Hello and good evening;
I am currently building plans to make my cmoy headphone amplifier better. As such, I am looking to change the operational amplifiers in it, and was wondering if a lowpass filter would be a good thing to have around 20KHz.

If so, using a capacitor and a resistor, how would I wire them to make one at this approx. frequency, and what values should I choose?

Should I go with as large a capacitor as possible?
Is there any type in particular (ceramic, electrolytic, etc..)
Where should I put it? Before amplification or after?

As a second point, I have a highpass filter to block out DC on the audio. I hear any resistors placed before the amplification stage should be matched with some other resistor to prevent some type of freaking out on the Op-amps parts... Does any one know about this, or can someone provide extra information?

Thanks in advance,
Tristan.

The amplifier works very well as-is. Unless you are just looking to amuse yourself, there is no point to adding a low pass filter.

What if I'm changing the amplifiers? Shouldn't I prevent ultrasonics from being amplified for some reason?

The CMOY uses a classic op-amp design that is stable for a very wide range of frequencies. At very high frequencies, the gain of the op-amp itself will become very low, eventually dropping below the limits imposed by the feedback loop. At that point high frequencies will be self-limiting.

If you still wish to limit very high frequencies, this would not hurt anything. I would use a simple RC combination at the input. No need for a large cap - in fact, that would suck at very high frequencies. Use a small ceramic cap to ground and a series resistor. For such a single-pole design, set the -3dB frequency well above 20kHz - I would shoot for 100kHz. In that case, you may use a 1K resistor and a 1.5pF cap.

Thanks for the precise answer, BradPDX.

That should be 1.5 nF , for a -3dB at 100KHz

If the amp gain drops off at high frequencies (and is well below unity at 1MHz) I wouldn't worry about it, unless you find that there is a load of noise above 20kHz which is introducing IMD in the amp or speakers. While the phase effects of a filter at 100kHz are highly unlikely to be audible, an extra 1Khz on input might add some extra noise, if your gain is high.

If the amp still has over unity gain above 200kHz (as many modern opamps, NE5532 for example, can) then a filter is probably a good idea. If this is the case, however, it might be better to change the amp design than add a simple filter on the input side.

Can I place the filter inside the feedback loop? (and can something like a 30kHz be okay?)

In many designs putting a simple filter in the feedback loop will reduce the phase margin causing the amplifier to oscillate. Why would you want it in the loop?

Also, if I was using a simple RC filter, I would set it higher than 30k. BradPDX's recommendation of 100kHz sounds sensible to me. Do you know what the gain-bandwidth graph of this amp looks like? If you could find me a schematic and some measurements, I could advise you much better.

I have three choices (basically)..

The OPA227 (which I'll use as a buffer on the output stage, I think, maybe as a power supply buffer as well, somehow)
The OPA228 or the OPA134 for the main amplification. Any ideas? (thanks in advance)

A power supply buffer? I would like to see this design, it sounds very - unconventional. I would choose OPA134 for the gain stage, but most of their opamps are very good indeed.

I couldn't find a datasheet for OPA227, are you sure you got the number right?

http://focus.ti.com/docs/prod/folders/print/opa227.html

The design I'm sort of building myself.. I figured that placing a buffer in the power supply would allow me to split the voltage more precisely.. Isn't that correct?

(I haven't made schematics yet - it's still in a partial thought-project-state (very vague indeed))