Bob Cordell's Power amplifier book

[Bob Cordell]

Originally Posted by abraxalito

Hi Bob

You say 'a feedback loop' and then 'the feedback loop'. Are these the same feedback loop? I ask because I've designed soft clipping circuits before that have what might be called an envelope feedback loop. This is in addition to the normal NFB which is applied to the power amp. So they operate in the same way as a compressor - sense the output envelope and feed that back to an AGC stage to lower the overall gain.

In a former job I used to design active subwoofer amps. One of the big challenges there is to handle overloads gracefully - the sound of the compressor played one of the biggest parts in the overall sound of the device.

With a subwoofer, the power supply is one of the most expensive parts so an engineering optimisation is called for to get the most out of it. Supplies sag when under load, and the drive unit's load varies with frequency so a clipper based on predicting what the power supply is going to do is almost impossible. I played with a feedback system from the output rails - I did this by using a rail-to-rail opamp, driven from a resistively divided version of the output rails. It can be arranged that this opamp clips at 90% of the maximum output voltage of the power amp - its clipping generates an error signal which feeds back to lower the gain at the input. There's absolutely no added distortion then below 90% of output, below the clip level of the modelling opamp.

In all cases in that post I was referring to the amplifier's global feedback loop. The kind of clipper I was referring to as my preferred approach is a soft clipper that precedes the amplifier, so that the amplifier, with its global feedback loop, never clips.

The use of a dynamic soft clip circuit means that the clipping threshold of the soft clip circuit tracks the average available rail voltage so that some degree of dynamic headroom is retained. The Klever Klipper merely uses passive silicon diodes for the clipping action. Those diodes are normally reverse-biased by the clip threshold control voltage.

The Klever Klipper is described in Chapter 17 of my book.

Cheers,
Bob

[spind]

Copyright 2011??? Did I miss a few months somewhere?

Steve.

[Bob Cordell]

Originally Posted by Damon Hill

errata: page 257 "for this reason, trimmer R22" referring to Figure 12.8 on page 256. There is no R22 in the schematic, so I guess this was referring to R12.

(I'm cheating a bit by reading ahead, but I was thinking to apply the error correction concept to a Leach amplifier circuit I'm using as a test bed, which will also be used to try out the On Semi Thermaltrak parts.)

I'm glad to see some introductory coverage of basic circuits and concepts, which was assumed knowledge in Self's and Slone's books. Not all of us are starting out with EE degrees.

Hi Damon,

Thanks for bringing that errata to my attention. You can also email me directly at bob [at] cordellaudio [dot] com.

I'm delighted that you liked the introductory coverage that I included in the book. As you can see, I've aimed to serve a broad spectrum of readers.

Cheers,
Bob

[jan.didden]

Got your book Bob! Wonderfull!

jan didden

[Rikard Nilsson]

I received my copy today and have just had a brief look through it, and I must say I´m deeply impressed! Loads and loads of useful info, good explanations that made me think in different paths, and a written language at just the right level to keep things interesting but never overwhelming. At least it suits me, not speaking English as my native language. To anybody out there not sure about buying the book, all I can say is go ahead and buy it! You won´t be disappointed! This is what I have been looking for the last few years. Thanks Bob!



/Rikard

[JPV]

Originally Posted by janneman

Got your book Bob! Wonderfull!

jan didden

Jan, where did you order it? I did it via amazone uk and i have to wait until december.


To Bob Cordell

In your book, do you analyze the influence of a reactive load on crossover distortion.
I am puzzeled by the fact that crossover distortion is happening when the current goes through zero in the output transitors. Due to the phase of the input impedance of the load, the voltage can be high at that moment and will then have a glich. Could it then happen that again due to the phase of the voltage input, pressure ouput transfer function of the system, the output pressure could be high and experience at that moment a glich which makes it louder than what would be expected from a zero crossing signal.

The same can be asked for the optimum bias current if the load is reactive.

If this is true, custom phase equalization of the system becomes a very important topic at least for diy.

JPV

[Bob Cordell]

Originally Posted by spind

Copyright 2011??? Did I miss a few months somewhere?

Steve.

Wow, you've really got sharp eyes. Good proof reader. I hadn't noticed that. I don't know whether it is a mistake or just the way they date things with copyrights. Go figure.

Cheers,
Bob

[Bob Cordell]

Originally Posted by Rikard Nilsson

I received my copy today and have just had a brief look through it, and I must say I´m deeply impressed! Loads and loads of useful info, good explanations that made me think in different paths, and a written language at just the right level to keep things interesting but never overwhelming. At least it suits me, not speaking English as my native language. To anybody out there not sure about buying the book, all I can say is go ahead and buy it! You won´t be disappointed! This is what I have been looking for the last few years. Thanks Bob!



/Rikard

Thanks, Rikard. These are very very kind words. I'm really glad you like the book and hope that you will enjoy it and that it will inspire you to good ideas.

Cheers,
Bob

[jan.didden]

Originally Posted by JPV

Jan, where did you order it? I did it via amazone uk and i have to wait until december.
[snip]

I got it from amazon US.

jan didden

[Bob Cordell]

Originally Posted by JPV

Jan, where did you order it? I did it via amazone uk and i have to wait until december.


To Bob Cordell

In your book, do you analyze the influence of a reactive load on crossover distortion.
I am puzzeled by the fact that crossover distortion is happening when the current goes through zero in the output transitors. Due to the phase of the input impedance of the load, the voltage can be high at that moment and will then have a glich. Could it then happen that again due to the phase of the voltage input, pressure ouput transfer function of the system, the output pressure could be high and experience at that moment a glich which makes it louder than what would be expected from a zero crossing signal.

The same can be asked for the optimum bias current if the load is reactive.

If this is true, custom phase equalization of the system becomes a very important topic at least for diy.

JPV

Hi JPV,

I do not recall saying much about the influence of a reactive load on crossover distortion, however your point is well-taken. Crossover distortion is a function of the output current of the output stage. While the net output current crosses through zero at 0V for a resistive load, this is not the case for a reactive load. So the answer is yes, the crossover glitch could occur at a point in the output signal where the output voltage is moderately high if a reactive load is being fed.

I don't think this observation has any implications for what the optimum quiescent output stage bias should be.

Cheers,
Bob