Discrete opamp - different topology

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

clintrubber

Well-known member
Joined
Jun 3, 2004
Messages
5,982
Location
The Netherlands
I came across this (please skip if it's been discussed before):

opa3.gif


which is an opamp. Inputs at the bases of Tr1 & Tr2, output from the collectors of Tr3,4.

Scroll down on this page, it's the third circuit
http://www.4qdtec.com/opamp.html

The author claims that "It is a circuit which is original to me and I have never seen its like used elsewhere! " and at least I myself haven't seen it like that before.


This opamp 'in use':

pre3.gif


from http://www.4qdtec.com/preamp.html

From what I've read so far I understood the fun is in the low transistor count and the unusual topology - no performance advantages as it seems.


Regards,

Peter
 
First reaction:

Although I've used plenty of "short-tailed" pairs like the two input devices, on inspection the rest of his circuit is a bit bizarre. Note that the paths after the input pair are conflicting: if you move the upper input device TR1's base positive, TR3 turns on harder and the output goes up. But for that same signal TR2's collector dumps more current into the TR4 stage, so its collector pulls down, fighting TR3! Also the local loop of R4, TR2, and TR4 constitutes positive feedback.

There may be some benefits to these internal feedback paths but they aren't obvious at first glance. The circuit may be justifiably neglected :grin:. I'm not saying it doesn't work though, but its operation depends, it would seem, on the upper path having higher gain than the lower. Maybe it's worth further investigation just to see how he got this far. (Note: PRR showed a somewhat similar differential amp with complementary input devices here a while ago, but of course his did not have conflicting feedback issues.)

If the TR4 stage were noninverting the circuit would make more sense. Balance would be achieved with the upper stage a mirror and the lower stage a common-base (TR2 collector to the emitter of TR4, base of TR4 biased by a ~constant voltage, resistor or i source in the TR4 emitter to the negative rail).

If that circuit were duplicated in a complementary fashion and the outputs combined you would have something resembling some topologies for certain very-high-slew-rate amps. Usually you need some R's between the emitters for stability. Unfortunately the noninverting stage is very hard to make with the appropriate properties of ~unlimited current, compared to a current mirror. So you wind up using two current mirrors for those paths (it gets messy and complex), or throw away two of the current signals from the input stages.

Whichever way you get his circuit to work, you have two Vbe's worth of offset and associated voltage drift, not any great problem for a.c.-coupled apps. And the slew rate will be very asymmetrical.
 
> operation depends, it would seem, on the upper path having higher gain than the lower.

That is fairly sure, if the top load is a resistor and the bottom load is a diode. Or if as in the second plan, the lower mirror has resistor degeneration.

Neat way to set current.

Current gain from feedback base to output is a few thousand. Useful but hardly universal.

Distortion (neglecting the positive feedback which hurts my head) is strongly even-order.

I can date the the E-E diff-pair to the mid-late 1960s, but where he goes from there is novel to me.
 
Good to see both your reactions. It's a bit of puzzling circuit indeed, but while some of the circuits of that website seem to be not completey tested thoughts, this opamp seems to be made for real - and I guess some obvious flaws would have shown up then.

'Made for real' at least in this version:
(be warned; the drawing style on that site might really get to you for this bigger circuit :wink: )
pwramp.gif


as from: http://www.4qdtec.com/pwramp.html.
For this power amp the lower transistor count (as compared to the more usual ways) has been lost by now of course...

Again, I guess the advantages are more in the 'unusual' realm than that there are benefits over the usual approaches, but especially since it's such a simple circuit (the first pic) it deserves some toying around with. I want to try it.

Regards,

Peter
 
Beware of a power supply sequencing/latchup problem with this arrangement. I found (in sims) that a bit of independent current into the TR4 base is needed to get the whole thing to come on. A 3.9M from +12 did the job, with R3 = 100 ohm.

Another issue is the large amount of low-frequency peaking with his d.c. feedback scheme, typical of such circuits. Better: the conventional d.c. block cap in series with the feedback divider, R5.

EDIT: There does seem to be a tiny advantage conferred by the local positive feedback loop on distortion, possibly because it raises the output impedance of Tr4 a bit.
 
I dont know...
Maybe it offers some benefit , but i will say thats alot of parts for a simple power amp.
 
Hi Brad,
That's nice, you've already thrown it in a simulator. Will as well & pay attention to the things you mentioned.

[quote author="Kit"]I dont know...
Maybe it offers some benefit , but i will say thats alot of parts for a simple power amp.[/quote]
Sure ! As said, the simple circuit may be fun, but that complete power amp... but I haven't read the page on which it was completely, might have been additional constraints.

Bye,

Peter
 
[quote author="clintrubber"]That's nice, you've already thrown it in a simulator. Will as well & pay attention to the things you mentioned.

Peter[/quote]

I was curious if there could be any benefits to the minor positive feedback loop. As mentioned they are small---the distortion reduction at 1kHz was about 20% for a selected case. It's possible this strategy could be pushed a bit by increasing the pos feedback (R in series with the base of Tr4, feed the Tr2 collector to the base) although instability is guaranteed eventually.

I do strongly recommend eliminating the 100k-100k-C4 network and just putting a cap to common in series with R5. The thing bounces like mad otherwise.

EDIT: Further sims suggest that the apparent 20% distortion reduction was an artifact. Simply taking the Tr2 collector to -12V works better than various amounts of positive feedback it seems. I used an 18k from +12 to the diodes to establish the output current.
 

Latest posts

Back
Top