Transistor Distortion

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Dominant LPF poles should be applied globally to deliver an accurate transfer function. Compensation caps to insure stability can be local to the op amp. It is unclear which that 47pF is. Caution, too much local compensation can compromise the overall transfer function.

I don't expect any of this is a major concern with this particular circuit.

JR
When I redesigned the API 318a distribution amp, moving the compensation cap from around the output drivers to just around the op-amp eliminated the instability and UHF oscillations that ABC was complaining about. The output devices were so much slower that the oscillation wasn’t even seen at the output, just pinching the headroom…
 
The lag from slower driver stages can cause instability if total phase shift of NF path exceeds 180' before open loop gain falls below unity gain. The remedy is a local compensation cap directly across the op amp. This cap should be small compared to overall global LPF capacitor to not alter the transfer function. This is pretty common in power amp design.


[edit- to be clear the local compensation cap is applying phase lead to the NF coming back into the op amp - input to compensate for the phase lag from the slow output devices, and perhaps stray capacitance to ground. This compensation cap is working against the global feed back cap so optimal size will interact. It only needs to provide enough phase lead to keep the amplifier stable, not alter the HF roll off. /edit]

JR
 
Last edited:
I am a little confused with some of the posts and criticisms in this thread because I remember very well that in my young engineering days the Revox PR99 tape recorders sounded very good. The line out amp in the PR99 uses the same topology criticized here. I still very well remember a fresh recording of The Boss's concert in the late 80s from Berlin connected to some JBL amp and UREI 811 speakers. I guess I was just inexperienced then, and I didn't really realize that I was listening to a MW transmitter. :cool:

1660548033231.png
 
Since Moamps put up that section of schematic (I used to have a PR99 and unfortunately gave it away) there are any number of aspects to that design that will affect the overall performance. The Op amps originally used were not particularly 'fast' (I forget which they were now LM748?) and C11 is I think feedforward compensation. The output transformer isolates the amplifier stage from a possible strongly capacitive cable loading so in itself probably assisting stability of the amplifier stage.
The original design brief used to be a compromise between sounding good (with whatever it might be connected to), have good measurable specifications (to brag about) and be cheap enough to make a profit. There are of course many different ways to achieve these goals.
Playing a 'devils advocate' it would be possible, but not advisable to omit R4 and feed in some HF oscillation from the bias/erase oscillator so that the 'switching' of the output transistors gets blurred. This would then require output filtering ) see also 'class D' power amplifiers. Yes I know that class D is different altogether but having just written this I am reminded of my STC 'amplifier designs' booklet from the early 1970's which shows a design for a power amp (where they had bravely shunned a driver transformer to have a multivibrator (2 transistor) oscillator feeding the output transistors. Their 200 watt design (2 Ohm load) recommeneded the use of 3 phase mains because continuous 12Amps DC from rectifier/capacitor design was deemed impractical. Thus 3, smaller power transformers were proposed.
 
I am a little confused with some of the posts and criticisms in this thread because I remember very well that in my young engineering days the Revox PR99 tape recorders sounded very good. The line out amp in the PR99 uses the same topology criticized here. I still very well remember a fresh recording of The Boss's concert in the late 80s from Berlin connected to some JBL amp and UREI 811 speakers. I guess I was just inexperienced then, and I didn't really realize that I was listening to a MW transmitter. :cool:

View attachment 97235
The 470 ohm resistor shunting between base-emitter of the output transistors provided 10x the current of the 4.7k in the first post.

JR
 
I am a little confused with some of the posts and criticisms in this thread because I remember very well that in my young engineering days the Revox PR99 tape recorders sounded very good. The line out amp in the PR99 uses the same topology criticized here.
I have heard folks say that the monitor outs are cleaner than the line outs on the PR99. But there is a lot of stuff between the monitor outs and the line outs so I can't say xover distorition is the cause.

And the effect of the 470R vs 4K7 between the bases and emitters has been explored extensively in this thread.
 
What is the purpose of C4 in the PR99 schematic? It looks like Current Feedback of the Positive type. Would this not exacerbate the non-linearities of the output transformer (and the non-linearities of the reflected load)?
 
OP replaced 4.7k resistor with 220 ohm and reduced the crossover distortion
The fact is that this topology is not adequately chosen in this case where the load impedance can be very small. Replacing the 4k7 resistor with 220 ohms is not a real cure, because if someone put on 32 ohm headphones, the distortion will appear again. The simplest solution IMO is to replace 072 with 4556 or similar, and to remove all redundant transistors and resistors.
 
The fact is that this topology is not adequately chosen in this case where the load impedance can be very small. Replacing the 4k7 resistor with 220 ohms is not a real cure, because if someone put on 32 ohm headphones, the distortion will appear again. The simplest solution IMO is to replace 072 with 4556 or similar, and to remove all redundant transistors and resistors.
I have designed multiple headphone amps over the decades, it is impossible to be all things to all people but I have come close. That said simple headphone amps like this use a significant build out resistor in series with the output to protect against extremely low Z cans. 68 ohm build out makes your hypothetical 32 ohm can more like 100 ohms. Purists headphone amp designers prefer low output Z. My Peavey headphone amp (HB-1) could happily drive loudspeakers to modest levels.

This is typical "value" headphone amp design... lowering the shunt resistance makes it serviceable as the OP's bench measurements confirmed. No one is claiming that this is audiophile, while it doesn't suck when used reasonably.

JR
 
The fact is that this topology is not adequately chosen in this case where the load impedance can be very small. Replacing the 4k7 resistor with 220 ohms is not a real cure, because if someone put on 32 ohm headphones, the distortion will appear again.
If I understand correctly the specs, the headphone output was designed to drive medium impedance headphones. the quoted figure is 220 ohms. Whether it's the actual oustput Z or the recommanded load, I don't know, but it seems consistent with the common (at the time) use of Beyer DT200 headphones, which standard impedance was 250 ohms. You had to make a special order to have 60 ohms or 600 ohms.
 
Last edited:
Although this document is well-documented and informative (it's not always the case with ESP), I am a tad dubious at the conclusion that driving a transformer with a NIC (or with any similar global NFB arrangement) only achieves minor perceived improvement. I have used different arrangements in the past : global NFB from a diff amp attached to the secondary, a form of NIC and a tertiary winding.
All have provided significant improvement, both in objective and subjective evaluation. I finally settled for the tertiary arrangement, because it was more stable in production.
I suspect Rod Elliott's evaluation (admittedly made by listening to radio) was not very sensible.
 
Last edited:
Back
Top