Boosting signal generator or op amp outputs to >+30 dBu

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Boosting signal generator or op amp outputs to >+30 dBu

Post by mediatechnology »

I'm playing around with JR's circuit fragment and found that the feedback connections to the op amp input are reversed.

Since the CE output stage is inverting the op amp input terminals need to be reversed with NFB returned to the non-inverting input.
As drawn, feedback is positive.
With an emitter follower buffer after the CE stage (T3 driving an NPN) it would be correct as drawn.

Image
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Boosting signal generator or op amp outputs to >+30 dBu

Post by JR. »

mediatechnology wrote:I'm playing around with JR's circuit fragment and found that the feedback connections to the op amp input are reversed.
yup :oops: :oops: :oops:
Since the CE output stage is inverting the op amp input terminals need to be reversed with NFB returned to the non-inverting input.
As drawn, feedback is positive.
With an emitter follower buffer after the CE stage (T3 driving an NPN) it would be correct as drawn.
not sure an emitter follower (a follower) would invert polarity.
Image
I hope I offered my schemo as unverified. :oops: I melted plenty of solder over such stuff but like 35 years ago,

I vaguely recall another variant. Where the common emitter pair was biased up by a fixed voltage divider, and the op amp output resistors connect instead to the emitters (like a cascode topology). This brings the output back into correct polarity while limiting the drive current to opamp output. Another trick that I recall playing with for the non-inverted topology, is directly connecting a cap from the opamp output to the overall output, to help at very high frequency, which is generally expected to be small amplitude. Analysis for the cap shunt is a little wierd...since it literally needs to decouple for high voltage outputs (probably Ok for GP wide band audio) and all of these circuits are a little Rube Goldberg, Perhaps just bite the bullet and design a discrete power amp.

FWIW at Peavey we made millions of channels of power amps with cheap opamp front ends (because a $0.15 opamp is a lot better than $0.15 worth of discrete circuitry). The all too common "driven supply", where output of modest voltage opamp drives a common emitter power stage, with collectors connected to the tops and bottoms of charged PS capacitors. The speakers connects to the common point between these two caps. This topology uses relatively low voltage components and is cheap... perhaps not very hifi, but adequate for general purpose amp channels.

JR

[edit- I screwed up explaining the driven rail amp topology :oops: more brain farts.
Last edited by JR. on Mon Apr 29, 2013 1:24 pm, edited 1 time in total.
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Boosting signal generator or op amp outputs to >+30 dBu

Post by mediatechnology »

not sure an emitter follower (a follower) would invert polarity.
You're right about that. Need to look at those BGW schematics to see where the polarity got re-inverted.

Works pretty well though. With +/-30V supplies, +27 dBu into 600R I'm seeing 0.001% THD - a lot of it generator.

I'm trying to find a good value of Re to provide Q-point stability without some form of thermal feedback. Right now I'm at Re=33R with an output stage Iq of ~25 mA.

With +/-31.5 supplies (max from the bench supply) I can get ~ +29 dBu.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Boosting signal generator or op amp outputs to >+30 dBu

Post by JR. »

There are many cobwebs in my memory. I used that BGW topology in a DIY amp I built in the early '70s. My recollection is opamp was relatively high speed (LM308? the one with emitter degeneration on input LTP) and opamp input pins were probably swapped to make polarity right. IIRC this amp was all of 0.25% THD at 20kHz and 8 ohms so nothing spectacular. While back in 70's 4x250W seemed like a lot of power amp to me. :D

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Boosting signal generator or op amp outputs to >+30 dBu

Post by mediatechnology »

This is a very poor scan of the BGW750C.
I found a "better" redrawn copy but its full of errors including "correcting" the op amp input pins.
The one below is original and correct.

It looks like that negative feedback is returned to the non-inverting input since the common emitter stage following the LM318 are inverting.
(You gotta zoom to see this and it will leave you with a headache when you do. Theres a dot under that green line and it extends into pin 3.)
Note the 100pF local NFB around the LM318 since there's probably a ton of phase shift through the output stage.
All further stages are non-inverting emitter followers.

Image
BGW750C Schematic
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Boosting signal generator or op amp outputs to >+30 dBu

Post by JR. »

Yup, that's the general level shift topology, I think mine was based on an even older model.

Mine used 2n3773, slow as dirt, NPN power devices, quasi comp to make PNP side. That looks like they used later fully complementary motorola pairs. Also the early BGW I copied didn't have current limiting detecting off the emitter resistors, in fact I added current limiting similar to what they used there.

The old BGW I based mine on used a whole amp crowbar protection scheme that just collapsed the entire power supply when it detected a problem. Not practical for my 4x250W kluge.

I might have used a LM318 front end. That has LTP degeneration and was relatively fast for it's time.. I don't remember small details.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Boosting signal generator or op amp outputs to >+30 dBu

Post by mediatechnology »

What do you suppose is a simple way to temperature compensate this?

Image

If the CE voltage gain stages were made small-signal and drove an additional output pair, it would be easy to have a Vbe multiplier between the output stage bases.
Then it looks remarkably like a power amp...

The only really ultra simple way I've tried thus far with a single-stage output is to heavily degenerate the emitters.
It works but is lossy.
I could get fancy with current sources but I think something simple is best.

I'm using an Re of 100R (R19), Rbe of 220R (R12) and an Rbase-drive (R19) of 10K.
Vcc ~31.5V and an Iq (Ie) of ~22.5 mA.
Performance is pretty darn good but limited to about +26 dBu due to the high Re.
With a good heatsink Iq is very stable but the values have to be tweaked for >63V operation and to approach +30 dBu.

Also added base-collector capacitors of 100 pF, a global Cfb of 100pF||10K and a local Cfb of 22 pF around the LME49710.

I should really post a schematic but I'm afraid someone would actually build it.
I don't consider it fully-tested.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Boosting signal generator or op amp outputs to >+30 dBu

Post by JR. »

mediatechnology wrote:What do you suppose is a simple way to temperature compensate this?
In general the Vbe should be small wrt Re so no need to compensate. The Re will prevent thermal runaway. The cap coupling is so you can have low quiescent DC current but still come up with lots of drive. Of course that drive voltage swing there comes out of Vp-p.
Image

If the CE voltage gain stages were made small-signal and drove an additional output pair, it would be easy to have a Vbe multiplier between the output stage bases.
Then it looks remarkably like a power amp...
IIRC the hifi amp I made for my little sister used power darlington, and perhaps large for power amp emitter Rs (something like .47 ohm) but again my memory of a one off build 3+ decades ago is not sharp.

The only really ultra simple way I've tried thus far with a single-stage output is to heavily degenerate the emitters.
It works but is lossy.
I could get fancy with current sources but I think something simple is best.
You can run the output stage starved and use a very fast opamp to keep distortion fair, but this will not be very hifi. For real hifi a more conventional gain stage and output stage. using the opamp for it's loop gain margin.

I'm using an Re of 100R (R19), Rbe of 220R (R12) and an Rbase-drive (R19) of 10K.
Vcc ~31.5V and an Iq (Ie) of ~22.5 mA.
Performance is pretty darn good but limited to about +26 dBu due to the high Re.
With a good heatsink Iq is very stable but the values have to be tweaked for >63V operation and to approach +30 dBu.

Also added base-collector capacitors of 100 pF, a global Cfb of 100pF||10K and a local Cfb of 22 pF around the LME49710.

I should really post a schematic but I'm afraid someone would actually build it.
I don't consider it fully-tested.
Please don't, just one of a hundred ways to skin this cat, and I hear you can buy an opamp that drops in... and works.

JR

PS: FWIW I wasted too much time while at Peavey working on a variant Vbe multiplier to set class A current in class AB output stage, where it actually regulated the minimum current in the power devices. Since it didn't drive the power devices with a DC voltage, it didn't have thermal runaway issues. My boss at the time was too much of a C_nt and wouldn't let me put it into a modest volume 35w amp to production test it with real volume builds. If it worked as good as I thought it would it obviously would have spread into other amps. The Peavey approach for cheap amps was a special diode engineered to track with the motorola power devices Peavey used. It worked well enough, but my was was better IMO. 8-)
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Boosting signal generator or op amp outputs to >+30 dBu

Post by mediatechnology »

In general the Vbe should be small wrt Re so no need to compensate. The Re will prevent thermal runaway.
Thank you. Yes, the IR drop across Re is ~ 3 to 4 times Vbe and it's stable and free from runaway with a modest heatsink.
The only reason for Iq regulation would be to provide a wider supply range without having to adjust resistors.

In response to my reluctance to post an interim work-in-progress schematic:
Please don't, just one of a hundred ways to skin this cat, and I hear you can buy an opamp that drops in... and works.
Some broader questions:

So are you suggesting that I just end this thread the same way I did with the precision rectifiers?
At least you didn't tell us we could do this in DSP. That's refreshing.
I saw no point in continuing that thread due to cold water being thrown on it.
I should have anticipated that this time and just not bothered.
It always seems to happen as a thread develops life. Why is that?
I hear you can buy an opamp that drops in... and works.
You mean the LTC6090?
It's HF distortion performance leaves a lot to be desired and its sole-sourced.
It's only good for about 10 mA peak.
Driving a 600R load, the LTC6090 is capable of a whopping +15 dBu before it craps out. We need about 60 mA.
My thought was that someone might want to do this with an off-the-shelf op amp and a couple of transistors they could buy almost anywhere and actually drive 600 Ohms.
I've seen SGs +/-40V op amp but it has a lot of bits and there's really no need for a fully-discrete solution.

Do I correctly infer that you don't think there's any value in posting a refined and working schematic or continuing the thread?
If so, why?
Are we boring you?
Why the flippant comment?

Two people here have asked for an output with +30 dBu capability and I've tried to make a circuit you posted - which had an error - tested and available.
There are tons of circuits capable of +30 dBu with low distortion.
But where are they?
Why are people asking for them if they are so obvious and easily-found?

I wanted to provide something simple and of value.
Once I had something tweaked I thought I'd be generous to share a schematic of it.
Maybe I shouldn't be so gracious with my time now.
juniorhifikit
Posts: 139
Joined: Sun Oct 04, 2009 8:37 am
Location: Oakland/Paris

Re: Boosting signal generator or op amp outputs to >+30 dBu

Post by juniorhifikit »

Well, I for one, am still following this thread with great interest and appreciate all of the input and background from all camps. My silence stems from the fact that I can't add anything useful. Please continue!
Post Reply