Miller compensation - pros/cons?

[fdegrove]

Hi,

For a full understanding basic Nyquist/Bode theory needs to be studied as suggested above.

I wouldn't hold my breath if I were you......

Anyway, the stage as shown is sometimes used as the first stage of a MM phono preamplifier where it makes perfect sense to use this kind of compensation.

For one, you can do with high input capacitance to compensate the rise in HF response from the MM cartridge and secondly with a single cap you can also tailor part of your RIAA correction network. Neat.

You'd certainly wouldn't want such high Cin in an amplifier under any normal audio circumstances bar perhaps an electric lead guitar amplifier.
Which is quite likely where Joel has got this from??

Cheers,

 

[Joel]

I wouldn't hold my breath if I were you......

Ah, Franky... still the same old troll I see.

You'd certainly wouldn't want such high Cin in an amplifier under any normal audio circumstances bar perhaps an electric lead guitar amplifier. Which is quite likely where Joel has got this from??

Arggh. May the rioters burn your car.
First, name me one guitar amp input that looks like this. Second, I have listened to the amp, and it sounds wonderful. It is not low-fi in any way.

Joel

 

[Joel]

In that case linearity, such as Miller compensation will improve, is not of consequence.

Huh? Negative feedback above 20kHz has no benefit?

Perhaps you want to revise that statement?

 

[tubetvr]

How's that sound?

Not so bad but I am missing one important point and that is that Miller compensation or any other way of compensating using dominant pole is not optimal, there are other ways that are more efficient.

In Opamps you normally have excess gain so you can waste more of it and use dominant pole which is easy and safe, with other methods you need to know the open loop behavior in detail which can be difficult to measure.

For tube amps where you needed compensation you didn't have the luxury of very high gain, gain was expensive and contributed to waste heat and size of the equipment, therefore it was more to gain by using better compensation methods than dominant pole.

In spite of what I just have written there are high performance opamps where the manufacturer recommends more advanced compensation so the old fact, (that dominant pole is not the optimum method) is still valid and of interest.

It is possible to show that an amplifier in order to be stable with feedback should have an asymptotic slope of about 9dB per octave which is more steep than what you achieve by dominant pole compensation, when the slope is steeper it means that the flat part of the open loop response can be wider.

What this 9dB per octave means in practise is that you need to add phase compensation at the point where the slope is steeper than 12dB per octave, this compensation can be done using more or less complicated networks, these are often described in older books about feedback circuits and servo loops, the whole method was invented by Bode in the 1940-ths. The often used method with a small cap in parallell with the feedback resistor is only one example of this kind of compensation where networks of often much more complicated nature is used.

Regards Hans

 

[jean-paul]

Arggh. May the rioters burn your car.

Joel

Please try to behave like a grown up person. This kind of language is not the language the members nor the moderators like to read. You have been warned.

Jean-Paul, DIYAudio moderator

 

[Johan Potgieter]

Huh? Negative feedback above 20kHz has no benefit?

Perhaps you want to revise that statement?

Hi Joel,

I said LINEARITY, not presence of NFB. I reacted to someone remarking that Miller compensation, compared to a C to common, improves linearity. [The argument is more valid in transistor amplifiers, where a C from signal to common for stability control sometimes gives rise to more distortion than a Miller-C - which falls outside this discussion.] Since we are talking of "the-way-in-which" and not the degree of attenuation, which would be the same for a passive C, all else being equal, the amount of NFB is a common factor.

Frank,

Quite true, although I do not recall having seen that method in practice as part of RIAA equalisation. What I do recall (and have used) is that simply diminishing the input resistor (thus loading the MM cartridge) can give the desired above 2 KHz attenuation, which will also damp tendency to h.f. peaks. Quite vaguely I seem to recall Goldring giving a specific R-value to this end. The value will be peculiar to a particular cartridge though since the inductances vary from type to type.

By the same token I am wondering whether using a capacitive load (whether Miller or passive) may not accentuate (cause) a cartridge coil resonance somewhere? It does bring in an LC! I have read something to that effect but it was lo-o-ong ago, and the grey matter does not improve with age! Perhaps you could comment.

Regards
Johan