Page 90 - Analog Circuit Design Art, Science, and Personalities

Bob Pease

profitable circuit. Every competitor-and many customers-realized that the P2
must cost a rather small amount to build, even allowing a few hours of work for
some special grading and matching and testing. So. people would invest their $227
and buy a P2 and take it home and pull it apart and try to figure out how it worked.
lhe story I heard (it might be partly apocryphal, but most of it probably has a lot of
truth) WiIS that Burr Brown hired a bright engineer, handed him a P2 and told him,
“Figure OLI~ how they do this-figure out how we can do it, too.” In a few days the
engineer had dismantled the circuit, traced it out, and had drawn up the schematic.
Then he anilyzed it and began experiments to be able to meet or exceed the P2’s
performance. But he couldn’t get it to work well. He tried every approach, but he
never could. After a full year. Burr Brown gave up and put the engineer to work on
some other project. Burr Brown never did get into the varactor-input amplifer busi-
nessl and I believe there is truth behind that story. Let me tell you why.
The P2 had an offset-adjust trim that was a little 20-turn trim-pot-that‘s not a
surprise. But it also had a “gain adjust.“ This was not any ordinary gain adjust. This
was a 20-turn variable trim capacitor-a differential piston capacitor-which the
user could trim. per the instruction sheet, with a tiny little 411en wrench or hexag-
onal key. But it did not just have a linear control over the gain. If you trimmed the
pot ovc: to one end. the gain might be at 300 or 500, and then as you trimmed it
closer 10 the center, the gain might rise to 700 or 900-and then, suddenly. the gain
would pop up to 71000 and then to 10,000 before the nonlinearity made the gain fall
off again, when you turned the adjustment too far. The test tcchs called this. “going
into mode.” 1 used to wonder what they meant by that.
Several years later, George Philbrick brought me in to help him on an up-dated,
upgraded version of the P2-the ‘’P2A”. Wc had 10 redesign the P2 because Philco
was stopping the manufacture of those old 2N344s, and we couldn’t buy any more,
so we had to redesign it to use the more rnodern (cheaper) Silicon Mesa transistors.
such as 2N706 or 2N760 or whatever. When we bought them from Texas
lnsirumcnts. they were labelled “SM0.387.” I had a new circuit working pretty well.
with !he help arid achice of Gecrgc Philbrick, because Hob Malter had passed awy.
sadly: sfter a iong boul with multiple sclerosis. about 1966. Anyhow, 1 was getting
some results with the silicon-transistor version. but the improvemcnts weren‘t
coming along as well or as kist as I expected. so I went back to fool around with
sewnil rea! P~s, and to study them.
Thc original P2 had an apparent imbalance at the output ol‘its demodulator. Weli.
that looked kind of dumb, Lhal the first DC transistor would he turn.ed off unless
thcrc WIIS a pretty big signal coming out of the demodulator. To turn on the DC
ti-msistor, you had to have a considerable imbalance of the RF. So I took onc unit
atid modified it to bias the demodulator about I V,, down from the positive supply.
so it would not have to handle a great amount ol’ signal just to drive the DC tran-.
sister Q7. Refer 10 rhe schematic of the P2 !Figure 9-3). Normally. Q6’s emitter was
connectcd directly lo thc +I 5 V bus. I disconnected it by removing link X--Y aiid
conncctcd ii to ;1 bias diode. Yes, the KF amplifier ran with less RF signal at bal-
aricc--lxit ~hc gain reluscd to “come inlo modc.” So chat ”in~pro\iei7ient” schcrnc
was unusable. Now. what was thut trying to tell me?
Ater WJX more study, 1 planned a few more experimentsl and thcn I tried pulling
apart thc iwo PC boards so 1 could access some of the signals down in between 1he
boards. As I ensed the two boards apart (with power i3N). the gain ”jumped out or
rnotle.“ i gradually realized the P2 amplifier was running. all thcse years: as a reflex
miplilkr. The ”gain adjust” consisted or changing the phase bctween the oscillator
ax: thc bridge. so when thc ampli!ied siglial came down to the end of the RF ampli-

71