Introduction

The Handy Lab Buddy is a tool every ECE should have. The four features of this tool include a
talking voltmeter, logic probe, voltage averager, and frequency measurer. As a cheap and accurate
device that outputs whatever being measured through speakers, it's one of its kind and an essential
tool for lab work.

Summary

Have you ever tried to debug a circuit and just wished your voltmeter would talk to you? Or have you
ever been too lazy to turn on and calibrate the oscilloscope to measure the frequency? Well, this is the
device for you. There are four operating modes of the Handy Lab Buddy. The first mode is a
voltmeter that measures voltages up to 5V and says the voltage measured through speakers. The
second mode is a logic probe that beeps once in a low tone if the output voltage is low and beeps
twice in a higher tone if the output voltage is high. The third mode, a feature unique to this device, is
a voltage averager that takes multiple samples in succession (perhaps a noisy voltage signal) and
calculates the max, min, and mean voltage and speaks the values over a speaker. The fourth and final
mode measures frequency with great accuracy from about 10 Hz to 100 kHz and speaks the value
over a speaker. A device capable of all these functions and a "talking" ability at a very cheap cost is
not available in the market today and is very useful to the ordinary engineer not wanting to invest in
an oscilloscope. As very enthusiastic electrical engineers, we decided to build this device - simply for
our own benefit, as it helps tremendously when debugging circuits.

We used quite a few features of the Atmel Mega644 MCU to implement this system, such as speech
generation and output, multiple ADC conversions, the output of the pulse width modulator (PWM),
and some others.

High Level Design

Rationale and Sources:

In all honesty, this was a secondary project idea for us, originally suggested by Bruce Land. Our
original project idea, the Electronic Dartboard (discussed later), had tremendous problems with the
hardware in the later weeks of the design cycle that forced us to abandon it. As our working
efficiency and debugging experience increased dramatically in those frustrating few weeks, we
managed to start and finish this new project within the final days of the semester.

Background Math:
The first big challenge for us was to get sound bites from AT&T and convert them for appropriate
use for the MCU. Using the speech generation document by Bruce Land, we utilized his MATLAB
programs to downsample from 16 kHz to 8 kHz and create header C files. The compression used in
the MATLAB program was a differential pulse-code modulation (DPCM) with a 4:1 compression
which sends 2:1 derivative samples.

To calculate the voltage generated, we had to use the built-in analog to digital converter in the MCU.
Simple bit conversions were done, and we had to divide the 10-bit digital voltage output by 1024
over VREF (5V) to get a floating point value for the voltage. For the logic probe, the same thing was
done, except an additional comparison was made to a threshold. The third feature, calculating the
min, max, and average of samples of voltages, used pretty much the same principle as above except
used additional variables to store the previous mean, min, max and sample number of samples.
Frequency is measured using an elegant combination of software and hardware. Our frequency
hardware first filters out the DC component of any input, then reapplies a DC bias, resulting in an AC
waveform around the biasing point of our high frequency inverter. The output of this inverter is then
stabilized. The final wave is thus a clean square wave, which is then used to trigger the external
interrupt on the microcontroller. The software then responds to the interrupt, first synching the timers
to the waveform, then measuring the the period of the waveform. Finally, the period is converted into
frequency. Using this method, we can reliably measure frequencies from 10Hz to 100,000kHz.

There were two ISRs, one of them controlled the part of the speech played using a table, based on a
timer, and the other, an external interrupt, controls triggers when to measure the time elapsed from
the previous trigger, to measure frequency.

Logical Structure:

The user-interface of the Handy Lab Buddy is as demonstrated above. In the first two modes, after
the analog voltage measurement is done, the signal needs to only go through an ADC before output to
the LCD and to the speakers. For mode three, the voltage averager, the user presses the button on the
probe when he/she is ready to take samples, then releases it to stop taking samples. The display
shows the final output and it is automatically spoken. The fourth mode, frequency measurment, the
user first has to change the banana plug connection to the frequency measurement pin (because of
circuit isolation, described later in detail), then the signal passes through an inverter circuit to digitize
the analog signal, which then triggers the external frequency ISR on the microcontroller (described
later). After measurement in software, it is outputted onto the LCD and speakers. The four buttons
that determine the modes are operated by a basic debounce state machine.

Hardware/Software Tradeoffs:

The biggest tradeoff during this design project was our decision to completely scrap our original idea
and use this one. Our original idea was to create an Electronic Dartboard, where a player would draw
concentric circles on a whiteboard, a infrared camera (removed form a WiiMote) would then be
calibrated to the circles and display them on an LCD. Then, the user would "shoot" an IR dart (laser)
at the whiteboard, which the WiiMote camera would have picked up and drawn the location on the
LCD. However, the WiiMote IR camera was not reliable enough outside of it's original controller. On
top of that, we concluded the large LCD we were planning to use was fried after much testing. We
worked on these getting these two hardware components to work by going to lab a lot of extra times
overthree weeks to no avail. Thus, due to faulty hardware and being incredibly pressed for time, we
decided to switch to this project, which our professor had initially suggested in class. In retrospect,
we're glad we made that decision, as we were able to come out with a complete, working product in a
short span of time. The detailed schematics and code for the old project are described later.

One tradeoff we faced was during the design of our frequency measurer. We wondered how to
calculate the frequency. One way of doing it was to use an Fast Fourier Transform algorithm as
linked on the course website. However, we realized that the DFT/FFT would take up too much
processing power and memory because of the look up tables. Besides, our frequency measurement
only needs to measure the most prominent frequency, and the DFT/FFT would be doing unnecessary
computation. Thus, we decided to take more of a hardware based approach, by digitizing the signal
by passing it through an inverter, then using an interrupt to calculate the frequency.

 Project Design
Hardware Design:

The hardware design for the Handy Lab Buddy was rather simple. To program the Atmel
ATMega644 Microcontroller, the STK500 development board was used. The Mega644 MCU was
then put on the Custom PC Board, provided by the lab. Soldering components onto the PC Board had
to be done with careful precision, as the pins were small and very compactly laid out.

Our other significant pieces of hardware in this project were the 2-line LCD display in lab, the 4
buttons, a speaker, an inverter, and probes (with another button). The overall interfacing of the MCU
with the rest of the hardware is shown below in the schematic.
Buttons

Buttons 1-4 were connected to port B7 to B4 (respectively) of the MCU. The top pin was grounded,
while each of the other pins had a 10kΩ connected to Vcc. The resistor was needed there because
when a button is pressed, the wire is shorted - to prevent this a resistor was placed there. These
buttons operated on a debounce state machine, described later in the Software Design section.

Another button was used as an interface for the voice output . This was also connected to a 10kΩ
resistor to Vcc, and to port B0. When this button was pressed, depending on what mode the system
was in, the system would speak whatever was on the LCD screen.

Frequency, Inverter, and Protector Circuit:

If a signal was to be measured for it's frequency, it needed to go through a circuit first to be
processed. First of all, the diode/resistor connected to Vcc in the schematic was just to protect the
MCU port A0 from being fried in case a voltage greater than 5V is fed into port A0. The signal
picked up by the probe goes through a 0.4uF total capcitance array to filter out DC values, then gets
DC shifted by the voltage divider, before going through the inverter to shift the center voltage of the
signal. The purpose of the inverter is to output a square wave from any periodic signal given. There
were a few ways of accomplishing this, one using an inverter or another using a Schmitt Trigger op
amp circuit. We chose the inverter because there was no analog circuitry to worry about. Thus, the
"analog" wave gets converted to a digital wave, as it converts every point on the wave below the
2.5V threshold to a logic 0 and above the threshold to a logic 1. The output of this input is then
stabilized by two 22 pF capacitors. The final wave going into port D3 is a clean square wave used to
trigger the external interrupt of the microcontroller.

Speakers:

The speakers are not shown in the schematic. They were simply connected via a headphone jack from
port B3 (output of the PWM) to ground.

Liquid Crystal Display:

The LCD used in this project was simply the 16-pin LCD used in ECE 4760 Spring 2010 Lab 1. The
pin assignments are as in the schematic.

Microcontroller Operation:

The Atmel ATMega644 microcontroller was the control module of the device. Port A0 was the ADC
used to implement the first three modes. Port D3 was used to implement the frequency mode. Port B0
was the button on the probe, that made the speaker output a voice signal, which was coming from
port B3. Ports C0-C2 and C4-C7 were used for the LCD. The diagram shows how the microcontroller
was laid out on the custom PCB board. The photograph after the probes section shows the completed
target board connected to the rest of the circuit. Two 1 uF capacitors were also used to connect Vcc
and Gnd to remove any external noise in the circuit as a whole.

Probes:

There were three probe inputs used: one to measure voltages, one for frequency, and the black was
always connected to ground as a reference. The input probe, depending on whether the device was
measuring frequency or voltage, would go on either the green connect or red connect (respectively).
This can be seen from the completed hardware circuit below.
 

Software and Program Design:

Speech Generation:

Speech generation was done exactly as outlined in Bruce Land's document. Once the voice files were
obtained and compressed using MATLAB, the header files generated were used to output the sound
later. The voices are all stored through the header file voice.h (attached at the end), which controls
what is being outputted to the PWM with different inputs.

Internal ISR:

This interrupt service routine was mainly for using timer2 to read voices from the table and output
them to the speakers using the PWM on B3.

Buttons Anti-Bounce:

In mode 1, if the probe button is pressed, the voltage displayed on the LCD is outputted to the
speakers. In mode 2, if the probe button is pressed, the a single beep is outputted if the voltage is
below a threshold of 2V or a double beep is outputted if the votlage is above the 2V threshold. In
mode 3, if the probe button is pressed once, then the state machine sends a signal to start collecting
data and update the min, max, and average values, then nothing happens when it's released, but stops
the data collecting when the button is pressed again, and outputs the data to the speaker. In mode 4,
only when the button is released is the data collected outputted to the speaker. The anti-bounce for
the four buttons determining is rather simple. If a button was pressed and not down before, then the
button counter variable incremented, so when it's released it triggers the functions in that mode.
External Interrupt ISR:

This external interrupt is simply for the fourth function of the device, the frequency measurer. If the
fourth button is pressed, then it triggers this ISR, which times the next period of the square wave, and
calculates the frequency. We also needed an overflow ISR because at the beginning we were having
problems with read the correct value for less than 250 Hz.

main() Function:

This function basically deals with the overall structure of the program. First of all, INT1 is turned on
for frequency detection. The PWM is turned on with a period of 256 cycles. Timer2 is set to overflow
at 7812 Hz (overflow ISR is turned on). Timer1 is also turned on for the frequency count (and
overflow ISR). For the voltage measurements, the ADC converters are initialized and a conversion is
enabled. The final part of the main() function has the outputs to the LCD at each stage. The button
state machine is also run multiple times with a delay of 20 ms (for the modes 3 and 4 primarily).

LCD:

The LCD used the same header file from the previous labs, and the same functions to write on the
LCD.

 Results

Speed of Execution:

Each mode of the Handy Lab Buddy was executed quickly. The LCD was able to recognize a change
in voltage very fast, and if the probe button was pressed, the speaker outputted the reading instantly.
For mode 3, our voltmeter took samples every 20ms and updated almost instantly on the screen the
max, min, and mean values for the number of samples at that moment. For the frequency mode, the
optimum range of operation was from 10Hz to 100kHz. If you were measuring out of that range, the
LCD would take about 1 or 2 seconds to update the value being fed in. However, the speaker
outputted whatever was on the screen instantly.

Accuracy:

Our voltmeter was very accurate to the hundredth value. A table of voltages measured by our
voltmeter in mode 1 versus the same voltages measured by the oscilloscope are shown below:

Our Voltmeter Oscilloscope % Error

(hundredth)

0.27V 0.25V 8%
1.04V 1V 4%
1.99V 2V 0.5%
2.5V 2.5V 0%
2.97V 3V 1%
4.05V 4V 1.25%
 4.73V 4.75V 0.42%

The error is slightly bigger for earlier values because they're more vulnerable to noise. The logic
probe (mode 2) was very accurate, and gave us 100% accuracy for every digital circuit we tested. The
variable voltage averager (mode 3) was more difficult to test, as there was no existing product to
compare it to, but the mean value responded as expected intuitively. Thus, our tests for mode 3 were
more qualitative, and deemed very successful. Finally, mode 4 was tested using the function
generator. Our frequency measure was very accurate from 10 Hz to 100 kHz (the operating range),
and then dropped off significantly after 100 kHz. The table below shows the error is actual and
measured values with our frequency measurer.

Our Frequency Measurer Function Generator % Error

No Signal (Too Low) 5 Hz N/A
10 Hz 10 Hz 0%
98 Hz 100 Hz 2%
996 Hz 1000 Hz 0.4%
20514 Hz 20000 Hz 2.5%
69323 Hz 70000 Hz 0.97%
105638 Hz 100000 Hz 5.6%
~135000 Hz 120000 Hz 12.5%

As you can see, the frequency measurer was very accurate for the operating range but dropped
significantly when the frequency was greater than 100 kHz.

Safety:

We took great care to see that our practices and final project met the highest standards for safety.
Every time we had to solder something onto the board, we would both wear safety goggles, make
sure the iron is heated appropriately and turned off when not needed. We would also always ground
ourselves when handling any circuitry, especially DIPs such as the Mega644s. As can be seen from
the final product, our soldering and wiring was very neat and properly laid out. This project doesn't
pose a significant harm to anyone else while in operation due to the low voltage input.

Interference:

We did not experience much interference when the Handy Lab Buddy was under voltage
measurement operation. However, when we were using the frequency mode, we could pick up the
60Hz noise from the lab environment.

Universal Usability:

This device is usable by any person who actively works in the lab. Since it measures only low
voltages, it does not pose much harm. Persons with special needs who can operate simple lab
voltmeters can use this device - and in fact get voice output and frequency measurement. So it is even
advantageous for some person to utilize this device instead of standard lab equipment.

 
Conclusions
Our talking voltmeter surpassed all expectations in accuracy and features implemented. However,
one thing we learned from this entire design project cycle is that there are always going to be serious
problems during the course of the design and you're going to have to make quick, appropriate
decisions regarding the future. Deciding to change our project completely was a wise decision on our
part, since we had no part of our original design completely working slightly more than a week
before our project was due. Although we felt frustrated that we had spent more than 3 weeks and 50
hours worth of work trying to get a simple IR camera and an LCD to work (simply the hardware), we
were satisfied with the final outcome of the design project. Another lesson also learned was the fact
that it is very difficult to reverse engineer a commercial product, such as the WiiMote. To pack that
many IC's and functionality into a small WiiMote package, the manufacturer use little tricks that are
impossible for us to capture. Although the schematics of how to use the IR camera and the I2C bus
were given online, we were unable to have a permanently working circuit within the alotted time (it
worked a few times and was unstable).

Old Project Design: WiiMote IR Camera/LCD Interface:

In our old design, we were reverse-engineering the IR camera on the WiiMote, and couldn't get it to
work. We were using the IR camera hack seen here, as well as the code for the I2C bus. For the LCD,
we were using the schematics and drivers written by a previous ECE 4760 Project, the Hard Drive
Clock, only to find very late in the game that the LCD was most likely fried, because it was not
processing code or even turning on.

LCD Drivers and Code:

Since the LCD we were using was the same one we used for labs in Spring 2010, we used the code
that was given to us (by Bruce Land). This included lcd.h and lcd.c (attached at the end).

Sound Bites:

The sound used for this project was obtained from the AT&T TextToSpeech website.

Future:

This project can easily be commercialzed for use in labs. It is useful for any electrical engineer, as the
talking capability comes in handy when debugging a complicated circuit and not being able to look at
the display. The averager and frequency meter are also one of it's kind features for this device, and
are definitely useful in lab. Thus, if properly bound and manufactured, this device has great potential.

Code:

lcd_lib.h, lcd_lib.c, voice.h, main.c