Unusual audio power amp configuration

Thread Starter

AnalogKid

Joined Aug 1, 2013
8,693
Unusual configuration for an audio power amplifier. I've seen this in other Peavey amps. The output of the amplifier is grounded, but the power transformer centertap is not. I sorta see how this works, but can't figure out why someone thinks this is an improvement over the traditional connectiions (either sonically, electrically, manufacturing-ly, etc.).

ak
1603326257757.png
 

ronsimpson

Joined Oct 7, 2019
1,017
why someone thinks this is an improvement over the traditional connections
Two engineers, after a hard day designing audio amps, working on their third drink;
Engineer1:"I hear said that ground is relative. You can actually call any point in a schematic ground. Its just a name."
Engineer2: "The speaker sits in-between the transformer-CT and the power transistors emitters. Why not call the other end of the speaker ground?"
Engineer1:"Call the patent Attorney"
---edited---
Engineer2:"By morning I could not remember what was so good about it."
 
Last edited:

AlbertHall

Joined Jun 4, 2014
10,558
I think what it does is to move the ±38V rails relative to ground.
So when Q4 is conducting, the voltage between the +38V rail and ground is less than 38V and vice versa for the -38V.
Therefore the voltage across Q4 will be lower when it is conducting and so less heat and greater efficeincy.
 

Ian0

Joined Aug 7, 2020
750
It makes precisely no difference to the power dissipation, the loudspeaker is connected between the same two points in the circuit. The difference is that the one normally connected to earth is the output, and the one normally called the output is connected to earth.
The big difference is in the drive circuit. The voltage amplifier no longer has to output a voltage swing equal to the power supply, it now only has to output enough voltage to turn a transistor from on to off - a couple of volts or so, a bit more for MOSFETs.
The drive circuitry can now be an op-amp running on +/- 5V rails, instead of discrete circuitry with +/-35V output.

There are, of course, disadvantages:
1) it needs a separate power supply for each channel (if it is stereo)
2) it needs both a high voltage supply referenced to the output AND a low voltage supply referenced to earth.
3) The phase is reversed, so the circuitry needed to keep it stable is different. You will note that the feedback returns to the non-inverting input of the op-amp, and most op-amps are designed NOT to have an inverting stage inside the feedback loop. Generally, if it is op-amp-driven the HF performance is not stellar.
 

Thread Starter

AnalogKid

Joined Aug 1, 2013
8,693
The only time I've seen this in band amplifiers, both mono; a keyboard amp from the early 80's, a bass amp from the early 90's, and this one (which looks to be about the same vintage based on the schematic style). all Peavey.

ak
 

Ian0

Joined Aug 7, 2020
750
Used extensively by C Audio ( if I remember correctly) with two transformers, and a lot of Chinese clones.
The Chinese clones tend to be rather blowy-uppy because many had been made with fake 2SC5200/2SA1943 transistors.
Changing the transistors for real ones became almost a full-time job for a while.

I built one, but even with a discrete drive stage, could never quite get the stability or low distortion that I could with the “conventional” arrangement.
 

Thread Starter

AnalogKid

Joined Aug 1, 2013
8,693
3) The phase is reversed, so the circuitry needed to keep it stable is different. You will note that the feedback returns to the non-inverting input of the op-amp, and most op-amps are designed NOT to have an inverting stage inside the feedback loop. Generally, if it is op-amp-driven the HF performance is not stellar.
The overall circuit is not inverting. The opamp is inverting locally, but so are the output transistors:

U2B - input goes down.
U2B output goes up.
U4 emitter tries to go up, but is bolted to GND, so ...
U4 collector goes down.
Transformer upper winding goes down.
Transformer centertap goes down, and speaker voltage goes down.
U2B + input goes down.

ak
 

Thread Starter

AnalogKid

Joined Aug 1, 2013
8,693
The big difference is in the drive circuit. The voltage amplifier no longer has to output a voltage swing equal to the power supply, it now only has to output enough voltage to turn a transistor from on to off - a couple of volts or so, a bit more for MOSFETs.
The drive circuitry can now be an op-amp running on +/- 5V rails, instead of discrete circuitry with +/-35V output.
And there it is.

ak
 

Dodgydave

Joined Jun 22, 2012
9,666
Unusual configuration for an audio power amplifier. I've seen this in other Peavey amps. The output of the amplifier is grounded, but the power transformer centertap is not. I sorta see how this works, but can't figure out why someone thinks this is an improvement over the traditional connectiions (either sonically, electrically, manufacturing-ly, etc.).

ak
View attachment 220258
Looks like the speaker is connected between the CT of the transformer and the output of the power transistors.
 
Last edited:

Ian0

Joined Aug 7, 2020
750
The overall circuit is not inverting. The opamp is inverting locally, but so are the output transistors:

U2B - input goes down.
U2B output goes up.
U4 emitter tries to go up, but is bolted to GND, so ...
U4 collector goes down.
Transformer upper winding goes down.
Transformer centertap goes down, and speaker voltage goes down.
U2B + input goes down.

ak
. In a conventional non-inverting audio amplifier, the voltage-amplifier stage Is driven from the transistor connected to the non-inverting input. In this case, it’s the transistor connected to the inverting input, because the output stage reverses the phase. so there‘s nowhere really convenient to put the dominant-pole capacitor, as it relies on a inverting voltage-amplifier stage.
It’s a while since I tried it, perhaps it just needed more patience and more gain/phase analysis. I wonder if it would respond to a technique like NDFL.
 
Last edited:

Ian0

Joined Aug 7, 2020
750
R60 and C39 form an 87 kHz corner. Plus, you're got C28 and C29.

ak
C26 is quite important as well, but its effect will be changed by the source impedance of whatever is connected to the input.
hope it‘s not a long cable with a lot of capacitance.
Because it’s non-inverting, C39 can‘t get the gain below unity, so won’t in itself stop it oscillating.
Perhaps I’ll have another look at it.
I think this design may be a guitar amp:
33Hz roll off on the input, and R59 Cr24/Cr25 which look like a soft clipping circuit
Either that, or this design has problems recovering from clipping.
 

Wolframore

Joined Jan 21, 2019
1,971
Opamp is a buffer.

the strange circled ground is the return for the signal. It could have been changed to any other symbol, its not really a ground. It’s confusing because we don’t normally see this used for the signal path.

For me this is just a standard output stage of a class AB with some interesting notation. If you look at the center tapped transformer ground (0v) it is the reference for the NI input to the opamp and the ground point for the speaker,

I don’t care for the +/- 15V being set with diodes.

This is for a Peavy Backstage 35W practice amp for guitars mainly but could be used for keyboards. Not a bad sounding amp. Had “tube-ish“ saturation built in. As with many of the amps they are susceptible to noise and RF, especially at high gain.
 

Ian0

Joined Aug 7, 2020
750
Opamp is a buffer.
Not so. The amplifier input is connected to its INVERTING input. There is no local feedback resistor from its output to its input. Not a buffer.

For me this is just a standard output stage of a class AB with some interesting notation. If you look at the center tapped transformer ground (0v) it is the reference for the NI input to the opamp and the ground point for the speaker,
I
The output transistors are driven from a +/-15V supply, if it were a conventional amplifier the output would clip at 13V.
The transformer centre tap is connected back through the feedback network R60/C39 and R61/C40 to the non-Inverting input of the op-amp. R61 is connected to ground. That sets the gain at 7.9. The amplifier input also is also referred to that same ground by R42.
 
Last edited:

Ian0

Joined Aug 7, 2020
750
Integrator, because it is the dominate pole in the amplifier. It creates the 6dB/octave roll-off that keeps it stable.
The feedback that sets the gain comes from R60/C39 and R61/C40.
 

Jony130

Joined Feb 17, 2009
5,186
Well, nothing new here. Changing the GND "force" the output stage to now work as common emitter amplifiers. And the power supplies are now floating.
The advantage of this topology is that now we can drive the output stage from the circuit that has the lower supply voltage. Thus, now we can drive the output stage directly using opamp powered only from ± 15V.

1..PNG

http://bee.mif.pg.gda.pl/ciasteczkowypotwor/#SM_scena/QSC/QSC_USA_Series_(370,_850,_1300)_Service_Manual.pdf

https://books.google.pl/books?id=-5UPyE6dcWgC&pg=PA91&lpg=PA91&dq=Transformerless+push-pull+(transistor+OTL)&source=bl&ots=wuIlvnAKh-&sig=s7nxQcmloFpnPsQyrbay4Eju6dc&hl=pl&ei=dbpUS6HhENTRjAez-NDUCg&sa=X&oi=book_result&ct=result#v=onepage&q=Transformerless push-pull (transistor OTL)&f=false
 

Attachments

Interestingly, the topology in Jony130's attached file, 88_1235904466..jpg , was used for many years by QSC in their commercial amplifiers. The key advantage was that the collectors of the output transistors were all at "ground" potential, and could therefore be used without mica (or other) insultation on a common heatsink - which was also grounded. Substantially better heat conduction without the insulators...
 
Top