Page 94 - Analog Circuit Design Art, Science, and Personalities

Bob Pease

learn anything much. 4nd in retrospect, well, Bob had a lot of good hunches, and he
probably had a good hunch in this respect also.
Of course, if you wait long enough, any good thing can become obsolete. -4s of
1989, yo9 could buy low-leakage amplifiers such as the NSC LMC660, with input
currents normally less than 0.004 pA, for about SO cents per amplifier ($2.00 for a
quad). But what do you expect after a 30-year wait'?

Notes on George Philbrick's P7 Circuit

1. The AC amplifiers arc all supplied through a single 4.7 kfl('?!) resistor! George
wanted to run all 4 AC amplifier stages on barely 1 mA total! In the P2, Bob
Malter was willing to spend 4 mA. I could never understand why George was so
unwilling to spend just 15 mW for the entire four-stage AC amplifier when he
spent 30 mW to bias up the output stage (27 k on Module 6). Maybe if he fed
any more current through those AC amplifiers, they would break into song and
oscillate hopelessly?! Because in the layout, the output of the fourth stage is
right next to the input of the first AC stage'?! Thosc of you who are not chess
playcrs might like to know that in the notation of chess. "?!" signifies a h1undc.i..
2. Likewise, the P7 oscillator was intended to run on less than 11.3 mA ('?!), whereas
the P2 spent 1.5 mA. I checked the actual P7 circuit to see if thesc values repre-
sented a typo error-but they didn't.
3. The P7 I have uses two AC amplifier modules-four stages of transistor-but
the arrangement of the upper and lower (mother and father?) boards left room
for three amplifier niodules-six stages of AC gain. You can see the gap in the
middle of the assembled unit: where another two stages could have fitted in. But
if you had six stages. then the possibility of oscillation would become hope-
lcssly bad. No wonder George backed up to go with just four stages.
4. The germanium IN100 diodes in series with the input.s are intended to act as
low-impedances near null but to act as current-limiters (just a few microam-
peres) when overdriven. That is what George intended--a neat concept. Rut in
actwlity. 1 bet they made bad errors when they rectified any ambient noise, and
f bet they had awfd leakages when operated in ambient light. Furthermore, if
you ever got around to running this amplifier with feedback. you would find they
add a lot of phase shift. At room tcmperature. they would cause a lag of perhaps
25 kil and 600 pF, or a IO kHz roll-of[. If you get it cold: the break at +5"C
wouid be at 2.5 kHz. This confirms my suspicions that George never really got
11ie DC: operation working okay, so he never even hqan to think seriously about
AC response. The circuit shows no cvidence of a big filter around the output
stage to give 6 dB-pcr-octa\;e rolloff.

Comments on Bob Malter's P2 Circuit

I. Obviously, 30 years ago this was a high-level industrial secret. Rut as I mcn-
tiorled: even if I gave you the schematic, that would not help you make a 1J2 thai
works. Since the P2 has been out or production for more than 20 ycars, this is
more like indu.strial archaeo!ogy than espionagc.
2. The doubled capacitors (several places where you see 17.5 pF in parallel with IO
pFj) were a:-ranged so that the test technicians could do some cwarse trims by
snipping out one or another of the caps. Much judgment and cxpcrience were
needed. Thcrc x.cre many other places ivhcrc discrctionary trim resistors and