Widlar super follower

Gerhard Hoffman was evaluating this topology in LTspice a few years back, so search his posts.

Am 02.09.26 um 23:08 schrieb JM:

Hoffmann with 2 n, pleaze :-)

probabbly this here: <

cheers, Gerhard

Or you could buy an opamp.

I like BUF602, a fast active follower.

What is D1 for?

Am 03.09.26 um 01:34 schrieb john larkin:

compensating the Q3 BE-drop?

Yes I should have remembered that from my time in Vienna. I remember scratching my head when told that a mann is a man, but a man can be a woman, or a man, or a mixed group. And a mann is also a husband - can't remember what a wife is called.

Just as well I'm not still there, I took a liking to Brühwurst and started working my way through all (800?) varieties. I would be like Mr Creosote had I stayed.

Probably. I've seen that type of thing often enough in high temperature (upto circa 200C) downhole tools.

But the low end is a current sink!

Many thanks Gerhard , yes that’s it!

The Bob Widlar's LM308 (LM108, LM208) relied on odd transistors with a very thin base which had very high current gain - around 3000 - but a very low breakdown voltage so that they could only be used in cascode. How you made them with a standard analog process, and mixed them with regular transistors is not something I can remember - the National Semiconductor applications notes didn't spell that out.

Yes. In his original monolithic designs the superbeta transistors he used had a Vce breakdown voltage of only about 1 to 1.4 volts due to the thin base region.

Makes no sense. The low end of the diode is a current sink.

And this isn't a superbeta design.

I wonder how it works without the diode.

I don't follow - this is just an attempt to copy a monolithic design (if this was built as is with the discretes shown I'm pretty sure it would oscillate) and in that design superbeta transistors would have been used, so without D1 you would likely get punch through of Q2. It's also good to try and balance the dissipation in Q1,2.

Sorfry, none of that makes sense.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

If you don't know about super-beta transistors, it probably won't.

Bob Widlar's Linear Application note AN-29 goes into the subject at length, and references his 1969 paper "Super Gain Transistor for IC" in the IEEE Journal of Solid State Circuits volume SC-4 number 4, August 1969.

I used the LM308 from time to time - it is a very neat op amp for monitoring high impedance current sources. John Larkin may have managed to miss it.

Am 04.09.26 um 08:09 schrieb Bill Sloman:

And yes, the circuit will require some compensation for stability. At f > 1.5 MHz the input impedance turns negative. ( real part of (input voltage / input current), top trace )

Paralleling more transistors does not bring the noise further down. More standing current does, but that hurts, also.

The top left corner of the circuit provides a noisy VCC. Noise level can be set directly by E1 in nV/rt Hz to check amplifier Vcc noise suppression. The slight offset provided by V3 prevents a div by 0 somewhere in LTspice.

Your posts would be more interesting without the constant venom. But then it takes 2 2 tango.

For those willing to experiment:

Version 4 SHEET 1 2052 1328 WIRE -368 -1136 -448 -1136 WIRE -288 -1104 -320 -1104 WIRE 16 -1104 -288 -1104 WIRE 224 -1104 16 -1104 WIRE -320 -1072 -320 -1104 WIRE -368 -1056 -368 -1136 WIRE -496 -1024 -512 -1024 WIRE -448 -1024 -448 -1056 WIRE -448 -1024 -496 -1024 WIRE -448 -1008 -448 -1024 WIRE 16 -992 16 -1104 WIRE -416 -928 -448 -928 WIRE -368 -928 -368 -1008 WIRE -368 -928 -416 -928 WIRE -320 -928 -320 -992 WIRE 224 -928 224 -1104 WIRE -416 -912 -416 -928 WIRE 16 -880 16 -912 WIRE 160 -880 16 -880 WIRE 176 -880 160 -880 WIRE 16 -864 16 -880 WIRE 224 -752 224 -832 WIRE 224 -752 -160 -752 WIRE 336 -752 224 -752 WIRE 384 -752 336 -752 WIRE -160 -704 -160 -752 WIRE 16 -704 16 -800 WIRE -496 -656 -512 -656 WIRE -448 -656 -496 -656 WIRE -352 -656 -368 -656 WIRE -304 -656 -352 -656 WIRE -224 -656 -304 -656 WIRE 224 -656 224 -752 WIRE 224 -656 80 -656 WIRE -352 -640 -352 -656 WIRE 224 -624 224 -656 WIRE -96 -608 -160 -608 WIRE 16 -608 -96 -608 WIRE -512 -544 -512 -656 WIRE -96 -544 -96 -608 WIRE -240 -416 -336 -416 WIRE -96 -416 -96 -464 WIRE -96 -416 -240 -416 WIRE 224 -416 224 -544 WIRE 224 -416 -96 -416 FLAG -320 -848 0 FLAG -336 -336 0 FLAG -288 -1104 vcc FLAG -496 -656 vgen FLAG -416 -912 0 FLAG -496 -1024 v3p FLAG -512 -464 0 FLAG -304 -656 rein FLAG 336 -752 raus FLAG -240 -416 VEE FLAG -352 -560 0 SYMBOL voltage -320 -944 R0 WINDOW 0 -32 56 VBottom 2 WINDOW 3 32 56 VTop 2 WINDOW 123 0 0 Left 2 WINDOW 39 -52 151 VTop 2 SYMATTR InstName V1 SYMATTR Value +6 SYMBOL voltage -336 -432 R0 WINDOW 123 -101 112 Left 2 WINDOW 39 -130 80 Left 2 WINDOW 3 -112 50 Left 2 WINDOW 0 -117 20 Left 2 SYMATTR Value2 AC 0 SYMATTR SpiceLine Rser=0 SYMATTR Value -6 SYMATTR InstName V2 SYMBOL e -320 -1088 R0 WINDOW 0 35 76 Left 2 WINDOW 3 35 98 Left 2 SYMATTR InstName E1 SYMATTR Value 2 SYMBOL res -464 -1152 R0 WINDOW 3 -66 -25 Left 2 WINDOW 123 -66 -4 Left 2 SYMATTR Value 60.347 SYMATTR Value2 temp=27 SYMATTR InstName R10 SYMBOL voltage -448 -1024 R0 WINDOW 0 -32 56 VBottom 2 WINDOW 3 32 56 VTop 2 WINDOW 123 0 0 Left 2 WINDOW 39 -57 138 VTop 2 SYMATTR InstName V3 SYMATTR Value 0.1µ SYMATTR SpiceLine Rser=0.1 SYMBOL voltage -512 -560 R0 WINDOW 123 41 64 Left 2 WINDOW 39 39 39 Left 2 WINDOW 3 41 92 Left 2 WINDOW 0 36 11 Left 2 SYMATTR Value2 AC 1 SYMATTR SpiceLine Rser=0 SYMATTR Value SINE(0 1 1000 0 0 0 20) SYMATTR InstName V5 SYMBOL npn 160 -928 R0 SYMATTR InstName Q3 SYMATTR Value fzt851 SYMBOL npn -224 -704 R0 WINDOW 0 -37 91 Left 2 WINDOW 3 -41 121 Left 2 WINDOW 39 -40 150 Left 2 SYMATTR InstName Q1 SYMATTR Value Q2sd2704k SYMATTR SpiceLine m=1 SYMBOL npn 80 -704 M0 WINDOW 0 38 115 Left 2 WINDOW 3 -51 142 Left 2 WINDOW 39 21 162 Left 2 SYMATTR InstName Q2 SYMATTR Value Q2sd2704k SYMATTR SpiceLine m=1 SYMBOL current 16 -992 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName I1 SYMATTR Value 20m SYMBOL current -96 -544 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName I2 SYMATTR Value 40m SYMBOL diode 0 -864 R0 SYMATTR InstName D1 SYMATTR Value bav99s SYMBOL res 208 -640 R0 SYMATTR InstName R1 SYMATTR Value 200 SYMBOL res -368 -656 R0 SYMATTR InstName R2 SYMATTR Value 10k SYMBOL res -352 -672 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R3 SYMATTR Value 0.1 TEXT 256 -1072 Left 2 !;tran 10m TEXT 264 -1104 Left 2 !.op TEXT 256 -1040 Left 2 !;noise v(raus) V5 dec 10 1 10e6 TEXT 264 -1016 Left 2 !.ac dec 1000 0.1 100Meg TEXT 256 -952 Left 2 !;.temp 20 40 60 TEXT 256 -984 Left 2 !.include d:\\lib\\spice\\ghf_spice_lib.txt TEXT -200 -392 Left 2 ;Model name reminder: BAV99s\nm_bfp640 bc850c fzt851 bc817 bfr93a \nZXTN2018F BF=270 No KF no AF\nZXTN25020DFL BF=550 no KF no AF\n2SD2704K BF=2300!!! no KF no AF TEXT -336 -1360 Left 2 !.MODEL Q2SD2704K NPN\n+ IS=250.00E-15 BF=2.3219E3 VAF=100 IKF=25.778E-3 ISE=270.34E-15\n+ NE=1.8069 BR=45.088 VAR=100 IKR=9.4796 ISC=411.09E-15\n+ NC=2.0254 NK=.54182 RE=.4 RB=3.1823 RC=58.585E-3\n+ CJE=8.7706E-12 MJE=.66324 CJC=15.709E-12 MJC=.53302 TF=3.8352E-9\n+ XTF=32.147 VTF=472.13 ITF=85.294 TR=7.0574E-9 XTB=1.5000\n+ AF = 2.0 KF=5.0E-9\n* AF and KF guessed by gerhard AF=2 == 1/f PSD TEXT -552 -792 Left 2 ;KF = flicker noise coefficient\nAF = flicker noise exp\nEF = fl. noise freq. exp SPECTRE? TEXT 32 -512 Left 2 ;Q1 und Q2 tragen gleich viel zum Rauschen bei.\n210 pV * 1.414 = 305 m=4 bringt nix. Aber mehr strom!\nzxtn25020 ist wesentlich besser als Q2sd2704k bei Rauschen\nbase bias 17uA sd2704

Gerhard

John Larkin does seem to experience it (and anything else that isn't fulsome flattery) as venom. It was more an expression of bewilderment on my part.

We probably also need to worry about the Early effect in that follower as well. It isn't big but it might matter in a "super follower".

FTZ851 is not a super-beta. If it were, one might want the diode b-e, reversed.

It's almost 60 years old!

I think I may have used it, decades ago. Aside from low Ib, it was a pretty bad amp.

My favorite gumdrop amp now is OPA197, which has 5 pA typ bias current. It's a pretty good comparator too.

There are some fA opamps now too.

What opamps are you using lately?

We accidentally met Widlar's girlfriend at the Washington Square Bar&Grille, not long before he died. She had no idea that he was famous. She was concerned about his drinking.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Not sure I follow you, the diff pair both operate at zero c-b voltage so no Early effect there and the emitter follower stage is inside the feedback loop?

Oh, that's what the diode is for. Clever.

But 0.6 c-b might work as well.

so no

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics