On Mon, 19 May 2008 12:42:39 -0500, flipper <flipper@[EMAIL PROTECTED]
> wrote:
......snip!.......
>
>I don't know what you mean by a "linear disturbance." If it's not
>following the signal then, 'disturbance' being 'linear' or not, NFB
>will work toward bringing it back in line. Otherwise, "linear
>disturbance," it seems to me, would be gain (if it were 'perfect')
>
Careful..... watch out for the terminology I use, "loop gains", open
loop gains, closed loop gains.
You can consider everything in the frequency domain, distortion ends
up being added harmonics (with specified phase and frequency). They
are signals that weren't in the original. They will be reduced by
factors depending on the loop gains.
.....snip!......
>> Now here's MY problem..... it's a mathematical one. It's my belief
>>that square root of sums of squares applies to uncorreleated signals
>>like white or thermal noise. Distortion is correlated (it's a
>>ploynomial base), and I don't think that the non-linear sqr root of
>>sums of squares (RMS) applies here.
>
>Correlated to what? What 'correlates' the distortion of a 12AX7 with a
>6BQ5?
>
They are very highly correlated, simply the amplitude of the
harmonics are dependent on the transfer function of the amplifying
device.
>Following stages certainly amplify the distortion of prior stages but
>they also add their own.
>
The harmonics will be at the same frequencies, but different
amplitudes. Phases may be different from the other circuit parameters.
.....snip!.....
>
>It will. That's not in question. It's that local feedback reduces (and
>PFB increases) the distortion of that stage which you seem to be
>disputing. That and how they add.
>
Local or global feedback will reduce the distortion, I don't
question that. My concern is that in juggling gains and local
feedback, you may be cutting global loop gain, and thus not reducing
distortion as much as you could.
In an op-amp, the designers go to considerable trouble to give as
large an open-loop gain as possible. They don't generally put in
local feedback (other than introducing a dominant pole for stability).
The point is to get the largest "loop gain" (openloop gain divided by
closed loop gain). By doing this, ALL the perturbations, noise, drift
are reduced by the loopgain.
I'll try an example, A and B. Overall gain is to be 1.
A has two gain blocks, each has a gain of 100. that makes open loop
gain=10,000, and loop gain=10,000. The 1st gain block has 10%
distortion. Overall distortion will be 1/1000 %.
B has two gain blocks each has a gain of 100, BUT with local
feedback, the first block has a real gain of 10, and the second one
has a real gain of 1. the 1st gain block has an open loop distortion
(as in example A) of 10%. with local feedback, the 1st gain block will
now have 1% distortion. Using the global feedback loop, its loop gain
will reduce the main outputs distortion to 1/10%. This amplifier is
substantially WORSE than 1st one because you aren't using the excess
gain of the second stage in the global "loop gain".
>Can we agree that you can reduce a tube's distortion with local
>feedback, as in it's Rk? Or any number of other means? (UL taps,
>plate-grid feedback, and so on)
>
>Well, if that's the case then if you reduce the distortion on each and
>every stage independently there's no need for 'global feedback'. At
>least from the tube distortion perspective (might want to for output
>impedance, damping factor, and the fact distortion comes from more
>places than just the tubes, like the OPT).
>
>Conversely, if you can make each and every one better you can also
>make them worse and both affect the overall distortion absent global
>feedback.
>
>
>> This is a tricky question, and it gets down to some hairy
>>theoretical issues..... do you know for sure that non-linearity
>>(distortion) is in fact added up in the sqrt or sums of squares? You
>>can decompose nonlinearity into a fourier series, and treat the
>>distortion as added sine waves, and they aren't random!
>
>Maybe not in ONE tube but there's nothing that makes tube A's good, or
>lousy, characteristics 'correlate' to tube B's behavior.
>
>You'd have a better chance of arguing there's correlation between the
>2'nd, 3'd, 4'th, and so on, harmonics in a single device (since
>they're multiples of the 'same signal' and, supposedly, 'not random')
>but those add as the square root of the sum of squares too. At least,
>that's what the text books tell me.
>
In this case the root of the sums of squares is just a convenient
way to show all the harmonics as one single RMS measurement on a meter
dial. Using a spectrum analyzer (or any other frequency domain device
like your ears), the harmonics should be simply vectorially summed.
arrrgggghhh! enough of the math.... if you take your circuit, and
run all the subsections open loop, look at the output of a distortion
analyzer, or spectrum analyzer, then compare (in %), the same amp with
local feedback loops, how do they compare?
My guess is that if you keep the global open loop gain the same (ie,
increase the gain of one stage at the expense of another stage), there
should be no difference. If you reduce the global open loop gain by
using local feedback, this circuit will probably show poorer
performance.
I'm curious.... have you ever tried to measure the perormance to
see what effect juggling the local feedbacks has on the output with a
global feedback loop?
I think it's the idea of positive feedback that I find so
disturbing, especially with devices that are supposed to be linear. I
think I'm too traditional to accept this idea without argument!
-Paul
-Paul
|
