Recently built a subwoofer amp. Placed a 33p capacitor over the 39k feedback res. Now at a certain amplitude, the amp starts to oscillate. Guess there is just one-to-many phase shifts happening.
I want to roll off the amp a bit since it is a sub amp. I saw somewhere, that a resistor in series with this cap before paralleling with feedback resistor, is used to mess up the Q. Any ideas on how to size this series resistor????
I want to roll off the amp a bit since it is a sub amp. I saw somewhere, that a resistor in series with this cap before paralleling with feedback resistor, is used to mess up the Q. Any ideas on how to size this series resistor????
Last edited:
You need to adjust your amp's compensation: it is not unity gain stable.
If it uses dominant pole compensation just increase a bit the value of the cap there.
If it uses dominant pole compensation just increase a bit the value of the cap there.
yes it is a very bad idea - increases feedback factor - likely the amp is not "unity gain stable " - as are most cheap op amps but few power amps
even if it did "work" it wouldn't give the right transfer function - with the usual non-inverting amp there would be a Zero in the response that would pass high frequencies with a constant attenuation = to your amp's gain (ie it would be "unity gain" for high frequencies instead of rolling them off)
roll off the signal input with a active filter circuit - XO needs to be coordinated with the rest of your system anyway - active analog or digital XO is the way most systems do this today
even if it did "work" it wouldn't give the right transfer function - with the usual non-inverting amp there would be a Zero in the response that would pass high frequencies with a constant attenuation = to your amp's gain (ie it would be "unity gain" for high frequencies instead of rolling them off)
roll off the signal input with a active filter circuit - XO needs to be coordinated with the rest of your system anyway - active analog or digital XO is the way most systems do this today
Last edited:
The task might be more complex, like compensation of input capacitance (between +input and virtual ground), or HF phase shifts. Then one would usually find a small FB cap, like 3pF, might help, and unity gain stability is not required.
In a typical amp, dominant pole compensation is done by the Miller capacitor on the VAS stage. The aim is to ensure that the loop gain does not enclose the critical point at (1,0) in the complex plane.
learning involves paying attention to answers, comments, taking in new knowledge, reformulating your question
the more info about the system, the amp, what your "need" is the better the advice can be
don't just repeat what you've previously concieved of until someone on the web encourages you to do it in the face of other good advice that it has problems
the more info about the system, the amp, what your "need" is the better the advice can be
don't just repeat what you've previously concieved of until someone on the web encourages you to do it in the face of other good advice that it has problems
Hi,
Its easy to roll-off the amplifiers response at the input without affecting stability.
This will have the same response as modifying the feedback loop.
To modify the feedback loop without stability issues you increase the gain at
low frequencies, allowing it to return to normal higher up, this will maintain
the gain margin required for stability. If you reduce gain the compensation
needs to be modified (increased) for the lower loop gain.
rgds, sreten.
Its easy to roll-off the amplifiers response at the input without affecting stability.
This will have the same response as modifying the feedback loop.
To modify the feedback loop without stability issues you increase the gain at
low frequencies, allowing it to return to normal higher up, this will maintain
the gain margin required for stability. If you reduce gain the compensation
needs to be modified (increased) for the lower loop gain.
rgds, sreten.
If your goal is to rolloff the frequency responce of the amp and limit it to sub-woofer frequencies, this should be done with a filter before the input, either passive or active. By adjusting the Miller cap value or the value of the feedback resistor (changing the feedback factor), you change the frequecy at which voltage gain = 1, and the phase shift of the output compared to the input is 180 degrees and so positive feedback with gain takes place. If the compensation does not match correctly the amp will be either under damped (unstable
) or over damped. Neither situation is optimal. Placing a resistor in series with the Miller cap creates a zero in the transfer which may help with phase margin about the unity gain frequency. Look up pole/zero compensation. Generally you can determine where the gain crossover frequency is and place the zero to be ~10 times higher. fz= 1/(2pi*R*C)
Using the compensation cap to set the audio signal bandwidth limit is barking up the wrong tree.
) or over damped. Neither situation is optimal. Placing a resistor in series with the Miller cap creates a zero in the transfer which may help with phase margin about the unity gain frequency. Look up pole/zero compensation. Generally you can determine where the gain crossover frequency is and place the zero to be ~10 times higher. fz= 1/(2pi*R*C) Using the compensation cap to set the audio signal bandwidth limit is barking up the wrong tree.
If your goal is to rolloff the frequency responce of the amp and limit it to sub-woofer frequencies, this should be done with a filter before the input, either passive or active. By adjusting the Miller cap value or the value of the feedback resistor (changing the feedback factor), you change the frequecy at which voltage gain = 1, and the phase shift of the output compared to the input is 180 degrees and so positive feedback with gain takes place. If the compensation does not match correctly the amp will be either under damped (unstable
Using the compensation cap to set the audio signal bandwidth limit is barking up the wrong tree.
Hi,
Not exactly accurate, your mixing up unity gain with fixed gain, and loop gain.
Miller degeneration is this case will have no useful effect. Its simple, the more
feedback you apply to reduce gain the more you need to nobble the open loop
gain with compensation to ensure stability.
It is also very simple regarding the feedback loop. Increasing gain at low
frequencies has no stability issues. With a feedback loop the compensation
has no bearing whatsoever on the effective audio response. To create a
bass boost and effectively a stepped roll-off put a parallel RC in series
with the higher value of the two feedback resistors.
e.g. say the resistors are 20K and 1K for a gain of x20. Adding a 47K with
a cap in parallel in series with the 20K will give a stepped bass response
with no stability issues, as the high frequency gain is the same.
For the above low gain will be x67 rolling off depending on the
cap to x20. The above will give around 10dB low bass boost.
(And is far more useful for compensating driver roll-offs.)
Its far simpler to simply roll-off the treble at the input.
(with a series resistor and parallel capacitor).
rgds, sreten.
Last edited:
- Status
- Not open for further replies.