diyAudio (
-   The Lounge (
-   -   Poles or feedback 101 (

tvrgeek 15th January 2012 11:24 AM

Poles or feedback 101
Looking for some educational help. The sound quality thread was getting a bit overloaded.

Thinking about total systems requirements for just a little, it occurs to me that:

1: I should not care much about distortion above 5K or so, and care not about distortion above 10K at all as even the second is out of audible range, provided it is not causing any undue strain on the rest of the system like the power supply, heat management, or driver. This would seem to be a handy requirement bound as apps with very low distortion below 10K seem pretty well within our understanding.

2: As our hearing is most sensitive in the 2K to 4K range, we should be very concerned with distortion in the decade below that as those artifacts will show up in our most sensitive range. Again, this seems well within our understanding.

3: Clean power supplies may be more important that is obvious because the harmonics if amplified in the VAS can wind up in our most sensitive range. Two of my amps have more rectifier noise that harmonic distortion right through the 2k to 4K octave. Something I think I can fix.

4: Did I mention poles? OK, looking at the text books and their discussion of adding poles, and about 30 sample schematics, I see a sprinkling of poles in the feed back in several locations all of which seem to be different in frequency and Q. I do not understand this. Logically, I would have thought we would pick some frequency sufficiently outside the required power range and used that for everything from the input infrasonic filter right through. How does this all relate to stability, harmonic distortion within the audible range and slew related issues? Say, pick 80K and provide 100% negative feedback above that. Please help with my quite naive understanding here.

There is a direction to these questions. From a systems view, an amplifier that is required to cover 10 or so decades is not an easy thing. By reducing the power bandwidth in a tri-amp system of each amp, it should be far easier to build better amps with less effort (cost) each. An obvious example is I need at least 6 dB less power for the tweeter amp than the mid just from driver sensitivity differences. The list is a yard long. Everything from cap costs to being able to put the crossover transfer function sprinkled around the system rather than as a separate section. Basically for free if one leverages the feedback poles I need for stability anyway. Think of this as a very high power active crossover problem.

Yea, the market likes to separate amps from drivers, so only a few self powered designs gets to think like this.

DF96 15th January 2012 02:45 PM

Are you asking about stability theory for closed loops, or applying that (already understood) theory to amplifier design?

Don't forget that harmonic distortion is usually accompanied by intermodulation too, as they have a common origin in nonlinearity. You might not hear the second harmonic of 15kHz, but you will hear the 1kHz second-order IM when 15kHz and 16kHz are put through the amp together.

tvrgeek 15th January 2012 02:58 PM

Both. Just learning the amp bits. Ah, IM. That makes perfect sense.
I am trying to assimilate a century of amp knowledge. Lots of non-obvious details. Self assumes you know half of it already.

Would not the IM artifacts be the harmonics of the difference? Not 1K but 2 , 3 etc? I need to WIKI that.

So, this means we still need to strive for absurd bandwidth and the feedback poles are primarily for stability of each given stage?

DF96 15th January 2012 04:14 PM

You can get IM at harmonics of the difference and lots of other frequencies too - higher order IM. For example, 4th order will give (among others) 2x16-2x15=2kHz. To understand how this works you need to revise your trigonometry, as it is all about multiplying sines and cosines together.

In a typical amp most of the poles are not in the feedback but the forward path. Typically, each stage contributes (unavoidably) at least one pole. However, once you close the loop they are all in the loop together. I'm not aware of any need for absurd bandwidth, either open or closed loop.

tvrgeek 15th January 2012 07:17 PM

Let's see, Oscar Has A Heap of Apples. Darn, where is the free version of PSpice....

Looking at my DH 120 as that is handy to study, I see a hiogh pass, and a low pass in the feedback (C4, C7) C5, C6 and D14 all of which I do not understand the function of as part of the input amp. And of course the mystry to me is R10 which consumes 1.8W all the time and will toast the board.

DF96 15th January 2012 07:50 PM

R10 is involved with bias sources for CCS, but may also play a role in diode clamps which prevent BJT collector saturation. D14 may be one of these clamps. The DH120 is quite complex, so you may better starting with a simpler circuit such as Self's 'blameless'.

tvrgeek 15th January 2012 08:59 PM

Yup, solid advice there. I am looking at that too. I figure it is about the "classic" of modern thought. A place where you can say, OK you think you can do better, how.

On the flip side, its not a bad idea to understand what some of the more clever gentlemen were doing. He must have had a good reason for using a dollar resistor or Hafler would not have allowed it being the master of few components himself. I am assuming Erno designed this one? He used a completely different design in the 200 and the AA series 60, DC100, and servo 100.

"When in doubt, plagiarize.
Let no ones work evade your eyes.
That's why God gave you eyes."
Tom Lehrer

tvrgeek 16th January 2012 12:37 PM

Bouncing between Cordell, Self, and a few WIKI pages is starting to make sense. Lots of details that as Self says, were left out as he expects you to know them already.

DF96 16th January 2012 12:46 PM

I always find when learning new stuff that a certain amount of time is needed for it to sink in, and join on to what I already know. Working harder, reading more etc., does not always shorten this time. Just keep at it, and you will get there.

tvrgeek 16th January 2012 01:51 PM

Terribly impatient. Easily distracted. Access to the kinds of experience this forum provides is almost addictive.

What is actually hardest is backing up to concentrate on the basics when you have some knowledge of them. It is too easy to skip over lightly as you want to get to the parts that are new. Bad idea, but that's how I seem to work. I also have a tendency to pick out a detail to understand that is not the focus of the current lesson. I have been driving professors crazy for years.

All times are GMT. The time now is 11:15 AM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 18.75%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio