi nt (p. 5)
RF-to-I F gai n compressi on (p. 7)
Conversi on gai n and noi se gure versus RF frequency (p. 9)
Total noi se power and noi se power spectral densi ty at the mi xer output (p. 12)
Spuri ous-free dynami c range (p. 13)
I ntermodul ati on di storti on, i nput- and output-referred TOI , conversi on gai n, and
l eakage versus RF i nput power (p. 14)
I ntermodul ati on di storti on, i nput- and output-referred TOI , conversi on gai n, and
l eakage versus LO i nput power (p. 16)
Thi rd-order i ntercept poi nt and conversi on gai n of a two-stage down-converti ng
recei ver (p. 17)
I mage rejecti on of a two-stage down-converti ng recei ver (p. 19)
2
Gilbert Cell Mixer Simulation
The si mul ati ons descri bed are appl i ed to the Gi l bert cel l mi xer [1] shown i n Fi gure 1.
Note that the si mul ati ons are done on a ni shed mi xer, but the anal yses coul d be used to
determi ne how the desi gn coul d be i mproved. Thi s Gi l bert cel l mi xer i s i n both MDS and
Seri es I V. I t i s the I AM-81018 mi xer, ori gi nal l y desi gned by two engi neers at Avantek [2].
Thi s part i s i n the HP Communi cati ons Components catal og [3]. Note that package
parasi ti cs are not i ncl uded i n thi s exampl e.
Figure 1. Gilbert cell mixer used in the simulations described in this document.
DC Simulation and Back Annotation
The schemati c i n the exampl e l e, DCTests, shows the DC si mul ati on of the mi xer. Note
that the new ADS back annotati on feature al l ows the desi gner to see the DC vol tages at
each node, as wel l as the DC currents i n each devi ce termi nal . Thi s feature i s accessed by
rst runni ng a DC si mul ati on, then openi ng the mi xer subci rcui t, and then sel ecti ng
Simulate/ Annotate DC Solution. The DC vol tage at each node as wel l as the current
i n each branch, as shown i n Fi gure 2, i ndi cates whether each devi ce i s bi ased correctl y or
not.
3
Figure 2. Back annotation of the DC solution to a schematic.
Conversion Gain and Noise Figure
The schemati c, ConvGain shown i n Fi gure 3, shows how to set up and si mul ate
conversi on gai n and noi se gure at a si ngl e frequency. The conversi on gai n i s cal cul ated
on the schemati c, so thi s parameter coul d be opti mi zed, i f desi red. The RF, LO, and I F
frequenci es as wel l as the RF power are al l dened as vari abl es, maki ng i t easi er to
change val ues and rerun the si mul ati on i f necessary. The two harmoni c bal ance
fundamental frequenci es are dened as the LO and RF frequenci es, respecti vel y. The RF
si gnal i s smal l , so i ts order i s just set to 1. The harmoni c bal ance control l er has been
congured to compute noi se gure. I n order for thi s computati on to be done, i t i s
necessary to speci fy the i nput and output ports as wel l as the i nput frequency (I nputFreq)
and output frequency (FreqForNoi se). Wi th these setti ngs, the si mul ator computes the
correct conversi on gai n, whi ch i s necessary to compute the noi se gure. I t i s not
necessary to speci fy a noi se node, unl ess the user wants to see the noi se vol tage computed
at a parti cul ar node. For noi se si mul ati on, onl y nodes speci ed here wi l l be output to the
dataset, not al l named nodes. The Kryl ov sol ver i s used i n thi s si mul ati on (on the
Harmoni c Bal ance control l er, set UseKryl ov=yes) i n order to speed i t up, and i t i s qui te
fast, requi ri ng onl y about 15 seconds on an HP 9000 Seri es C180 workstati on.
The noi se gure to each port i s computed automati cal l y. I n thi s case onl y the noi se gure
to port 2 (nf(2)) i s of i nterest. The conversi on gai n i s computed as the power at the I F port
at the I F frequency mi nus the power avai l abl e from the RF source. Thi s can be computed
usi ng the measurement equati on shown, ConvGai n2. Power_RF can be used i n thi s
equati on as l ong as i t i s i ncl uded as an OutVar i n the harmoni c bal ance control l er bl ock.
The mi x() functi on i s used to speci fy whi ch spectral tone at the mi xer output i s the desi red
one. I n thi s case, the mi x(Vi f,{-1,1}) functi on speci es the tone at -1*(Freq[1] i n the HB
anal ysi s control l er) +1*(Freq[2] i n the HB anal ysi s control l er), or -1*LOfreq +1*RFfreq.
The si mul ati on resul ts are shown i n Fi gure 4.
4
Figure 3. Mixer conversion gain and noise gure simulation set-up
Figure 4. Mixer output spectrum, conversion gain, and noise gure results. The conversion gain can
be computed on the schematic page or here on a data display page, yielding identical results.
5
Note that the i nput si gnal s coul d be speci ed as vol tages rather than powers, and the
conversi on gai n coul d al so be computed usi ng vol tages rather than powers. Al so, smal l -
si gnal mi xer mode coul d be used for thi s type of si mul ati on, whi ch al l ows a reducti on by
one of the number of l arge-si gnal tones si mul ated i n the ci rcui t, when the tone that i s
bei ng el i mi nated can be assumed to be smal l rel ati ve to the other tones. There i s an
exampl e showi ng how to set up such a si mul ati on i n the exampl es/RFI C/
Mi xerDi ffMode_prj exampl e project.
The schemati c, ConvGain_wFilt, i s the same set-up as ConvGai n, except that an i mage-
rejecti on l ter has been added at the i nput, and a bandpass l ter at the I F has been
added.
Third-Order Intercept Point
The schemati c, MixerTOI shown i n Fi gure 5, shows how to set-up a thi rd-order
i ntermodul ati on di storti on si mul ati on. I n thi s si mul ati on, two tones are suppl i ed at the
RF port. Nonl i neari ti es i n the mi xer generate thi rd-order i ntermodul ati on di storti on, and
the thi rd-order i ntercept poi nt i s a measure of how much di storti on i s generated. Thi s
si mul ati on i s very fast, wi th UseKryl ov set to yes.
Figure 5. Mixer Third-Order Intercept (TOI) simulation set-up.
Thi s si mul ati on set-up i s si mi l ar to the one for conversi on gai n, except that now the i nput
si gnal has two l arge-si gnal tones separated by fspaci ng. The measurement equati on
cal cul ates the output-referred thi rd-order i ntercept poi nt, conversi on gai n, and i nput-
referred thi rd-order i ntercept poi nt. A bui l t-i n functi on, I p3_out() uses notati on si mi l ar to
the mi x() functi on to speci fy the desi red output si gnal as wel l as the undesi red
6
i ntermodul ati on tone. I n a more real i sti c si mul ati on, the i nput and output l ters shoul d
be repl aced by l ess i deal ones, or measured S-parameters coul d be used.
Fi gure 6 shows the output spectrum wi th i ntermodul ati on di storti on products at the I F
frequency. Markers have been added, and the output-referred thi rd-order i ntercept poi nt
has been cal cul ated based on si mpl e geometry and the assumpti on that the
i ntermodul ati on products wi l l ri se three ti mes as fast as the desi red I F tones, for each dB
i ncrease i n RF power. The TOI cal cul ati ons are i denti cal , whether done on the schemati c
page or usi ng markers here on the data di spl ay page. The output-referred thi rd-order
i ntercept poi nt i s cal cul ated here usi ng the marker readouts and the di spl ay shows that
the measurement functi on on the schemati c cal cul ates the i denti cal resul t. The i nput-
referred thi rd-order i ntercept poi nt i s just the output-referred thi rd-order i ntercept poi nt
mi nus the conversi on gai n. (The cal cul ati ons are shown graphi cal l y i n Fi gure 7).
Figure 6. Mixer TOI Simulation Results
Note that when usi ng thi s approach for computi ng TOI , the user shoul d veri fy that the
upper and l ower i ntermodul ati on di storti on terms i n Fi gure 6 are the same ampl i tude. I f
they are not, the number of harmoni cs of each i nput tone (i n parti cul ar the LO) and/or the
MaxOrder setti ng i n the Harmoni c Bal ance control l er shoul d be i ncreased. Al so, thi s
extrapol ati on assumes that the mi xer nonl i neari ti es generate i ntermodul ati on di storti on
terms that track the i deal sl ope of Fi gure 7, whi ch may not al ways be the case. I t may be
necessary to do a more thorough si mul ati on, i n whi ch the RF power i s swept. Thi s i s
shown i n the Intermodulation Distortion, Input- and Output-Referred TOI,
Conversion Gain, and Leakage versus RF Input Power secti on bel ow.
7
Figure 7. Output-referred TOI is calculated by extrapolating the results of a low input power
simulation.
RF-to-IF Gain Compression
The schemati c, RFIFcompression i n Fi gure 8, shows an easy way to determi ne the 1 dB
gai n compressi on of the mi xer, usi ng the new Gai n Compressi on anal ysi s component. The
Gai n Compressi on component uses harmoni c bal ance to automati cal l y sweep the i nput
si gnal power l evel unti l the conversi on gai n from the speci ed i nput port and frequency to
the speci ed output port and frequency i s reduced from i ts smal l -si gnal val ue by the
speci ed number of dB. The si mul ati on resul ts are shown i n Fi gure 9.
I nput-Referred TOI
I nput Power (dBm)
Output Power (dBm)
Output-Referred TOI
Output-Referred TOI = m1+ 0.5(m1-m2)
Desi red output si gnal
sl ope=1:1
3rd-order
i ntermod si gnal
sl ope=3:1
m2
m1
+
+
8
Figure 8. RF-to-IF gain compression simulation set-up.
Figure 9. RFIFcompression.dds window showing results of RF-to-IF conversion gain simulation.
The conversion gain when the RF power is small (ConvGain_linear) is different from in Figure 4
because RF and IF lters are included here, and the order (number of harmonics) of the RF signal is
2 instead of 1.
9
Conversion Gain versus Swept RF Frequency
The schemati c, SweptRF i n Fi gure 10, si mul ates the conversi on gai n of the mi xer as the
RF frequency i s swept from 200 MHz to 7 GHz.
I n thi s case, the RF frequency i s dened as a vari abl e, Frf, and i t i s swept i n the harmoni c
bal ance control l er. The I F frequency, Fi f, i s al so dened to be a constant at 70 MHz, and
the LO frequency i s set to one I F bel ow the RF frequency. Two sweep pl ans are used, to
attai n a ner-resol uti on sweep when the RF i s bel ow 1 GHz, and a coarser sweep when
the RF i s above 7 GHz. The conversi on gai n i s computed usi ng the measurement
equati ons shown, and i t i s output to the dataset. The conversi on gai n versus RF frequency
i s shown i n Fi gure 11.
Fi gure 12 shows correct and i ncorrect ways to compute the power of the I F tone. HP ADS
outputs a spectrum at each named node on the schemati c. However, to compute the
conversi on gai n, we need to know the power at the I F frequency, 70 MHz. Thi s i s al ways
the RF frequency mi nus the LO frequency, and the correct way to speci fy thi s spectral
component i s to use the mi x functi on, as shown i n the measurement equati on for Vi fTone
i n Fi gure 10, and i n the equati on for I F_tone i n Fi gure 12.
Figure 10. Simulation set-up for swept RF, constant IF conversion gain simulation.
10
Figure 11. Conversion gain versus RF input frequency.
Figure 12. Two ways of specifying spectral components. The mix() function, while more
complicated, avoids selecting the wrong spectral component, which can occur if explicit indexing is
used, as in the equation for IF_toneOld.
11
Noise Figure versus RF Frequency
The schemati c, SweptRF_NF shown i n Fi gure 13 bel ow, si mul ates the noi se gure of the
mi xer as a functi on of RF frequency, wi th the I F xed at 70 MHz. The si mul ati on set-up i s
very cl ose to that shown i n Fi gure 10 above, except that the l arge-si gnal RF tone has been
repl aced by a Term component, and the harmoni c bal ance control l er has been congured
to cal cul ate noi se gure. Removi ng the l arge-si gnal tone from the RF i nput speeds up the
si mul ati on and shoul d have very l i ttl e effect on the computed resul t provi ded the
ampl i tude of the RF i nput si gnal that i s removed i s at l east 10 dB bel ow the ampl i tude of
the LO. Al ternati vel y, noi se gure can be computed usi ng the Fi gure 10 si mul ati on set-
up, just by conguri ng the harmoni c bal ance control l er to cal cul ate noi se. Noi se gure i s
cal cul ated usi ng the fol l owi ng equati on:
Where Na i s the noi se added by the mi xer
kTB i s the noi se power avai l abl e from the i nput noi se source, and
G i s the conversi on gai n from RF to I F.
Figure 13. Set-up for simulating noise gure versus RF frequency. Without a large-signal tone at the
RF input, this simulation is faster than a set-up that includes the RF input tone.
The si mul ator computes Na, whi ch i ncl udes the noi se that i s frequency-transl ated to the
I F by the l arge-si gnal LO tone (and any other l arge-si gnal tones.) The conversi on gai n, G,
i s al so computed, and kTB i s computed from Bol tzmanns constant, the anal ysi s
temperature i n degrees Kel vi n, and the user-settabl e noi se bandwi dth (usual l y 1 Hz.)
Note that thi s computed noi se gure assumes that there i s onl y one i nput frequency that
NF = ____________________________________________________________
Noise Power at output due to noise at RF input at RF frequency
Total Noise Power at output port at I F frequency
NF = ____________
Na + kTBG
kTBG
12
transl ates to the I F, unl i ke i n an actual measurement, where i t woul d be necessary to
i ncl ude an i mage-rejecti on l ter at the i nput.
Fi gure 14 shows the resul ts of the si mul ati on. Note that nf(2) i s a response wi th two
i ndependent vari abl es, Frf and noi sefreq. Addi ng the [0] syntax el i mi nates the noi sefreq
i ndependent vari abl e (i t has onl y one val ue anyway), enabl i ng the noi se gure to be
pl otted and l i sted more conveni entl y.
Figure 14. Noise Figure versus RF input frequency.
Output Noise Power and Spurious-Free Dynamic Range
The schemati c, NoiseFloor, si mul ates the total noi se power at the output of the mi xer.
Thi s resul t i s combi ned wi th the Mi xerTOI si mul ati on resul t to cal cul ate the spuri ous-
free dynami c range. Note that i t i s possi bl e to compute the spuri ous-free dynami c range
from just a si ngl e si mul ati on, but by runni ng the noi se si mul ati on separatel y, onl y a
si ngl e l arge si gnal tone (for the LO) i s requi red, l eadi ng to a much faster noi se si mul ati on.
Fi gure 15 shows the si mul ati on set-up.
Figure 15. Simulation set-up for calculating output noise power.
13
The noi se vol tage spectral densi ty wi l l be computed i n a 10 MHz bandwi dth centered on
the I F frequency, i n 0.1 MHz steps. The total noi se power at the output i n thi s bandwi dth
wi l l be computed, and thi s resul t al ong wi th a previ ous TOI si mul ati on wi l l be used to
compute the spuri ous-free dynami c range. Fi gure 16 shows the noi se spectral densi ti es
and noi se output power cal cul ati ons.
Figure 16. Noise spectral density plots and output power calculations.
There are several thi ngs to poi nt out concerni ng Fi gure 16. Fi rst, i f there i s onl y one
speci ed noi se node i n the si mul ati on, the dataset wi l l just have an entry cal l ed noi se,
whi ch i s the noi se vol tage spectral densi ty at thi s node. Thi s val ue i s pl otted on the l eft i n
Fi gure 16. I n order to cal cul ate the total noi se output power, we must rst convert thi s to
a power spectral densi ty. Si nce ADS outputs rms noi se vol tages (unl i ke resul ts from
harmoni c bal ance anal yses whi ch are peak vol tages), and the output resi stance i s 50
ohms, the noi se power spectral densi ty i s computed by squari ng the noi se vol tage and
di vi di ng by 50. The resul t i s pl otted on the ri ght i n Fi gure 16, where the trace type has
been set to hi stogram. The total noi se output power i s computed by summi ng the areas
under each of the rectangl es, si nce there i s no bui l t-i n i ntegrati on functi on i n ADS 1.0.
The Noi sePowerTotal _dBm equati on carri es out thi s summati on and converts the power
i nto dBm. Note that i f the si mul ati on i s rerun wi th a di fferent frequency step si ze, the
new step si ze wi l l have to be entered i nto thi s equati on. (I t woul d be better to make the
step si ze a vari abl e and pass i t i nto the dataset vi a the Other=OutVar= syntax on the
Harmoni c Bal ance control l er, as shown i n Fi gure 5.)
Calculating Spurious-Free Dynamic Range
One deni ti on of spuri ous-free dynami c range i s the di fference between a desi red I F
si gnal at the output and the noi se oor at the output, when the spuri ous i ntermodul ati on
di storti on si gnal s generated by the mi xer or recei ver have the same power as the noi se
oor. So i t i s necessary to combi ne resul ts from both a noi se oor si mul ati on and an
i ntermodul ati on di storti on si mul ati on to compute the spuri ous-free dynami c range.
Fi gure 17 shows the cal cul ati on. The dynami c range cal cul ati on woul d be just the
di fference between the two markers i f marker two where at the same power l evel as the
noi se oor. I n general , thi s i s not the case. So we assume that for each dB change i n the
i nput si gnal s to the mi xer, the si gnal at marker m1 i ncreases 1 dB and the si gnal at
14
marker m2 i ncreases 3 dB. Based on thi s we can determi ne where marker m1 wi l l be
when the spuri ous si gnal at marker m2 has the same power l evel as the noi se power. Thi s
cal cul ati on i s carri ed out i n Fi gure 17.
Figure 17. Spurious-free dynamic range calculation.
Intermodulation Distortion, Input- and Output-Referred TOI,
Conversion Gain, and Leakage versus RF Input Power
The schemati c, IMDRFSwpHB, si mul ates the mi xer's i ntermodul ati on di storti on,
conversi on gai n, RF-to-I F l eakage, LO-to-RF l eakage, and LO-to-I F l eakage versus RF
i nput power, usi ng harmoni c bal ance. The si mul ati on set-up i s shown i n Fi gure 18. Thi s
set-up i s very si mi l ar to the TOI si mul ati on set-up i n Fi gure 5, except that Norton
equi val ent sources are used for the RF and LO sources. Thi s reduces the number of nodes
i n the ci rcui t by one per source, whi ch l eads to a sl i ghtl y faster si mul ati on ti me. Al so, the
power of each of the RF i nput tones i s swept.
Figure 18. Figure 18: Simulation set-up for intermodulation distortion, input- and output-referred
TOI, conversion gain, and leakage versus RF input power
Fi gure 19 shows the TOI and conversi on gai n si mul ati on resul ts. The l ower l eft pl ot
shows one of the desi red si gnal s and one of the undesi red i ntermodul ati on di storti on
si gnal s at the output, versus RF i nput power. The i nput- and output-referred TOI poi nts
15
are computed vi a si mpl e geometry as descri bed i n the Third-Order Intercept Point
secti on above. The TOI poi nts can be computed ei ther on the schemati c or on the data
di spl ay. As shown here, the resul ts are the same.
Figure 19. TOI and conversion gain simulation results.
Fi gure 20 shows the RF-to-I F l eakage, LO-to-I F l eakage, and LO-to-RF l eakage as a
functi on of RF i nput power. As can be seen, these l eakage terms vary onl y sl i ghtl y wi th
RF power. Agai n, the mi x() functi on i s used to conveni entl y speci fy parti cul ar spectral
components for these cal cul ati ons.
Figure 20. Various leakage terms versus RF input power.
16
Intermodulation Distortion, Input- and Output-Referred TOI,
Conversion Gain, and Leakage versus LO Input Power
The schemati c, IMDLOSwpHB, si mul ates the mi xer's i ntermodul ati on di storti on,
conversi on gai n, RF-to-I F l eakage, LO-to-RF l eakage, and LO-to-I F l eakage versus LO
i nput power, usi ng harmoni c bal ance. I t i s i denti cal to Fi gure 18, except that the LO
power i s swept i nstead of the RF power. Fi gure 18 coul d be modi ed to sweep any other
parameter i n the ci rcui t, i nstead. Fi gure 21 shows that the TOI poi nts and conversi on
gai n do depend on the LO power. Fi gure 22 shows that the l eakage terms do vary
substanti al l y wi th LO power.
The schemati c, IMDRFSwpEnv si mul ates the mi xer's i ntermodul ati on di storti on versus
RF i nput power, usi ng the Ci rcui t Envel ope si mul ator. Whi l e thi s was a good approach
when usi ng HP MDS, thi s type of si mul ati on i s general l y faster wi th the New Harmoni c
Bal ance (Kryl ov) si mul ator, usi ng the set-up shown i n Fi gure 18.
Figure 21. TOI points and conversion gain versus LO power.
17
Figure 22. Various leakage terms versus LO power.
Third-order intercept point and conversion gain of a two-stage
down-converting receiver
The schemati c, DoubleConvHB, si mul ates a doubl e downconversi on recei ver's thi rd-
order i ntercept poi nt and conversi on gai n. The si mul ati on set-up i s shown i n Fi gure 23.
There are two RF i nput si gnal s, centered at 2 GHz and spaced 100 kHz apart, and two
LOs, one at 1800 MHz and a second at 180 MHz. Wi th the New Harmoni c Bal ance
(Kryl ov) si mul ator, thi s 4-tone si mul ati on takes onl y 100 seconds to run.
Fi gure 24 shows the ful l output spectrum, the spectrum near the I F, the output-referred
TOI poi nt, and the conversi on gai n.
18
Figure 23. Double-conversion receiver simulation set-up.
Figure 24. Double-downconversion receiver simulation results.
19
Image Rejection of a Two-Stage Down-Converting Receiver
The schemati c, DblConvImag, si mul ates the i mage rejecti on and conversi on gai n of the
two-stage downconverter. The i mage rejecti on i s determi ned al most enti rel y by the
rejecti on of the i nput l ter at the i mage frequency. The si mul ati on set-up i s i denti cal to
Fi gure 23, except that one RF i nput frequency i s set to RFfreq=2 GHz and the other i s set
to 10 kHz bel ow the i mage frequency (2*LOfreq1-RFfreq-(10 kHz)). The 10 kHz offset
term makes i t easi er to di sti ngui sh between the desi red si gnal and the i mage si gnal
appeari ng at the I F. Fi gure 25 shows these two si gnal s as wel l as a cal cul ati on of i mage
rejecti on and conversi on gai n.
Figure 25. Image rejection simulation results.
Filter Frequency Responses
The schemati c, RFBandFiltTest, si mul ates the RF l ter's frequency response (S21 and
group del ay), and the schemati c, FirstIFFiltTest, si mul ates the rst I F l ter's frequency
response.
Summary
There are numerous si mul ati ons that can be appl i ed to a mi xer or recei ver. These
schemati cs and data di spl ays show how to use HP Advanced Desi gn System to set up
vari ous si mul ati ons and di spl ay resul ts. Wi th the speed i mprovements wi th the New
Harmoni c Bal ance (Kryl ov) si mul ator, none of the si mul ati ons shown here requi red more
than a mi nute or two to run on an HP 9000 Seri es C180 Uni x workstati on. Users shoul d
be abl e to copy and modi fy thi s exampl e project to easi l y si mul ate the same resul ts on
thei r mi xers or recei vers.
20
References:
[1] B.Gi l bert, A Preci se Four-Quadrant Mul ti pl i er wi th Subnanosecond Response, I EEE
J ournal of Solid-State Circuits, Vol . SC-3, pp. 365-373, December 1968.
[2] James Whol ey and I ssy Ki pni s, Si l i con Bi pol ar Acti ve Mi xers, Applied Microwave,
Spri ng 1990, pp. 287-293.
[3] Hewl ett-Packard Co., Communications Components GaAs & Silicon Products
Designers Catalog, HP l i terature number 5091-4574E, 1993.
[4] Paul R. Gray and Robert G. Meyer, Analysis and Design of Analog I ntegrated Circuits,
Third Edition. Wi l ey, New York, 1993, Chapter 10.
[5] Hewl ett-Packard Co., HP EEsof Di vi si on, Low-Power Mi xer Desi gn Exampl e usi ng
HP Advanced Desi gn System, 1998.
[6] Hewl ett-Packard Co., HP EEsof Di vi si on, Desi gners Appl i cati ons Gui de, May, 1998.
Thi s i s a descri pti on of each exampl e i n HP Advanced Desi gn System, Rel ease 1.0, and i s
avai l abl e from your Hewl ett-Packard el d representati ve.
For more i nformati on, cal l your l ocal HP
sal es ofce l i sted i n your phone di rectory,
or an HP regi onal ofce l i sted bel ow for
the l ocati on of your nearest sal es ofce.
United States
Hewl ett-Packard Company
Test & Measurement
5301 Stevens Creek Bl vd.
Bui l di ng 51L-SC
Santa Cl ara, CA 95052 USA
800-452-4844
Hewl ett-Packard Company
29 Burl i ngton Mal l Road
Burl i ngton, MA 01803 USA
617-270-7000
Canada
Hewl ett-Packard Canada Ltd.
5150 Spectrum Way
Mi ssi ssauga, Ontari o
L4W 5G1 Canada
416-206-4725
Europe
Hewl ett-Packard
European Marketi ng Centre
P.O. Box 999
1180 AZ Amstel veen
The Netherl ands
J apan
Yokogawa-Hewl ett-Packard Ltd.
Measurement Assi stance Center
Takakura-cho, Hachi oji -shi , 9-1
Tokyo 192, Japan
81-426-48-0722
Latin America
Hewl ett-Packard
Lati n Ameri can Regi on Headquarters
5200 Bl ue Lagoon Dri ve
9th Fl oor
Mi ami , FL 33126 USA
305-267-4220
Australia/New Zealand
Hewl ett-Packard Austral i a Ltd.
31-41 Joseph Street
Bl ackburn, Vi ctori a 3130
Austral i a
13-1347 ext. 2902
61-3-272-2895 ext. 2902
Asia Pacic
Hewl ett-Packard
Asi a Paci c Ltd.
17-21/F Shel l Tower
Ti mes Square
1 Matheson Street
Causeway Bay, Hong Kong
8522-599-7070
Data subject to change
1998 Hewl ett-Packard Company