Differential amplifier behaving weirdly

Sep 07, 2007 3 Replies
T

So, in this circuit:

formatting link
if I breadboarded (eww!) it today, omitting the JFETs and their current sinks, using 680 ohms for the collector loads, 470 ohms for the emitter follower loads, +60V for the HV supply, and replacing the common mode servo with a



500 ohm variable resistor in the current sink's emitter (its base is instead biased with two diodes, like the JFET CCS's). So the signal path is about as shown.

Checking the diff alone, the waveform on the driven transistor looks just fine, not even needing compensation (not suprising; I wasn't expecting to need any with the 5179's anyway). But the opposite side (which is biased with a 100k pot (from +/-9V) bypassed to GND with 0.1uF), its collector waveform looks notably different. It has a long time constant (I forget what), which has nothing to do with compensation: in fact, adding 220pF across the emitters causes a steep peak which falls off, then rises up the slow slope!



BTW, I first tried the output cascoded (using 3904's for the bottom). Not very useful: it oscillated like heck. RF parasite at all times, weird squegging behavior, touch sensitive, etc. As shown seems to behave slightly nicer! ;-)



Tim


-- Deep Fryer: A very philosophical monk. Website @

formatting link


Vote
Twitter Facebook LinkedIn
G

circuit:

formatting link
I can't bring up that link.

Vote
Twitter Facebook LinkedIn
T

circuit:

formatting link
%2...

Apparently, it line wrapped.

Re: OP: evidently, it was the CCS. Looks like one of the diodes went open, somehow. Plain 1N914 I've used before, but a whole 9V across it. No, not backwards! I don't know how that would explain most of the performace I was getting, or the reduction in output speed on the far side.

Today I increased the diff pair's bias current (470 ohm collector resistors as shown) added inductors to them (not exactly peaking inductors, but L in any case) and added compensation to the emitters. Now the output drivers (measured without the 2SC1569's in place) show a 10ns rise/fall time, which is roughly that of the signal generator as well, so it's going quite fast. With the 1569's in and tweaking for optimal performance, I get 40, or was that 20, ns edges at the output. Bandwidth is over 15MHz. I think I'm ready to solder this thing to FR4 and see how it does in the Heathkit scope.

Tim

-- Deep Fryer: A very philosophical monk. Website @

formatting link

Vote
Twitter Facebook LinkedIn
M

circuit:

formatting link

It was broken. All you have to do is splie it back together.

Service to my country? Been there, Done that, and I\'ve got my DD214 to prove it. Member of DAV #85. Michael A. Terrell Central Florida
Vote
Twitter Facebook LinkedIn

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required