Bob:
While my full-time job is writing for Electronic Design, I
still teach part-time and work on an NSF grant that is
attempting to update the electronics curricula in community
colleges. From my observations in my own college and
across the country, most curricula are out of date with what is
going on in the industry. (True, but not disastrously bad. We can’t
ask that education for techs or for EEs be really up to date. That
has almost always been impossible. /rap) There is too much
emphasis on BJTs and little or poor coverage of MOSFETs and ICs.
(Yeah, but you can get and buy and solder BJTs, and the circuits
will work. You can’t get MOSFET kit parts worth crap! And if you do,
they don’t work well. In most cases, even Spice works better than
making breadboards of MOSFETs. Of course, we know the exceptions...
If a kid has learned a little about MOSFETs and a lot about
BJTs, we can convert him over. /rap) I would love to get your opinion.
Could you answer a few questions? First, what would you say
the percentage mix of BJT/MOSFET circuits is in ICs (or discretes)
for linear and digital? It must be close to 100% digital, but what
about linear? (I think it is about 45/55, but we work on weird projects.
Many people do not recognize that to make low-power circuits,
CMOS is not inherently low-power. It takes a lot of work. So,
we make micropower op amps using BJTs—and very fast ones, too.
When you go that fast, you have to trust Spice, somewhat. /rap)
Second, have you seen any engineering techs recently? And does
NSC employ them? (Yes and yes. I have interviewed some and lectured
to some recently. /rap) These are the guys that help engineers
with breadboarding, test, etc. I haven’t seen many in years,
although I used to be one. (I actually repaired Philbrick K2Ws and
related gear way back years ago.) What is your take on this? (Our
technicians can repair ANYTHING—yes, even K2Ws. (I still have
some and use them.) /rap) Third, how important is it for a tech to
know detailed BJT and MOSFET biasing, etc.? (Generally, not. Biasing
FETs is almost impossible because a good bias depends on the
match of the FETs, and that happens well on only one chip. /rap)
Fourth, most of us working on the NSF grant think that a tech
needs more of a systems view today as opposed to a detailed circuit-
analysis background. Do you concur?
• Louis E. Frenzel, Communications/Test Editor
• Pease: I tend to agree. Often, a good technician must use and
understand op amps—and measuring equipment, DVMs, spectrum
analyzers, and automated test equipment. Circuit analysis
is a specialty, and even for engineers, this is challenging. Of
course, they should be aware of circuit analysis and bias setup.
Even 20 and 40 years ago, we did not demand them to be
experts at that. We’re now asking our senior techs to do more
and more analysis of data and to tell us when things look right—
and when things look suspicious or “funny.” They’re usually quite
good. Of course, we’ve had the luxury of several excellent techs
out of the College of San Mateo, and with just a few years of
mentoring, many of them have become circuit engineers.
Dear Bob:
Your article “What’s All This Input Impedance Stuff, Anyhow?”
(Sept. 7, 2004, ED Online 8576) describes a single op-amp
differential amplifier circuit with a gain of 100; resistor pairs of R1
= 1k and R2 = 100k; and front-end buffers ignored. I do not argue
about circuit gains of voltages. They seem to be okay. But the
impedances are not. Various sources (like NS’s Linear Application
Handbook: AN-20, AN-29, etc.) state that the input impedance of
the circuit at the inverting input is equal to R1. I do not accept this
unconditional statement! When inputs are equal and in opposite
phase, as in an ideal case, the input impedance is actually 500 Ω.
(Ah, but that is a special condition! /rap) In your article, the impedance
is claimed to be R1 = 1k (according to the handbook statement),
which is incorrect. When inputs are driven to 10 V dc each,
the impedances are 101k on both sides, just as in your article. I
agree. But this 101k impedance is not equal to R1 = 1k. If gain = 1
(all four resistors equal), then impedance is 2/3 * R. I argued
about this with Dr. Michael Ellis a few years ago, and after some
calculations and simulations, we agreed. (First, I was wrong when
believing the unconditional statement above.) But when the input
voltages are not equal, the situation changes further. For example,
if positive input is 100 times negative input and in opposite phase
(gain = 1 circuit), then the impedance is only 1/50 * R. You may
simulate various conditions as I did and find out that the impedance
varies vastly. (Well, if you pick the right “special conditions,”
anything can happen! /rap) Obviously, we must come to the conclusion
that the impedance at the inverting input node is not equal
to 1, neither constant, but depends on magnitudes and phases of
input voltages and is therefore largely variable. In general, when
inputs are not correlated with amplitude or phase (random input or
noise), one can not guess the impedance. Do you agree?
• Eero-Pekka Mand
• Pease: I tend to agree. But (a) there is nothing simple about
this, and (b) you have waited three years to comment! So (c),
shall we consider the case where the input is a transformer
winding, not center-tapped? Or (d) cap-coupled? If the caps are
big enough, it may still work. But you may need 100X bigger
capacitors on the negative input. Let’s discuss. This might lead
to “What’s All This Z In Stuff, Anyhow? (Revisited).”
Comments invited! rap@galaxy.nsc.com —or:
Mail Stop D2597A, National Semiconductor
P.O. Box 58090, Santa Clara, CA 95052-8090
BOB PEASE obtained a BSEE from MIT in 1961 and is Staff Scientist at
National Semiconductor Corp., Santa Clara, California.