I have just finished reading Linear Audio Volume 11, and once again Jan Didden is to be congratulated on a generally first-class job.
But- as you might imagine, I make an exception for the Michael Kiwanuka article, which contains a lot of criticism aimed at me.
The amplifier circuit Kiwanka is complaining about is Fig 15.1 in Audio Power Amplifier Design. (6th edn) A copy is attached. The biasing system controls TR1 input-pair tail-current, with the bias voltage passed to VAS current-source TR5, crucially via resistor R23. This looks like a sound piece of penny-pinching but actually has a unexpected limitation in positive slew-rate, which I duly explain and cure. The same scheme was used in the original EW article in Sept 1994. I think all my amplifier designs, for the last 10 years at least, have used separate biasing for the two current-sources to avoid this issue.
Kiwanka's Figure 1 has a biasing system that controls the current through the VAS source TR5, and no bias sharing, which is of course a completely different situation. Whether this change was deliberate or accidental I don't know, but the rest of the article is naturally irrelevant to anything I have written, so you needn't try to make sense of it.
If you are wondering about all this 'Thompson' business, the Russell & Solomon paper just makes the bare statement that a current-mirror in the input-pair stage for differential-to-single-ended conversion was first used in an opamp by J E Thompson at Motorola in 1966- so far as the authors are aware. No reference is given. I don't know who first applied it to a power amplifier, and if anyone can tell me I would be most interested. However that seems to me no reason for calling the configuration a Thompson. Many people are known to have contributed, and you might as well call it a Blumlein (input-pair) or a Miller. (compensation) My Blameless name refers to later developments such as heavy local feedback in the input pair.
[IMGDEAD]http://www.douglas-self.com/system/APAD6-Fig15-1.gif[/IMGDEAD]
But- as you might imagine, I make an exception for the Michael Kiwanuka article, which contains a lot of criticism aimed at me.
The amplifier circuit Kiwanka is complaining about is Fig 15.1 in Audio Power Amplifier Design. (6th edn) A copy is attached. The biasing system controls TR1 input-pair tail-current, with the bias voltage passed to VAS current-source TR5, crucially via resistor R23. This looks like a sound piece of penny-pinching but actually has a unexpected limitation in positive slew-rate, which I duly explain and cure. The same scheme was used in the original EW article in Sept 1994. I think all my amplifier designs, for the last 10 years at least, have used separate biasing for the two current-sources to avoid this issue.
Kiwanka's Figure 1 has a biasing system that controls the current through the VAS source TR5, and no bias sharing, which is of course a completely different situation. Whether this change was deliberate or accidental I don't know, but the rest of the article is naturally irrelevant to anything I have written, so you needn't try to make sense of it.
If you are wondering about all this 'Thompson' business, the Russell & Solomon paper just makes the bare statement that a current-mirror in the input-pair stage for differential-to-single-ended conversion was first used in an opamp by J E Thompson at Motorola in 1966- so far as the authors are aware. No reference is given. I don't know who first applied it to a power amplifier, and if anyone can tell me I would be most interested. However that seems to me no reason for calling the configuration a Thompson. Many people are known to have contributed, and you might as well call it a Blumlein (input-pair) or a Miller. (compensation) My Blameless name refers to later developments such as heavy local feedback in the input pair.
[IMGDEAD]http://www.douglas-self.com/system/APAD6-Fig15-1.gif[/IMGDEAD]