Lab Manual 04

Download as pdf or txt
Download as pdf or txt
You are on page 1of 60

EN162—Analysis and Design of Electronic Circuits

Spring 2004

LAB MANUAL

Wm. R. Patterson III


B.H. 449 ext. 3-1449
[email protected]
Engineering 162 Laboratory Manual Spring 2004

Table of Contents

General Information: ..................................................................................................................3


EXPERIMENT NO. 1- Oscilloscope Use..................................................................................7
EXPERIMENT NO. 2- Characterization of a Diode and a Transistor ....................................14
USE OF THE TEKTRONIX 575 TRANSISTOR CURVE TRACER................................19
EXPERIMENT NO. 3- Half a Cheap Stereo ...........................................................................22
EXPERIMENT NO. 4- Audio Frequency Amplifier ...............................................................25
EXPERIMENT NO. 5- Video Amplifier .................................................................................28
EXPERIMENT NO. 6- Sine Wave Oscillator .........................................................................33
EXPERIMENT NO. 7- Telephone Anti-Aliasing Filter ..........................................................40
EXPERIMENT NO. 8- Design of a Bipolar OTA Circuit.......................................................46
APPENDIX: DEVICE DATA SHEETS.................................................................................49

2
Engineering 162 Laboratory Manual Spring 2004

General Information:

Admonition: We have to ask that you be especially careful about neatness in the lab. The lab
can be a difficult experience for everyone if you don’t put things away when you are done.
There are around 30 people in the course and that number can make quite a mess. We try to
have enough parts, but we cannot accommodate your habits of dropping used materials wher-
ever you happen to be when finished. You are not done with a lab until your breadboard is
disassembled and shared parts are returned to their boxes. This is especially true of capacitors.

Introduction: There are eight labs set in this manual, two measurement labs and six labs for
which you design, build, and test a circuit to meet fixed performance criteria. You must com-
plete and document six laboratory exercises in order to get credit for the course. (Although
there are eight exercises set, I will base grades on only six reports. You will not receive extra
credit for extra reports.) You must do Lab 2 on the characterization of devices, but the others
may be your choice from all the experiments. THERE WILL BE NO EXCEPTIONS TO
THIS POLICY. In addition to this manual, you will receive a protoboard on which to assem-
ble your circuits, a pair of wire strippers, and a small packet of semiconductors. Please return
the breadboard and strippers at the end of the semester. Passive components and some semi-
conductors must be shared communally. We have labeled, multi-compartment cabinets and
boxes for storage of these materials. Your cooperation in returning parts to their proper
place is absolutely necessary for the success of the lab. Also, please do not keep components
longer than necessary to do your lab. Capacitors in particular are bulky and expensive so that
we do not keep enough available for everyone to have complete sets simultaneously.

The labs are done in the Hewlett Electronics Laboratory, room 196 of the new Giancarlo addi-
tion. The lab will be shared with Engineering 164 and Engineering 158 this semester. Equip-
ment and parts for this course will be on the benches farthest from the doors. This room is
open all the time during the semester. There does not have to be a lab instructor there for you
to work, but such assistance will be available eighteen or so hours per week. (The hours will
be announced on the class website.) You hand in lab reports in a box in the lab that is marked
"EN162 Lab Reports," and they will be graded by a TA under guidelines I set.

3
Engineering 162 Laboratory Manual Spring 2004

Collaboration: Design concepts and measurement techniques may be discussed freely. Each
student must build and test the circuits on his or her own protoboard. ALL lab work and all
written reports must be done individually. All data reported must be your own measurements
– copying data and representing it as yours is a cardinal sin among scientists of all types.

Concise, complete, polished lab reports will be rewarded with higher marks. A TA must wit-
ness the measurements you present to prove that you have met the lab specifications. He
or she will record the fact that you did the lab on a spreadsheet and will give you an ID number
to print with your data plots. While you can design, build and test your circuits without a TA
if you wish, the final run of measurements does require a TA for this recording process.

Lab reports must be typeset in a word-processing program. The format for the design lab
write-ups would be appropriate for an in-house report prior to a design review at an electronics
firm:

I. Introduction: Discuss briefly the major goals of the lab.


II. Schematic: Show component values and, where appropriate, the measured quiescent
node voltages.
III. Tables: Compare design versus measured values of all important quantities
IV. Design. Description of the design procedure and design equations. Prove that the
worst-case conditions are met by design. You do not have to show the algebra by
which you may have fit the equations, but you do have to show that a suitable equation
set is satisfied.
V. Questions. Answer the specific questions set in the handout.

For the characterization experiments, follow the directions in the lab handout.

For each experiment there is a deadline by which your report must be turned in. These will be
announced in class and posted on the class web site. Reports that are turned in after the dead-
line will have penalties assessed against them. The penalty may be as much as 10 % per week
off! Be forewarned -- your classmates also like to wait until the last minute to do their work,
so start early and try to get a bit ahead of the deadlines -- overcrowding in the lab will not be
accepted as an excuse for turning in lab reports late. Also remember the deadlines have been
set as late as practical. The labs are intended to teach, and you will have the benefit of that
knowledge earlier if you do them earlier. This is particularly true of doing some of the earlier
labs before the mid-semester exam, which comes a little late in the semester.

There are several general textbooks, which I will put on reserve in the Sciences Library, in-
cluding older texts by Neaman and Savant. Gray and Meyer's Analog Integrated Circuit De-
sign, now in its 4th edition, is also an excellent reference for some of the later topics in the
course. It is considered a classic in the field.
4
Engineering 162 Laboratory Manual Spring 2004

Several of the lab exercises, particularly Labs 4, 5 and 7 require measurements of the gain and
phase shift of amplifiers as functions of frequency. We have assembled a computer driven sys-
tem for automatic measurements of these quantities. You can take that data home either by
emailing a file or by putting it into your Engineering account. The TA still has to witness the
measurement and issue the ID number, and in doing so will check your circuit board serial
number. The parameter analyzer supplies a similar set of files with data for Lab 2, and labs 1
and 6 require making plots and FFT calculations from a digital oscilloscope. All files are sub-
ject to the same checkoff system and properly formatted printouts of them are a major part of
your reports.

Components: Of the various passive components in the lab, the capacitors require special
comment. There are three principal types: electrolytic capacitors, film capacitors and ceramic
disk capacitors. Electrolytic capacitors are made with aluminum or tantalum plates by electro-
lytically forming an insulating oxide on one plate. (If you don’t know what the term “electro-
lytically” means, then please either ask me or look it up.) Since that oxide is very thin, a large
capacitance per unit volume is possible. These are the only types of capacitor available for
you that have values over a microfarad. Unfortunately, the chemical reaction that forms the
oxide is reversible if the potential across the capacitor is of opposite polarity to the forming
voltage for an appreciable length of time. The capacitors are marked for the terminal that was
positive during manufacture. In use, one must be careful that the applied voltage is in the
same direction. Failure to do so will ruin the capacitor and make the circuit malfunction.
Aluminum electrolytics are often unsuitable for some high frequency applications because of
high series inductance. Another feature of electrolytic capacitors based on aluminum is that
the tolerance on their capacitance is very poor, typically -30% to +80% of nominal value.

Disk ceramic capacitors use simple blocks of a ceramic with a very high dielectric constant.
All are nearly ideal capacitors to quite high frequency (10's of Megahertz). However, inexpen-
sive types usually have poor tolerance (-20% to +50%) and very high temperature sensitivity.
They are used primarily where exact values of capacitance are not critical. A common appli-
cation for these is power supply bypass. There are a couple of large reels of bypass caps in the
lab – you should never lack for enough.

Film capacitors have generally good performance with tolerances no worse than +/-10%.
Their primary disadvantage is that they are usually physically large for a given value of capaci-
tance. You are likely to find capacitor tolerance a significant issue only for lab 7.

Resistors: Resistors have three principal ratings: resistance in ohms, tolerance in percent, and
power dissipation in watts. Most of the resistors we have in the lab can dissipate ¼ watt and
have tolerances of ± 5 %. They are manufactured only with certain standard resistance values
that are separated by about 10 %. We have most of these values and it usually does not make
5
Engineering 162 Laboratory Manual Spring 2004

sense to put resistors in series to come closer to calculated values. The attempt is illusory be-
cause the tolerance variation is bigger than the adjustment you would be trying to make. Re-
sistance values are coded on the resistors with color bands. The first two bands give two sig-
nificant digits of the value, the third band is a multiplier expressed as a power of 10, and the
fourth band is the tolerance. The table below shows the color code. For example, red-red-
orange-gold is 22 Kohms ± 5 %. Sometimes ¼ watt is not enough for a circuit application and
then a physically larger resistor is required. We have some 2-watt resistors in a relatively nar-
row range of values anticipating a need for such a device in Lab 5.

Significant Digit Value


Color (Band 1 or 2) Multiplier
Black 0 1
Brown 1 10
Red 2 100
Orange 3 1000
Yellow 4 10000
Green 5 100000
Blue 6 1000000
Violet 7
Gray 8
White 9

Resistor Color Codes

6
Engineering 162 Laboratory Manual Spring 2004

EXPERIMENT NO. 1 - Oscilloscope Use

I hope that the structure of this experiment gets you familiar with the full range of
measurements possible with a digital oscilloscope and also demonstrates some of the problems
involved in their use. The first problem is that oscilloscopes can affect the circuit you are
measuring so that what you see is different from how the circuit works when you are not look-
ing at it. Such effects can be serious and while they can be minimized with the proper cables
and probes, they cannot be eliminated. The second problem is a set of complications that
arise from sampled data. Our Agilent scopes sample the input signal and convert it to discrete
measurements that are stored in RAM memory, allowing one to read that data out into files on
a PC. (The scope can also do mathematical operations including Fourier transforms on its
stored data, but getting those results out is difficult. You will read some of that information
from the screen but you will then do the operations over in Excel with more insight.). One
must be sure that the sampling rate and voltage resolution are adequate to represent the signal
you are measuring. With our new scopes, this is not usually a problem. The scopes seem so
automatic and your calculations so easy that you may think finding the spectrum of a signal
accurately is a piece of cake. When you do it however you will find the results subject to errors
from inadequate sampling or mismatch of sampling time to the period of the waveform are
significant.

We have built two circuits that generate complex, periodic waveforms roughly analo-
gous to a slow scan television video signal. (There are two independent circuits so that more
than one person can work at a time.) You are to measure several features of the signal includ-
ing its shape, its fundamental period, its vulnerability to corruption by capacitive loading, its
relationship to a second, "synchronizing" signal, and finally its Fourier spectrum. We have
two types of probe for connecting the scopes to the generator. One is a “10X” attenuator probe
that reduces the signal amplitude by a factor of 10 before it reaches the scope. The other
probe, called a 1X probe, passes the signal without attenuation. The difference is not only the
attenuation but also the amount of capacitance that the circuit “sees” when you connect the
probe. You will find that the tradeoff for the reduced gain of the 10X probe is that its lower
input capacitance affects the circuit less. I have tried to design the experiment to lead you to
think about why this is the case.

Measurements:

1 There is synchronization or “synch” pulse output available from a BNC connector on


the waveform generator. Using a coaxial cable, connect that signal to one channel of
the scope and set the scope for a stable measurement. With the scope cursor controls
and digital readout, determine the maximum and minimum voltages (be careful about
offset), the period, and the pulse width at the 1.5-volt level. (The 1.5-volt reference

7
Engineering 162 Laboratory Manual Spring 2004

level is standard for 3.3-volt CMOS logic and the older 5-volt TTL logic. Do not con-
fuse the reference voltage level for time measurements with the pulse amplitude.)

2 Now connect the synch signal to the “Ext, Trigger” input of the scope’s time base and
set the scope to trigger on the falling edge of this signal, i.e., use external synch mode
with negative slope and appropriate level. Connect the actual waveform, available
only as a terminal on the protoboard itself, to a vertical input of the scope (i.e., Ch1 or
Ch2), setting the vertical sensitivity and horizontal time base to display between one
and two periods of the signal with as large a height as can be set without clipping.
Measure the period and the maximum and minimum amplitudes.

3 The signal source has an appreciable output resistance. Its Thevenin equivalent circuit
is: ROUT
VOUT
Vopen_circ

Determine Rout by loading the circuit with a suitable empirically-determined resistor


and inferring the value of Rout from the change in output voltage. WARNING: think
carefully and even experiment to determine which probe to use for this measurement.

4 On the PC connected to your oscilloscope, open a new EXCEL spreadsheet to collect


the data for your write-up. Capture a trace of the waveform measured with the 10X
probe taking 2000 samples of data. Leaving the scope controls unchanged, make this
measurement again connecting the signal with just the 1X probe, and again one final
time with both probes connected to the generator. (Obviously, you still only need one
trace for this last measurement as both probes see the same signal.) Each captured
waveform will appear in your workbook as a separate sheet; you probably want to an-
notate each sheet with what measurement it contains.) Notice that the waveform
changes shape as you change probe! This is the effect on the actual signal of connect-
ing it to the scope and it is non-negligible. Be sure to record the horizontal time base,
the vertical sensitivity, and the screen size in cm for both axes, as you will need that
data for your calculations.

5 A change of shape in a signal implies a change in its spectrum. Measure the frequency
spectrum of the signal from first the 10X probe and then from the 1X probe as follows.
(Do NOT have both probes connected to the source at the same time!) Change the
8
Engineering 162 Laboratory Manual Spring 2004

scope time base so that many cycles of the waveform are displayed (try 100 µs per di-
vision, for example). Set up a fast Fourier transform (FFT) of the waveform using the
scope’s “Math” menu. Try a span of 500 kHz and a center frequency of 250 kHz.

What are the frequencies of the peaks in the spectrum, and how do they relate
to the fundamental frequency measured in part (1)?

Note the sampling rate that is indicated on the scope display. Compute how many
samples are taken during the displayed time interval of the waveform. Watch what
happens to the spectrum display and the sample rate as the time base control is
changed. Also compare the spectra obtained with a “Hanning” window versus a “flat-
top” window. These windows are specific ways of weighting the samples at the ends
of the time interval in order to minimize the effects of having a finite sample interval.
The Hanning window is preferred for accurate frequency measurements, while the flat-
top window is used for amplitude measurements.

Set the flattop window and a sample/sample rate that gives good resolution of the
peaks. Set the channel sensitivity so that the waveform occupies most of the screen
without any clipping. Note these settings for reference later. Measure the amplitudes
of the first 5 harmonics relative to the fundamental in dB for the waveforms from both
probes.

6 Capture in EXCEL the same traces you analyzed with the scope’s FFT module so that
the data is available to analyze yourself.

7 Measure the length of the probe cables with a ruler. (You will use this to find their ca-
pacitance per foot.) Record the values of resistance and capacitance marked on both
the 10X and 1X probes and the input of the oscilloscope. These values are the equiva-
lent impedance seen at the probe or scope input expressed as a parallel combination of
a resistor and capacitor. For the probes, the circuits are for the probe and scope to-
gether. They do not necessarily correspond to any actual physical components.

8 I do not wish to offer an occasion of sin and take a very dim view of copied data.
Therefore, you must have your work signed by the TA in one of two ways, depending
on how fast I get some software to work. Either the TA will initial a graph printed out
from your EXCEL file in the lab or he will paste a software-generated ID number
unique to you into your spreadsheet. If the latter, you must print that on one of the
graphs in your write-up, as discussed below.

Guidelines, Issues, and Questions for Write-Ups:

9
Engineering 162 Laboratory Manual Spring 2004

Please begin your write-up with a short introduction that talks of the features of measurements
and spectra that seemed important to you from this lab. Then use the following list as guid-
ance for the rest of the report. (Doing so will make it easier to write and easier for us to
grade.)
1. Tabulate your synch pulse measurements.

2. Show the measurements and calculation by which you found the output resistance of
the source.

3. Analyze the operation of the 10X probe. The equivalent circuit of the 10X probe is
shown below. The components R1 and C1 are actual physical elements in the probe
body. The capacitor CCABLE represents the cable capacitance. Let CT be the total ca-
pacitance of cable and scope given by CT = CCABLE + CSCOPE . Please do the following
for this circuit:

• Derive the transfer function of this system and show that it is independent of
frequency if CT/C1 = R1/RSCOPE.

• The real probe is designed and adjusted to meet this constraint. (There is a
screwdriver adjustment of CCABLE, an adjustable lumped capacitor in parallel
with it, that we try to keep set properly.) Prove that this condition also makes
the input impedance equivalent to a single resistor and capacitor in parallel.
The probe is marked with the values of these equivalent components rather
than any of the values of the actual components: R1, C1, RSCOPE, CSCOPE or
CCABLE. The scope itself is marked with its input resistance (1 megohm) and
capacitance.

• Find the values of the real components of the probe and scope system (R1, C1,
CSCOPE and CCABLE) from your theory and measurements.

4. Typical coaxial cable usually has about 30 pf of capacitance per foot. How does that
compare to your estimate of the capacitances per foot of the two probe cables? What
is the total capacitance of the 1X probe, including both its cable and the scope connec-
tion?

10
Engineering 162 Laboratory Manual Spring 2004

C1

R1

vsig vout
CCABLE CSCOPE
RSCOPE

Source Probe and Cable Scope

Equivalent Circuit for a 10X Probe Connected to an Oscilloscope

5. Using your EXCEL results, plot the data you took of the single cycle traces. Overlay
the data taken with the 10X probe, the 1X probe, and both probes on the same graph.
Distinguish the traces by color, line type or small vertical offset. Try to design the
presentation so that the change in signal features is clear as the load capacitance in-
creases. Comment qualitatively as to why the curves change. I had you synchronize
the horizontal sweep using the “External” input primarily for this part of the measure-
ment. What advantage for this plot does such synchronization have?

6. Look at your raw data and determine what the minimum voltage step is that the scope
stored, that is, what is the vertical resolution of the scope on this scale. How many
such steps are in your trace? How many fill the screen? How many bits per sample
does the scope store?

7. Show that output impedance of the generator, R0, in conjunction with the scope probe
forms a low pass filter and compute the cutoffs for the three cases you just plotted.
How do those frequencies compare to the fundamental frequency of the waveform?

8. From the measurements you took of the amplitudes of the first five harmonics in sec-
tion 5 of the measurements, what is the lowest harmonic frequency at which the 1X
11
Engineering 162 Laboratory Manual Spring 2004

measurements were attenuated by 3 DB (a factor of 1


2
) or more than the 10X ones?
Is this consistent with the calculations of the cutoff frequencies of your probe setups?

9. Prepare FFT spectra from the data you took over long traces with the 10X and 1X
probes in section 6 of the measurements. You can do this within EXCEL using the
Fourier Analysis tool that is part of the Data Analysis add-in or you can import the data
into MATLAB. Because that transform is done with an FFT – Fast Fourier Transform
– algorithm, the number of data points must be an integer power of 2 – in your case
2048 points. You may pad the waveform with zeros or repeat a portion of the wave-
form. The transform produces 2048 complex numbers, but because the input is real,
only the first 1024 contain unique data. The rest contain duplicate data. You are in-
terested in the magnitude spectrum and can use the IMABS() worksheet function to
show only the magnitude of a complex number. Scale your results so that your plot
will give the RMS value of each frequency component. (Scaling is probably by
2
2048 .) The lowest FFT component is DC and is not reliable because it may be
changed with the scope’s vertical offset control. You may ignore it. The remaining
frequencies are separated by constant steps given by 1T where T is the total time in-
terval of the original measurements. There are a total of N 2 such frequencies where N
is the number of points in the set. For example, a trace taken with 2000 samples at 100
µsec per cm horizontal scale, will have a frequency step size of
1/(10cm*100usec/cm*2048/2000) = 976 Hz. Its data will span frequencies to 1 MHz.
Plot the 10X and 1X data.

10. Because the waveform is known to be perfectly periodic, the proper spectrum is given
by a Fourier series. I have used sine and cosine coefficient calculations to derive the
spectrum quite accurately from the single cycle data taken with the 10X probe. The
graph below shows my results and the data on which it is based is available on the
class web site. Overlay the 10X probe spectrum you calculated on a single graph along
with my results. Comment on the accuracy of your calculation with respect to both
harmonic frequencies and amplitudes.

11. As a final test of your powers of observation, here is a challenge question. The scope
automatically compensates its gain for whether you use a 10X or a 1X probe. (It does
not work for any other probe range – there are, for example, 100X probes for high volt-
age that do not display correctly.) How does the scope know what probe you are
using?

12
Engineering 162 Laboratory Manual Spring 2004

0.6

0.5

0.4
Volts RMS

0.3

0.2

0.1

0
0 100 200 300 400 500 600 700 800 900 1000

Frequency (KHz)

Magnitudes of the Harmonics of the Lab 1 Waveform Calculated from its Fourier Series

13
Engineering 162 Laboratory Manual Spring 2004

EXPERIMENT NO. 2 - Characterization of a Diode and a Transistor

The object of this experiment is to measure of some of the device characteristics dis-
cussed in class for a typical diode and for typical bipolar and MOSFET transistors. The diode
is a 1N4003 diode designed for rectifier service with a nominal breakdown rating of 200 volts
and an average forward current rating of 750 ma. The bipolar transistor is the 2N3904 a low
power transistor with a VCB0 rating of 30 volts, a maximum collector current rating of 150 ma
and a power dissipation rating of 0.3 watt maximum. The MOSFET is one of four found on an
integrated transistor array from Advanced Linear Devices, their ALD1106PB. Full data sheets
on these devices are on the class website and I have put extracts in an Appendix to this man-
ual. The MOSFET is not something we will have talked about before you do the lab. Before
the class is over, I hope to have you compare this data to the simulation models from the
SPICE program. (The MOSFET measurement will also give you the basic idea of the device
used in lab 3.)

Current and voltage measurements are done with one of two instruments, a Hewlett Packard
parameter analyzer for low voltages and currents and a much older Tektronix Model 575 curve
tracer for higher voltages or currents. Data from the analyzer can be captured on the PC that
controls it and you can email those files to yourself. The software for the analyzer is still
evolving and the TAs will have to explain its condition and operation to you as soon as I tell
them. (It is actually in pretty good shape now and I hope to finish it soon.) Most of the meas-
urements will be made with the analyzer. Instructions for the curve tracer are included with
this write-up. As much as possible we will try to let you do the measurement yourself.

1. Measure the capacitance of a reverse biased diode using our Boonton capacitance
meter. This instrument applies a small 1 MHz signal to one side of the unit under test
and senses the current in other lead, while that lead is effectively grounded. Readout is
by a panel meter graduated directly in picofarads. Be sure the meter is zeroed before
putting in your diode and make use of the mirror scale to minimize parallax. Two ter-
minals on the back of the meter, labeled "bias," allow one to apply a DC voltage to the
unit under test while the capacitance is being measured. There is a high voltage power
supply available to give bias voltages over 100 volts, but its minimum output is about 6
volts. Both you and the meter need protection from this voltage and I have an appro-
priate network for that purpose connected to the back of the meter already. Please do
not disturb it.
There is a second supply to use for bias in the range from 0.2 volts forward to 15 volts re-
verse. The meter on the high voltage supply is not very accurate and the protection net-
work I have on the meter reduces the voltage at the diode. Therefore, use a separate digital
voltmeter to measure the voltage being supplied to the Boonton and to determine the po-
larity of the bias on the diode. (Remember you are interested in the reverse bias capaci-
14
Engineering 162 Laboratory Manual Spring 2004

tance.) To avoid problems with including the capacitance of the meter with that of the di-
ode, connect the DVM to the bias terminals on the back of the meter. You can use the me-
ter across the diode terminals to determine polarity before connecting the diode. Measure
and plot the reverse bias capacitance of the 1N4003 (or equivalent) diode from zero to 100
volts reverse with enough points to determine whether it approximately follows the ex-
pected curve for an abrupt junction:
CJ 0
C = C0 +
1 − V / VBI

where Co is a fixed capacitance due to the case of the diode, CJ0 is the junction capaci-
tance at zero bias, and VBI is the junction built-in potential. Because this equation varies
more rapidly near zero volts, you will have to take points more closely spaced near zero
than you do for higher voltages. VBI must be less than the bandgap potential (1.18 volts
for silicon) and is likely to be more than .7 volts. Using Excel or a math package, find the
best estimates for C 0 , C J 0 , and VBI . Then overlay your data with a theoretical plot based
on these estimates.

2. Using the semiconductor parameter analyzer, measure the forward current of the
1N4003 as a function of bias zero volts until the current reaches 100 ma. Does this
curve fit a relation of the form:
I D ∝ I S e qV / nkT

where n is a constant? What value of n fits best? Over what range of V and ID is this fit to
within 20 %? (You can take this data home as a file for later processing.)

3. Again using the semiconductor parameter analyzer, make a Gummel plot for your
2N3904 transistor with VCE = 3 volts, keeping IC under 100 ma. Begin with VBE at zero
volts so that any heating effects will affect only the high current measurements. (A
Gummel plot has overlaid curves of IC and IB as functions of VBE at a fixed, usually low
value of VCE. The curves are drawn on a log-linear scale so that a wide range of cur-
rent can be shown.) From this plot, make a sketch of hFE versus IC. Over what range
and with what value of n is IC proportional to exp(qV/nkT)? Does IB behave the same
way?

4. Still with the parameter analyzer, measure hfe for the 2N3904 at collector currents of
5µamp, 0.5 ma, 10 ma, and 80 ma with VCE = 5 volts. Note that hfe is the differential
gain ∆IC/∆IB that you approximate as ∆I C / ∆I B where the incremental currents are
their changes for one or two steps of the base current. It may not be very different
from hFE.

15
Engineering 162 Laboratory Manual Spring 2004

5. Measure IC as a function of VCE from 0 to 10 volts using IB as a parameter, taking


enough data to be able to draw one of the sets of curves we use in class to show ampli-
fier operation. Cover the range of IC up to 10ma with 5 – 7 base current curves. Plot
this in your report and derive from it the Early voltage at I C ≈ 5 ma and VCESAT at the
same current.

6. In principle and in practice the emitter and collector of a bipolar transistor can be in-
terchanged, and the device will still show appreciable common emitter current gain. In
this configuration, the old collector is, of course, the new emitter. This is called "re-
verse operation" of the transistor, and the current gain is called hRE or βR. Measure
this quantity for IC = 1 ma. How does this compare to hFE as derived from your
Gummel plot? Calculate the corresponding common base gains αF and αR? Why are
the relative magnitudes of hFE and hRE what they are?

7. For the MOSFET using the parameter analyzer, measure ID as a function of VGS for a
fixed drain voltage of 4.0 volts and a gate voltage from 0.3 to 2 volts. Use a logarith-
mic current scale so you see the full drain current behavior. Set the analyzer current
limit for the drain current at 10 ma. Plot this data on the same log scale in your report.
Replot it on linear axes as I D versus VGS.

8. Define the breakdown voltage of a junction to be the voltage at which the reverse leak-
age current rises to 10 µamp. Using the curve tracer, determine the breakdown voltage
of a 1N4003 and of the collector-base (VCB0) and emitter base (VEB0) junctions of a
2N3904. (NOTE: Any diode in the 1N4001 to 1N4007 family is suitable for this
measurement. The curve tracer has a maximum supply voltage of 200V. This may
not be sufficient to cause breakdown, in which case, report an approximate upper
limit to the leakage current at 200V instead.) Why do the two junction breakdown
voltages for the 2N3904 have such different values?

NOTES:

1. Please try to avoid damaging components by limiting reverse, leakage current to 10


µamp. None of these junctions are designed with sufficient thermal dissipation to
serve as a zener diode.

2. Here are the base diagrams (bottom view) for the 2N3904 and the MOSFETs:
16
Engineering 162 Laboratory Manual Spring 2004

Gate

Drain
collector
emitter Source
base
2N3904 - bottom view TO-18 MOSFET - bottom
view

16 9

ALD1106PB

1 8

Quad MOSFET array

17
Engineering 162 Laboratory Manual Spring 2004

Gummel Plot of a 2N3904

1.00E+00

1.00E-01

1.00E-02

1.00E-03

1.00E-04
Current (amp.)

1.00E-05

1.00E-06
Collector Current
1.00E-07
Base Current
1.00E-08

1.00E-09

1.00E-10

1.00E-11
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Base Voltage - VBE

Example of a Gummel Plot

18
Engineering 162 Laboratory Manual Spring 2004

USE OF THE TEKTRONIX 575 TRANSISTOR CURVE TRACER

The following is a brief description of the procedure that should be followed to find the
characteristics of a typical transistor.

STEP 1: Study the accompanying Figures 1 and 2 until you understand how the tracer
works and until you can associate each control on the instrument with the por-
tion of the system it affects. It may take you some time to fully understand it,
and actually trying to use the instrument may help clarify its operation.

STEP 2: Turn the curve tracer on. While the tracer is warming up, set the vertical and
horizontal scale controls for appropriate ranges of collector current and collec-
tor voltage. These controls set the sensitivity of the display and are analogous
to the sensitivity controls of an oscilloscope. Normally one knows in advance
what range of current and voltage is of interest. The voltage or current per di-
vision is set so that full scale of 10 divisions encompasses the range of interest.

STEP 3: Even before putting a transistor into the test socket set the "Peak Volts Range"
and "Dissipation Limiting Resistance" knobs to limit the voltage and power
available to the device. Doing so will reduce the changes of burning anything
out. The peak volts should be kept below the breakdown rating of the device.
There is a graph on the top of the 575, which shows the relation between
power, peak volts and resistance. Use this to select a resistance to limit the
power. Small plastic transistors generally withstand about .36 watt while lar-
ger metal ones withstand about .6 watt. Power transistors without heat sinks
may absorb 1 watt but over 1 watt is usually possible only with heat sinks.

STEP 4: For common emitter measurements, set BOTH polarity knobs on -(minus) for
PNP and + (plus) for NPN devices

STEP 5: Turn the base generator "Step Selector" knob to a current range about 1/50th of
the vertical current sensitivity range. Set the step generator flick switch to "re-
petitive" and the "steps per family" control to mid-range.

STEP 6: Check that the switch next to the test sockets is in the "emitter grounded" posi-
tion.

STEP 7: Place your transistor in one of the test sockets with collector, base and emitter
leads in the proper slots. You may need to consult a data sheet to find the ap-

19
Engineering 162 Laboratory Manual Spring 2004

propriate connections. Turn the transistor on by flicking the selector switch


toward the test socket.
STEP 8: Use the position knobs to shift your beam. A PNP transistor should have the
beam in the upper right-hand corner, while an NPN transistor should have it in
the lower left.

STEP 9: Adjust the "Peak Volts" control to obtain a trace of the collector characteris-
tics. Then set focus, intensity, astigmatism and scale illumination to suit your
personal taste.

STEP 10: If you do not obtain a trace recheck these steps, particularly Steps 4 and 5. It is
always possible that your transistor is burned out, and you may want to try an-
other. Even when you do get a trace you may want to make slight adjustments
to the "Step Selector" and "Dissipation Limiting Resistor" to have the trace fill
the screen conveniently.

MISCELLANEOUS REMARKS

The curve tracer can be used to make common base measurements by interchanging
the base and emitter leads and using the base generator to drive the emitter. The interchange
of leads can be made with the "base grounded" knob for transistors in one of the sockets. For
transistors attached to the binding posts, it is necessary to interchange the actual wires. When
using the base step generator to drive the emitter, one has to reverse the polarity of the genera-
tor (+ for PNP, - for NPN) and to increase the current level to the same range as the vertical
sensitivity.

The curve tracer can also be used to measure diodes, JFETs and MOSFETs Diodes
connect between the emitter and collector terminals, and the base generator is ignored. MOS-
FETs connect drain to collector terminal, gate to base and source to emitter. Then the step
generator is set to voltage steps rather than current steps. It is important to use a base current
drive for bipolar transistors since collector current varies so rapidly with base voltage, and to
use a voltage drive on the gates of FETs since they are voltage controlled devices whose gates
are not supposed to draw current and can be damaged if such gate current flows.

20
Engineering 162 Laboratory Manual Spring 2004

dissipation-
mA C
limiting resistor
per
step
B
measure
+
collector collector

+ volts
volts sweep (volt-
age source)
horizontal
per
base step generator

measure
(current source)

step
base base
series
resistor
measure
E –
current source base measure
or volts volts current-
sampling collector mA
horizontal or vertical horizontal resistor
or vertical vertical


Figure 1. Functional block diagram for the Type 575 curve tracer, shown with an npn transistor under
test. This instrument places the transistor in a common-emitter circuit with a continuous sweep for the
collector voltage and a base current that remains constant during one (or one-half) of a collector sweep
but then changes by a fixed amount. Every call-out in a shaded box refers to a corresponding label on
the front of the instrument. The up position of the switch gives constant base current steps while the
down position gives constant voltage steps through a series resistor.
Collector
sweep

Collector sweep waveform—120 sweeps per second


Base step generator

STEPS/SEC switch at the upper 120 position

STEPS/SEC switch at the lower 120 position

STEPS/SEC switch at the lower 240 position


Figure 2. Waveforms of the collector sweep generator (top) and the base step generator (bottom three).
The number of base steps is selectable.

21
Engineering 162 Laboratory Manual Spring 2004

EXPERIMENT NO. 3 - Half a Cheap Stereo

This experiment introduces the idea of biasing for linear amplification. It is intended
primarily to introduce the capabilities of FETs (Field Effect Transistors), to remind you of the
rules for Q point selection in circuits that use matching transformers, and to give you a chance
to think about simple RC circuits. Build the circuit shown below, designing the required RC
network and measuring its performance. It has a transformer to drive a low-resistance speaker
with an amplifier made from a single MOSFET. Using the primary taps marked 2.5 and 5
watts and the C and 8 Ω secondary connections to the speaker will result in a transformer pri-
mary with a primary magnetizing inductance of 0.5 H, a turns ratio 4.6:1 from primary to sec-
ondary, and a primary wire resistance of 30 ohms. (By comparison, the secondary resistance is
negligible.) The speaker acts roughly as an 8 Ω resistance.

There are also a number of MTP2N35 (Motorola) power MOS field effect transistors
or similar devices in the lab that are already mounted on heat sinks with extension wires on
their leads. These wires let you plug the transistor into the rest of the circuit on your proto-
board. Unmounted transistors are intended only as backups. The pin connections and basic
ratings of the MOSFETs are shown below. The gate bias voltage in operation is usually about
3.5 V but may vary up or down by several tenths of a volt. You may have to adjust your cir-
cuit to get the correct (optimum!) quiescent current.

There is a tape player in the lab that we have modified for use as a signal source by adding an
output cable and limiting impedance. It is equivalent to a 2.5 volts peak-to-peak signal with a
source impedance equivalent to 1600 pf capacitor. The figure below shows the connection of
the MOSFET to the coupling transformer and speaker. Designing an RC network to connect
the tape player to the MOSFET and to provide MOSFET bias is part of your job. It should
take either three resistors and one capacitor or two resistors and two capacitors. (I rather like
the latter circuit for aesthetic reasons. At least, you should know what the two choices are.)
The MOSFET gate can be considered an open circuit for this design. The signal voltage from
the tape player at full volume is somewhat bigger than the MOSFET can handle without dis-
tortion. To fix this problem, your RC network should attenuate the signal by a ratio of 5:1. In
your write-up, you are expected to show this is attenuation is realized by design. The low fre-
quency cutoff associated with the capacitor of the signal source (1600 pf equivalent source
impedance) has to be placed at 50 Hz.

Select the bias from the network to make the quiescent current optimal for maximum possible
output. The choice of an appropriate value for the ratio of the minimum dynamic drain current
to its quiescent value is up to you. Determining the bias voltage that corresponds to the opti-
mum drain current may be done empirically. You may use the curve tracer or parameter ana-
lyzer to determine an appropriate value of gate voltage VGS or you may use trial and error to
22
Engineering 162 Laboratory Manual Spring 2004

find the right IDS. This is the one lab that allows a cut-and-try approach of adjusting one resis-
tor value while measuring the quiescent drain current IDS. Note that the selection, adjustment,
and confirmation of IDS are an important part of the basis on which your report will be graded.

In your report, you should show data for the quiescent gate and drain voltages, and the
quiescent drain current. You may measure the latter by inserting a 10-ohm resistor in series
with the power supply VDD line and measuring the voltage drop across it with a digital volt-
meter. If you wish, you may use this same method in setting the drain current value. In addi-
tion to a qualitative evaluation of the functioning of the circuit, you should indicate how you
designed your RC network. Also calculate the expected low frequency cutoff due to the trans-
former magnetizing inductance.

VDD +12 V
4.6 : 1 (Turns ratio)

Cs
1600 pF
RC Speaker
MTP2N35 8Ω
network
Vs
2.5 V p-p

Signal Source MOSFET Amplifier


(Equivalent circuit to tape player)

The circuit has been designed without a volume control because, as nearly as I can tell,
such controls are of no interest to anyone under thirty. If, however, you are amused by the
idea, you might, optionally, add one. The challenge will be to prevent the volume control
from changing the bias point. I will try to remember to have a 1-megohm potentiometer in the
lab for the purpose. If you are not satisfied with the program material we have supplied, feel
free to try some of your own.

23
Engineering 162 Laboratory Manual Spring 2004

NOTE: MOSFETs are discussed in chapter 4 of Sedra and Smith. If you have done lab 2, you
will already have some idea of how the drain current of a MOSFET is controlled by the volt-
age from gate to source. We will discuss MOSFETs in class near the end of March. You do
not need to know much about them to do the lab since the lab is largely independent of the de-
tails of the device! Field effect transistors differ from BJTs in a couple of ways that are impor-
tant here and which make the lab easier. The gate of an FET is its control terminal in the same
sense as the base of a BJT is for it. However, the gate draws no steady DC or bias current at
all. Superficially the drain current vs. drain voltage curves with VGS as the curve parameter
look just like those of a BJT. The main difference besides the lack of gate current is that the
gate voltage for a given drain current is larger than VBE for a BJT and tends to be less repeat-
able from unit to unit. Interestingly nothing in the consideration of the "best" Q-point for a
single stage amplifier depends on the differences in the device types.

MTP2N35: n-channel enhancement-mode MOSFET made by the DMOS (diffused MOS)


process

Maximum VDS 350 V


Maximum IDS 2A
Maximum P 40 W (on heatsink with ambient temperature 25 °C)
Threshold, VT 2.5 V (typical)
Saturation, VDSSAT 1 V (approximate - VGS dependent)

MTP2N50 MOSFET
Front View

G D S

24
Engineering 162 Laboratory Manual Spring 2004

EXPERIMENT NO. 4 - Audio Frequency Amplifier

This lab introduces you to some very common aspects of design through a very simple
single transistor amplifier. In this lab you will get your first design experience with the con-
cepts of small signal analysis and biasing. These concepts are extremely fundamental to cir-
cuit design, so make sure you understand them, and the distinction between them. This lab has
been introduced into the course primarily because students have had trouble with these con-
cepts. In addition, this lab requires you to use some common measurement techniques. The
experiment is quite simple relative to some of the later ones, and should not take you much
time in the lab. As your first design, it may require substantial thought beforehand.

The Experiment

You must design, build, and characterize an amplifier having the following specifica-
tions

Gain: A v = 15 ± 20%, (23 DB +/- 1.5 DB) from 100Hz to 10kHz


Input Impedance: 10 kΩ ± 2 kΩ @ 1kHz
Output Impedance: <6kΩ @ 1 kHz
Output Voltage Swing: at least ±2.5 volts from quiescent point
Low Frequency Rolloff: -3dB point (relative to midband gain) must
be kept below 30Hz

Your amplifier should run from a single 12V power supply, and be based on the 2N2222A
(npn) transistor, which has the following specifications:

BVCEO = 30V
70 < hFE ≈ hfe < 230 @ l mA
PD = 300mW
1/hOE > 150kΩ typical @ 1 mA

Through the proper choice of biasing components, your circuit should meet design cri-
teria for the full range of hFE (70 to 230) and for 0.50V < VBE < 0.70V. It is not sufficient for
you to make your circuit work for one particular set of components - it must be designed to
work for any components that meet the appropriate specifications (i.e. 65 < hFE < 250, resistor
values ± 5%, etc.). You will be graded stringently on this point.

25
Engineering 162 Laboratory Manual Spring 2004

A common emitter amplifier with a partially bypassed emitter circuit will work. You
must choose the passive component values so that your circuit functions as required. Your
report should contain:

• Schematic

• Description of how you chose your component values, along with calculations of qui-
escent voltages and currents, gain, input impedance and output impedance

• Measurements of quiescent voltages

• Printout of the Bode plot of the amplifier taken on our computerized plotter showing
that the amplifier conforms to the design specifications. The printout must show your
protoboard number and be initialed by the TA.

• Measurements of input and output impedances at 1kHz.

HINTS

1. It is often easier, in this type of design, to satisfy the small-signal design


specifications first, and then choose biasing components, which, consistent with
small-signal requirements, fulfill the remaining requirements of quiescent point
stabilization and large-signal properties such as output voltage swing.

2. Input impedance is most easily measured by placing a resistor comparable to the


expected input impedance in series with the signal source. You measure the in-
put current from the voltage across this resistor. Use two oscilloscope probes,
one connected to each side of the resistor, and set the scope to measure the dif-
ference in the voltages (invert one scope channel and sum them). Be sure both
channels are set to the same sensitivity. (This approach to measuring the input
current is suggested because it is simple to predict when it will work. Instru-
ments such as bench voltmeters and DVMs usually have too much capacitance
and too little frequency response to be used except at quite low frequencies. The
differential scope method works well to quite high frequencies.)

3. The output impedance can be measured using a similar technique to what you
used in lab #1. The only trick is to connect the load resistor so that the quiescent
point is not disturbed so much that the amplifier ceases to work properly.

26
Engineering 162 Laboratory Manual Spring 2004

VCC

RC
R2

CIN
2N222A

R4
vOUT
vin
R1

R3 CE

Circuit Topology for a Common Emitter Amplifier Based on the 2N222A

27
Engineering 162 Laboratory Manual Spring 2004

EXPERIMENT NO. 5 - Video Amplifier

The bandwidth for television video signals in the United States (NTSC standard) is
approximately 30 Hz to 4 MHz. Congress has recently passed legislation defining a new digi-
tal TV standard that will require a 3 DB bandwidth of 30 Hz to 6 MHz. This is comparable to
other video applications, particularly computer displays. (Video displays are often even wider
because of their high resolution and high frame rate.) When the display is a large-screen, cath-
ode ray tube, a common design problem arises in which a modest video signal level on the or-
der of 2.2 volts pp is to drive the grid of a CRT electron gun directly. An amplifier capable of
providing the required gain, bandwidth, and output voltage is a standard feature of all TVs and
monitors. Suppose that our new 6 MHz TV requires a CRT drive level is from +20 to +65
volts relative to ground and that in addition to the +12V supply (from which to derive base bi-
ases), 80 VDC can be obtained from the high voltage supplies in the lab. Design and build an
amplifier that will give a 30 Hz to 6 MHz bandwidth as defined by its lower and upper 3 DB
points. The circuit is to have a minimum input impedance of 1,000 ohms over the entire fre-
quency range. Simulate the capacitive loading of the CRT grid with the 10x oscilloscope
probe (about 15 pf capacitance).

You will probably find that these design requirements cannot be met with a simple
common emitter amplifier, and that a cascode configuration will be necessary. The cascode
connection reduces the Miller effect, which helps keep the input impedance high. Also, it is
difficult to combine high cut-off frequency, high breakdown voltage, and high power handling
in one transistor. The cascode circuit eases these demands by dividing the requirements be-
tween two transistors. Design the circuit around 2N3440 and 2N2222A transistors. The
maximum power dissipation for a 2N3440 is 1 watt in an ambient temperature up to 50°C.
Design your circuit conservatively to have the 2N3440 dissipate no more than about .65 W.
Design around typical values of all transistor parameters. A table of some of the parameters is
given below; more complete data is available in the Appendix.

A complication in the problem is that the capacitances shunting the output are substantial. In
addition to the scope probe and COB of the 2N3440 there can be up to 2 pf between terminals
of the breadboard. You will probably find that this capacitance limits the upper frequency
cutoff from the collector load pole to something under 3 MHz. There is a standard method for
increasing the bandwidth by putting a coil in series with the collector resistor as shown below.

28
Engineering 162 Laboratory Manual Spring 2004
VCC

For this circuit, the data shown below


gives the relation between the inductance LC
LC and the bandwidth. The frequency
fmax is the cutoff frequency with no in-
1
ductor, i.e., f max = where RC
(2π RC Cshunt ) COB
Cshunt is the total capacitance to ground
from the output node. Use this configura- Scope probe and
tion to move the 3DB cutoff frequency breadboard ca-
(i.e., the frequency at which the gain de- pacitance
creases by a factor of .707) out as needed.
There is a selection of small RF chokes
(coils) available from the TA. You may
check out and return what you need. (The coils look much like resistors ($ .02 ea.) but cost
over two dollars each. You must return these and not treat them as casually as resistors.) The
color code on coils is the same as for resistors; the units are microhenries. (The coil color
code can be difficult to be sure of because the colors are not well printed. There is an R-L-C
meter in the lab to check with.)

Note that 6 MHz is an appreciable fraction of the 250 MHz common base cutoff fre-
quency of the 2N2222A. In calculating input impedance you will have to take account of that
fact. The Q-point will have to be stabilized in such a way as to account for the range of hFE of
the common emitter transistor and for .5 < VBE < .7 volt. ICBO should not be a problem. Be
sure to show you have met this requirement by design.

Wiring can cause problems in a circuit like this. You will reduce or avoid such prob-
lems if you keep it neat with short leads. One unhappy fact of life at high frequencies is that
power supplies may not be high frequency signal grounds unless one puts a small capacitance
(typically .1 to .01 µfd.) between the supply and ground physically located right at your circuit.
The best kind of capacitor is a ceramic disc type; it is commonly called a bypass capacitor.
This observation applies to both the 12-volt supply on the breadboard station and to the 80-volt
supply, which will have to come from one of the large Lambda supplies. Any electrolytic ca-
pacitors should be bypassed as well, because they often have significant series inductance.

Measure the gain over the range from 10 Hz to 10 MHz. This requirement must be
met with a plot from the automated Bode plotter. Make a separate measurement to show that
you have met the input impedance requirement. Also, record the quiescent circuit voltages.
Please try to minimize component damage by checking your wiring before applying power. If
you have a problem, check the devices on the curve tracer.

29
Engineering 162 Laboratory Manual Spring 2004

Write-up: In your write-up do the following:

I. Indicate how you arrived at component values.

II. Calculate the input impedance of the whole circuit for typical transistor parameters at
6 MHz. Compare this to the measured value. (Note that if you meet the input imped-
ance requirement at 6 MHz, you do so automatically for all frequencies below that
too.)

III. Calculate the limits on the quiescent current due to transistor parameter variations.

IV. Derive the relative gain formula for the circuit including the effects of the inductor.
(See discussion below.)

V. Report your measurement results for both the Q-point and the frequency response
(Bode plot) with an explanation of any discrepancy with calculation.

Gain versus Scaled Frequency for Several Values of kL

1.6

1.4 kL = 0.9
kL = 1.0

1.2 kL = 0.7
Relative Gain

1 kL = 0.5

0.8
kL = 0.3
kL = 0.0
0.6

0.4

0.2

0
0.1 1.0 10.0
Scaled Frequency (f/fMAX)

30
Engineering 162 Laboratory Manual Spring 2004

Relative Gain Calculations: The collector impedance, ZC, made up of RC , LC , and CSHUNT ,
dominates the high frequency gain of this circuit and its dependence on frequency determines
the relative gain of the amplifier. This combination of passive components occurs at various
scales in many applications and is best analyzed in dimensionless terms. The relative ampli-
fier gain will be the dimensionless ratio Z C / RC . If the inductor is not present the circuit has
single pole behavior with a cutoff frequency given by f MAX defined above. Scaling all fre-
quencies by this value makes the frequency scale dimensionless too. Finally, one can express
the inductance as the product of a dimensionless coefficient k L as:

LC = k L RC2 Cshunt
As part of your write-up, show that the relative gain is given by the expression:

1 + [k L2 ( f / f max ) + (1 − k L )( f / f max )]2


3
ZC
=
( f / f max ) + [kL ( f / f max ) − 1]2
2 2
RC
This formula was used to calculate the family of curves on the graph above. The graph is quite
handy for picking LC based on how much additional frequency response is required.

Selected transistor data. See the data sheets for more details.

Transistor Parameters
type
COB (pF) fT (MHz) BVCBO (V) BVEBO (V) PD (W)

2N2222 7 @ VCB=10 250 60 30 0.3


V

2N3440 4 @ VCB=40 25 300 250 1.0


V

HINTS:

1. The collector resistor is fixed by the maximum power dissipation requirements. Use the
maximum power theorem, which states that a voltage source in series with a resistor R and
31
Engineering 162 Laboratory Manual Spring 2004

a load will transfer maximum power to that load when half the voltage is across R and half
across the load. That power is V2/4R. The trick is to identify the voltage that goes with
applying this theorem to your circuit.

2. Pay attention to heat in the collector resistor – it may be too much for the standard quarter
watt resistors you use for most of the labs. See the comments on resistors in the introduc-
tion.

3. The quiescent current is set by the midpoint of the output drive levels.

4. VCB for the 2N2222 is set by a compromise between too low a value increasing the value
of COB and too high a value giving an unstable Q point. A reasonable compromise here is
≈ 4 to 8 V.

5. Input current can be determined by placing a resistor in series with the input and using two
scope probes with the scope set to measure the difference in voltage across the resistor. In
making this measurement, remember that the voltage across the amplifier input is out of
phase with either the generator output or the voltage across the series resistor. You must
measure the amplifier voltage separately of the drop across the resistor. The capacitance
of the scope probes may not be negligible in the measurement of input impedance at 8
MHz. (The reactance of 15 pf at 6 MHz is 1700 ohms!) Be sure to account for this.

6. The output of the oscillators in the lab is not completely independent of frequency espe-
cially above 1 MHz. The Bode plotter compensates for this, but input impedance and your
own checkout measurements may be affected by it.

32
Engineering 162 Laboratory Manual Spring 2004

EXPERIMENT NO. 6 - Sine Wave Oscillator

Sinusoids are useful waveforms, because they are convenient basis functions for ana-
lyzing the response of (linear) systems involving time derivatives and/or integrals of signals.
Any sufficiently well-behaved signal can be represented by a linear combination of sinusoids
(through Fourier transforms) and derivative and integral operations on any of the basis func-
tions yields another basis function. In fact, sinusoids are the only functions that retain their
original shape during processing by a linear system, which makes it easy to characterize the
effect of the system by specifying only the amplitude and phase change. In mathematical lan-
guage, sinusoids, or rather imaginary exponentials, are the eigenfunctions of linear systems.
Consequently, sine wave generators are useful tools for circuit testing and diagnosis (as you
have already seen in the previous labs). In this lab, you will design a variable frequency sine
wave generator. Although the approach you use cannot yield a sine wave of sufficient purity
(i.e. low harmonic content) for use in a high quality generator, the quality of its waveform is
sufficient for many less demanding applications. This is a fairly inexpensive way of producing
a versatile waveform generator (sine, triangle and square waves simultaneously) over a broad
frequency range.

The approach you will use involves the generation of a triangle wave followed by a
conversion of the triangle wave to a sine wave using a nonlinear conversion circuit. It is rela-
tively easy to generate a high-quality variable frequency triangle wave using an op amp inte-
grator circuit with a variable input resistor. (Alternatively, it is easy to convert a variable fre-
quency square wave to a triangle wave using an integrator). The triangle wave generator will
be made using a combination of an integrator and a Schmitt trigger as shown in the figure be-
low. Although in practice the integrator and Schmitt trigger functions can be integrated into a
single fairly simple circuit, the approach outlined in this lab will give you exposure to two
opamp applications (integrators and Schmitt triggers separately).

A circuit having the block diagram shown on the next page can generate a triangle
wave. When the Schmitt trigger output is negative, the integrator output ramps positive.
When that output reaches the positive-going threshold of the Schmitt trigger, the Schmitt trig-
ger output goes positive. This causes the integrator output to ramp downward until the
Schmitt trigger’s negative going threshold is reached. This sends the Schmitt trigger output
negative, and the cycle repeats. The symmetry of the triangle wave depends on two things.
First the negative and positive outputs of the Schmitt trigger must be equal in magnitude to
obtain equal slopes for the positive and negative going parts of the triangle wave. Second, the
positive and negative switching points of the Schmitt trigger, which must be equal magnitude,
control the positive and negative peaks of the triangle wave. Although an opamp-based
Schmitt trigger will not precisely meet these two conditions, the accuracy will be acceptable
33
Engineering 162 Laboratory Manual Spring 2004

for the purpose of this experiment. Your task is to determine how to implement the integrator
and Schmitt trigger in such a way as to yield a triangle wave generator having a frequency
range of at least 100Hz to 10kHz. You will be allowed one 10 kΩ potentiometer for adjusting
the oscillator frequency. The amplitude of the square wave need only be greater than 5 volts
pp, however, the triangle amplitude should be as near a practicable to 5 volts pp.

The triangle to sine conversion circuit you will use is an example of a class of nonlin-
ear circuits. It utilizes the nonlinear I-V characteristics of the base-emitter junctions of bipolar
transistors. Attached is a copy of the original journal article (W.M.C. Sansen, S. Lui, S. Peet-
ers, and Robert G. Meyer, The Differential Pair as a Triangle-Sine Wave Converter, IEEE J.
Solid-State Circuits, Vol. SC-11, June 1976, pp. 418-420) describing this triangle to sine con-
version technique. Using a single CA3086 NPN transistor array chip, design and construct a
circuit following the technique outlined by Meyer, et al. It is almost certainly easiest to use the
collector output option rather than the emitter option that they also mention. The combination
of this circuit with your triangle wave generator will be a variable frequency sine wave genera-
tor. You probably have to put something between the triangle output and the sine-converter
input to reduce the signal amplitude to an appropriate value. The output sinusoid shall be 5
volts pp too and have a zero mean value. The output source impedance shall not exceed 100
ohms. (This does not mean you could load the circuit with 100 ohms! You may meet this re-
quirement by design.) You very likely need another opamp to do this. It is probably best to
use it as a differential amplifier, taking the difference in potential between the two collectors
of the sine converter. This reduces distortion and saves a capacitor. As a difference amplifier
is a little harder to design, I will reserve points on the lab grade specifically for doing this suc-
cessfully.

In demonstrating your circuit to the TA, you must show reasonably distortion-free sine
wave generation from 100 Hz to 10 kHz. Distortion must be reduced by proper selection of
passive components in the triangle wave oscillator. To establish a quantitative measure of the
distortion of the sine wave, use an oscilloscope to capture the signal when your generator fre-
quency is about 1 kHz. Use a sampling rate of at least 100 kHz and try to adjust for as nearly
ten cycles of the waveform on screen as practicable. Fill the vertical scale as much as possible
without clipping. Use Excel to calculate the FFT and from the peaks in that transform calcu-
late the ratios of the second through fourth harmonics to the fundamental. To meet the re-
quirements of the lab, the amplitudes of the second and third harmonics must be at least 23 DB
below the fundamental. Also capture the square and triangle outputs of your circuit at 1 KHz.
In your report, overlay them with the sine wave data all on the same time axis. As in lab 1,
the TA will either initial a printout of your waveforms or give you an electronic signature to
include on your graph when you demonstrate your triangle and square wave outputs to him.

In the section of your report that has the tabulation of measured and designed values,
please include at least:
34
Engineering 162 Laboratory Manual Spring 2004

1. Amplitudes of all waveshapes at 100 Hz, 1 kHz, and 10 KHz.


2. Distortion of the sine wave -- FFT printout and the derived ratio of the second and
third harmonics to the fundamental amplitude. Make your FFT printout have a DB
scale and normalize so that the fundamental is at 0 DB.
3. Waveshapes of all outputs at 1 KHz overlaid on the same time axis.
4. Maximum and minimum frequencies.
5. Quiescent values of the voltages in the sine converter
6. DC or mean value of the sinusoidal output -- i.e. is it really zero mean?

Hints:

1. Bypassing the power supplies with small ceramic capacitors on the breadboard will help
avoid spurious oscillations.

2. Keep the bias currents in the CA3086 transistors below 1mA for best results.

WARNING: Pin 13 (SUBSTRATE) must be connected directly to the


VEE (-12 volt) supply line in order to isolate one transistor from another. This limits
how one can use transistor Q5. Failure to observe this requirement may destroy the
IC. We do not take kindly to wholesale destruction of CA3086s. You check these
out from the TA, who is able to check their operation before you receive them and
after you return them.

3. One technique for realizing a variable time-constant integrator is shown below:

CI

RI
R2

vIN
+
R1 vOUT

35
Engineering 162 Laboratory Manual Spring 2004

Variable Time Constant Integrator

Positive feedback

Sine wave
Schmitt Variable time output
trigger constant inte- Waveshaping
grator circuit

Function Generator Block Diagram: the Schmitt trigger and integrator form a square
and triangle generator. The waveshaping circuit uses a differential pair with current source
biasing and an emitter resistor. (See attached article.) You should have an opamp buffer
in that circuit to minimize distortion and lower the output impedance.

36
Engineering 162 Laboratory Manual Spring 2004

37
Engineering 162 Laboratory Manual Spring 2004

38
Engineering 162 Laboratory Manual Spring 2004

39
Engineering 162 Laboratory Manual Spring 2004

EXPERIMENT NO. 7 - Telephone Anti-Aliasing Filter

WARNING: I have started on redesigning this lab using MOSFET array transistors. If I
complete the redesign before anyone does this lab, it will be superceded by the new exer-
cise.
It is now standard practice to send telephone voice service over digital links by filtering the
analog signal and sampling this bandwidth-limited signal with an ADC. The standard service
uses an 8 Ksps sampling rate, which requires that the signal bandwidth be restricted to below 4
kHz. (Although you are not expected to know this as a result of prerequisite courses, all sys-
tems which use analog to digital conversion must be careful to limit the high frequency content
of the original signal. It is a sampling theorem due to Nyquist that the original signal can only
be recovered from the digital samples if the signal had no components beyond one-half the
sampling frequency. If this condition is not met, then not only is data lost, but false informa-
tion is folded into the real data. If you played around with the FFT scope in Experiment 1, you
may have seen this effect.) Voice service requires that frequency components in the original
signal be maintained as well as possible out to beyond 3 kHz. Usually the 3 DB down point is
placed at around 3.1 to 3.3 KHz. In this lab you are to design and build a filter which might be
used for this purpose using active circuits based on operational amplifiers.

The main difficulty in realizing such a filter is the need for out of band suppression. The mini-
mum suppression I will consider acceptable is 9 DB at 4 kHz. and 30 DB at 8 kHz. The mid-
band gain of the filter subsystem at 1 kHz. should be 20 DB (or 10X) +/- 1.5 DB. At the low
frequency end, there should be at least a single pole high pass response so the DC components
are removed from the signal. The low frequency response should be down no more than by 6
DB at 100 Hz. You are free to choose any standard transfer function that will be reasonably
smooth. If you choose one which exhibits passband ripple (the Chebycheff response, for
example) then that ripple should not exceed 1 DB. Probably the simplest acceptable solution is
a Butterworth filter of odd order. (Odd order because most of the complication of construction
goes into the quadratic pole pairs, and extra rolloff comes with minimal effort with one more
simple pole.) You are free to make that filter from cascaded sections of any of the designs that
are discussed in your text or in class. There is an automated response measurement available
as with labs 4 and 5.

To meet the requirements for response with the tolerances of the components in the lab may
require adjusting some resistor values. (You should use mostly film capacitors, as their toler-
ance of +/- 10 % is the best we can manage. Resistor values come in roughly 10 % incre-
ments.)

While I am setting the rolloff requirement fairly leniently, you should be aware that real sys-
tems require more suppression near 4 kHz. This extra attenuation can come at some increase
in the transfer function at higher frequencies. The way this is usually done is to use a filter that
40
Engineering 162 Laboratory Manual Spring 2004

has one or more zeros on the imaginary axis. A zero of this kind means the filter has no re-
sponse to sinusoids at the zero frequency. The figure below shows two transfer functions over-
laid. The first curve drops very rapidly past 3.1 kHz, going to zero transmission at 4.05 kHz
before rising to -37 DB at a little over 5 kHz. This is the transfer function:

1 (sτz )2 + 1 1
H(s) = • •
1 + sτ1 sτ sτ 3
(sτ 2 )2 + 2 + 1 (sτ3 )2 + + 1
Q2 Q3

where

τ1 = 1.51 • 10-4 sec [1053 Hz]

τ2 = 7.05 • 10-5 sec [2256 Hz]

Q2 = 1.57

τz = 3.88 • 10-5 sec [4100 Hz]

τ3 = 5.14 • 10-5 sec [3096 Hz]

Q3 = 8.07

The second curve is a 5th order Butterworth filter with its cutoff at 3.1 kHz. Notice that even-
tually the Butterworth will attenuate more than the first curve, but the filter with a zero does
better up to 10 kHz. I am including a reprint from a book, Principles of Active Network Syn-
thesis and Design by Gobind Daryananni (John Wiley, NY, 1976) on how to build the section
of the filter with the zeros using three or four opamps. If you are feeling adventurous, I would
encourage you to try this circuit. It actually is only two or three more opamps than the least
complicated acceptable solution. Please feel free to take or ask for more opamps as you need
them. You can make the two factors of H(s) that have no zeros in any of the usual low pass
topologies – they are not particularly high Q. The section with zeros can be done by following
the discussion below.

41
Engineering 162 Laboratory Manual Spring 2004

Telephone anti-aliasing filter response


20

-20
Attenuation (dB)

-40

-60

-80

-100

-120
2 3 4 5
10 10 10 10
Frequency (Hz)

Realization procedure for zeros on the jω axis from Daryannani, op. cit.:

42
Engineering 162 Laboratory Manual Spring 2004

43
Engineering 162 Laboratory Manual Spring 2004

44
Engineering 162 Laboratory Manual Spring 2004

45
Engineering 162 Laboratory Manual Spring 2004

EXPERIMENT NO. 8 - Design of a Bipolar OTA Circuit

WARNING: I have started on redesigning this lab using MOSFET array transistors. If I
complete the redesign before anyone does this lab, it will be superceded by the new exer-
cise.

When an integrated circuit was being designed, it was once customary to breadboard it
using standard chips with transistors similar to those expected to be made by the ultimate pro-
duction technology. Now it is usually done by computer simulation alone, but you can still try
one the old way. The point of breadboarding is to be sure that there are no undesirable operat-
ing modes for the device. It also gives the designer a chance to try the device in practical ap-
plications and to revise the specifications if necessary to make them more useful without hav-
ing to commit to the large costs involved in a set of masks. Although some of the techniques
for such breadboarding are not readily available to us, we do have two types transistor arrays
(3 npn's and 2 pnp's on one chip) available.

Design and breadboard an IC circuit which is a voltage input to current output ampli-
fier. (This type of amplifier is called an operational transconductance amplifier or OTA.) If
v1 and v2 are two input voltages relative to ground and i0 is an output current running through
an arbitrary impedance to ground, then your circuit should realize the function i0 = A(v1 - v2)
where A is a constant with units of conductance. If the amplifier output voltage is v0 and the
power supply is ± 12 V then this formula shall hold for v0 ≤ 10 V and i0 ≤ 1.5 ma. The
common mode rejection ratio is to be ≥ 200 at 100 Hz. The input offset voltage is to be ≤ 10
mvolts; the common mode voltage ≥ ±7 volts; the maximum differential input voltage 5 volts;
the minimum value of A = 0.4 amp/V and the minimum differential input impedance 30
kohms.

WARNING: DO NOT USE A


MILLIAMMETER OR A VOLTMETER
ON A MILLIAMPERE RANGE
TO MEASURE CURRENT IN THIS CIRCUIT

Milliammeters have very low input resistance. Connected between the wrong points in
this circuit they can blow out an IC instantaneously. These integrated circuits are expensive so
use them with care. Check wiring before applying power and ask a TA to check your circuit
46
Engineering 162 Laboratory Manual Spring 2004

configuration for safety. (He will not tell you if it is right, only if he thinks it would destroy an
IC.) For current measurements use a voltmeter across a 1.5 K ohm resistor.

In class, in the discussion of operational amplifiers, we cover circuit configurations for


all but the current amplifier. The creative part of the experiment is devising a current output
stage that can both source and sink current. (Hint: The output can be the difference between a
constant current and a unidirectional, variable current amplifier.)

A block diagram of the circuit is given below. A simple version of the circuit can be
built with 8 transistors from 2 chips. The super fanciest version might take as many as 20 tran-
sistors. Build the circuit according to the rules of the next paragraph and test all the above
parameters except the maximum input voltage. We expect to have an automated arrangement
for measuring the differential gain and input offset. A circuit is given below for measuring the
CMRR. It may give you ideas for testing other parameters.

Since this is an IC design, there are certain restrictions on the components that lead to
the following rules:

1. No capacitors bigger than 40 pf may be used.

2. Resistors fall into two classes according to the voltage across them. Class 1 re-
sistors may have any voltage across them, but no individual resistor may ex-
ceed 39 K and the sum of all Class 1 resistors in the circuit may not exceed 70
K. Class 2 resistors may only have ≤ 3 volts across them, but may have any
value from 2 K to 100 K with essentially no restriction on the total number.

Also, in using the CA3096 transistor arrays, you must connect the substrate (pin 16)
to the most negative voltage in the circuit in order to keep the transistors isolated. Those of
you who find the problem interesting might optionally consider how to add the following fea-
tures to the circuit:

1. Two nearly identical outputs instead of one, so that one might be a true current
output while the other is used for feedback. Alternatively, some OTAs have
two differential outputs, that is, one output is the negative of the other.

2. Ten times as much input impedance with three times the overall gain.

3. Provision for external adjustment of the input offset voltage.

4. Higher output current and lower power dissipation with two more current mir-
rors instead of the current amplifier.
47
Engineering 162 Laboratory Manual Spring 2004

VCC

v1 Differential Current Current iOUT


 Amplifier
Input voltages  (Diff. Pair) Mirror
Amplifier
v
 2

Bias Network

VEE
Block Diagram

10 K
Variable
+
Power
Supply – UUT
Scope
0 – 20 VDC 100 Ω Ext.
(Not grounded)
1K

Sinewave
Generator –
100 Hz

Synch

Common Mode Rejection Test Circuit

48
Engineering 162 Laboratory Manual Spring 2004

APPENDIX: DEVICE DATA SHEETS

1N4154
2N2222A
2N3440
2N3904
ALD1106
ALD1107
CA3086
CA3096
LF353N

Note: these are abbreviated datasheets. Complete datasheets are on the class website.

49
Engineering 162 Laboratory Manual Spring 2004

50
Engineering 162 Laboratory Manual Spring 2004

51
Engineering 162 Laboratory Manual Spring 2004

52
Engineering 162 Laboratory Manual Spring 2004

53
Engineering 162 Laboratory Manual Spring 2004

54
Engineering 162 Laboratory Manual Spring 2004

55
Engineering 162 Laboratory Manual Spring 2004

56
Engineering 162 Laboratory Manual Spring 2004

57
Engineering 162 Laboratory Manual Spring 2004

58
Engineering 162 Laboratory Manual Spring 2004

59
Engineering 162 Laboratory Manual Spring 2004

60

You might also like