This is a fascinating thread. The "Cherry" compensation has been known for a long time. Here is an example from the "RCA Solid State Power Circuits Designer's Handbook" (1971).
Ed
Ed
This RCA note is from a period when stability of transistorised amps was little understood and designers used old fashioned methods which they didn't understand. I don't think you can claim this is an example of Cherry. It has little, if any, attention to stability apart from two dubious zeros which are probably accidentally in 2 nested loops.
I built a couple of the amps in this note and had serious problems with stability. The Motorola notes of the same period where much better and closer to 'modern' practice. The Motorola output devices were also better.
I built a couple of the amps in this note and had serious problems with stability. The Motorola notes of the same period where much better and closer to 'modern' practice. The Motorola output devices were also better.
I did not read Cherry's articles, and so I cannot comment on what constitutes Cherry compensation.
In the RCA class B amplifier, the 820pF capacitor and 18K resistor form the dominant pole at 11Khz. The 10pF capacitor adds a phase lead at 160KHz. This looks like standard one-pole compensation except that the integrator includes the output transistors.
The big downside of the RCA class B design is that it has horrific crossover distortion!
The RCA book also describes a better class AB design.
Here too the integrator (39pF capacitor) includes the output stage. The big mistake is that it also includes the 10uH inductor! Otherwise, it is standard one-pole compensation.
I agree with your sentiment that solid-state amplifiers were not very good in the 1970s.
Ed
In the RCA class B amplifier, the 820pF capacitor and 18K resistor form the dominant pole at 11Khz. The 10pF capacitor adds a phase lead at 160KHz. This looks like standard one-pole compensation except that the integrator includes the output transistors.
The big downside of the RCA class B design is that it has horrific crossover distortion!
The RCA book also describes a better class AB design.
Here too the integrator (39pF capacitor) includes the output stage. The big mistake is that it also includes the 10uH inductor! Otherwise, it is standard one-pole compensation.
I agree with your sentiment that solid-state amplifiers were not very good in the 1970s.
Ed
I tried to include a capacitor from the output, alike Cherry, in a TPC. https://www.diyaudio.com/community/threads/oitpc-output-inclusive-tpc-not-tmc.317335/
Damir
Damir
@dadod - Your frequency compensation method is nicely documented but defies hand analysis.
I can see why the RCA amps included the output stage in the integrator - doing so enables extremely high slew rate at the VAS output because the VAS is allowed to "chatter" within the crossover notch. This is the opposite of what one would consider "stable".
Outside of the crossover notch, the slew rate of the class B amp is only 1V/uS. The class AB amp does better at 26V/uS.
In my amps, I use one-pole compensation because it can be analyzed by hand.
Ed
I can see why the RCA amps included the output stage in the integrator - doing so enables extremely high slew rate at the VAS output because the VAS is allowed to "chatter" within the crossover notch. This is the opposite of what one would consider "stable".
Outside of the crossover notch, the slew rate of the class B amp is only 1V/uS. The class AB amp does better at 26V/uS.
In my amps, I use one-pole compensation because it can be analyzed by hand.
Ed