Cmoy volume control

[PRR]

> the pot itself, doesn´t affect at all at the frequency resnponse?

You don't trust my counting-on-fingers? I don't. Let's see what the idiot computer says.

I think I have set-up your circuit accurately, except I "know" there is some stray capacitance and I assumed 100pFd at the chip input. (It is probably less.)

With the 10K pot-full-up, response "extends to infinity". This is only "true" if V1 is a ZERO impedance source. I should have allowed a few ohms, but even 100 ohms would hardly show a droop on this scale.

With the 10K pot at -4dB, top-response is down 3dB at 706K. The pot makes a difference.

With the 10K pot at -8dB, top-response is down 3dB at 678K.

The idiot computer forgot to run -6dB setting, the worst-case. I figure -3dB at 653KHz, not too different from the -4dB and -8dB results the computer plotted. -3dB at 650KHz is -1dB near 325KHz and -0.5dB near 170KHz, insignificant loss even farrr above the audio range.

BTW, pot at -6dB is also the most-extended bass. But in this case the difference is less than 3% shift of -3dB frequency. I sure can't see it on the full-size graph.

With the 10K pot at -20dB, top-response is down 3dB at 1,780K. In principle, when pot is ALL the way down, "infinite loss", frequency response is again "infinite"... but who can prove it?
__________________________

But if we keep the 100pFd, and change the pot to 100K, all those frequency top-limits move down by a factor of 10. We may not care 650KHz or 1,700KHz. We might care about -3dB at 65KHz, because that is -1dB at 33KHz and -0.5dB near 17KHz. Guitar-amps with 1Meg pots can have top-loss that no hi-fi guy would accept.

 

[lagoausente]

  Thanks for your analysis, PRR.  I understand your explanation, but what´s the conclusion?  I´m hearing 0.5db?  Or I must suppose that frequency loose is not posible because of the pot, and I must check other things?

 

[sahib]

The conclusion is that we have no idea what the problem actually is. And based on your description of the problem "less clear" all we can do is to "throw darts" around it as the man said.


Grab the soldering iron and try the following;


1. PRR's way; reduce the gain so that the offending spot on the volume pot is shifted somewehere less offensive. This is the easiest. If it works forget about the other three.

2. bcarso's way; reduce the impedances on the input to minimise the loading. Replace the pot with 5K.

3. My way; change the input configuration.

If all above fail to satisfy your ears then consider using a stepped attenuator. You can configure the law in a way that you eliminate the offending spot. Although you will be shifting the volume level too.

 

[lagoausente]

Undertand.  I´ll try all that this weekend, and will come with results.

When you refer to step attenuator, would be a replace in the pot position?  Or for example a gain control with a variable R3?

 

[mshilarious]

This won't have much to do with your high-frequency loss, but why use OPA2132 as a can driver?  With an 18V supply?  Let's say 9VRMS into 32 ohm cans, that's 280mA, which is a) way more than that opamp can supply, and b) ridiculously loud if it could.  So I presume you are using high impedance headphones?  Otherwise, check out something like TLV4112, which I use in an otherwise similar circuit to good (loud) effect, with a lot less wasted power.  Rail-to-rail, figure 2VRMS into 32 ohms is a bit more than 60mA or 120mW.  Loud.  It all depends on the cans you're using, but most are low impedance these days.

OK, the high frequency loss.  You're going to have to try to measure it and see what it is.  Loop the output back to your computer and analyze, see what you get.  Play with the pot, try it with the amp unloaded and with a dummy load.  That will help determine if there really is a loss, and what might be causing it.  It would also be interesting to see the difference in distortion at high input/low pot settings vs. low input/high pot, as PRR has conjectured re: distortion of the signal source.  Heck, just loop the signal source straight back to itself and measure that first.

 

[Silvas]

The conclusion is that we have no idea what the problem actually is. And based on your description of the problem "less clear" all we can do is to "throw darts" around it as the man said.


Grab the soldering iron and try the following;


1. PRR's way; reduce the gain so that the offending spot on the volume pot is shifted somewehere less offensive. This is the easiest. If it works forget about the other three.

2. bcarso's way; reduce the impedances on the input to minimise the loading. Replace the pot with 5K.

3. My way; change the input configuration.

If all above fail to satisfy your ears then consider using a stepped attenuator. You can configure the law in a way that you eliminate the offending spot. Although you will be shifting the volume level too.

I took PRR´s way....

Used stock values for everything except the 10k, which i changed for a 2k. 10k was waaaaay too much gain. 2k is more than adequate for the output of my PC soundcard (old SB LIVE with KX drivers) but i think a 1k will work nicely. The headphones i use my CMOY with, are fostex T20, not the most efficient phones on the world, but the amp can drive them loud. Ah, i bypassed the carbon film 10R resistors, and the tracking now is cool. Done in my own layout. I love how this thing sounds. The .1´s are vishay metaliized polys and the opamp is opa2277

 

[lagoausente]

  Solved!,  tried PRR´s solution.  Replaced 10k resistor for 1.3k one. Now it´s DA output from the computer is equilibrated. Sounds well and pot position doesn´t affect the sound. Thanks a lot all for your efforts.

  I had OPA2132 originally. tried NE5532 and LT1358.  I found OPA2132 too much colored and fat sound. Ne5532 sound more neutral and defined, but LT1358 was much better of the three, with high definition at high frequencies, easily to discern instruments and feeling of depth and space that the other two didn´t have.
  I don´t know if the specs of the NE5532 and LT1358 are adequated for supply headphones, I asume any op-amp could be used, I´m wrong with this?
  I use DA24QS  DA converter from Uwe Beis (DIY kit), what I like a lot. Sounds very clear, although needs a clean power supply if don´t want to hear hmmm.
  I have realized that on that DA he uses LM4562 opamp,  do you think it´s a opamp to try even on CMoy or in Supergreen Pre?  What any other opamps do you recomend to get a clear and defined and neutral sound?

 

[ej_whyte]

Dead-thread revival, saves starting a new one though.

My thinking is that, I get rid of R2 altogether, by pass C2 and connect the input of the opamp directly to the wiper of the potentiometer. I increase the pot value to say 22K. Now through the low side and the wiper of the  potentiometer the opamp gets its DC path to the ground. Although that varies, it does not matter very much. One thing important though, I am betting on the opamp being standard consumer grade opamp. I increase the value of C2  and move it to the input and the opamp is protected against DC that may come from the source. As far as the source is concerned, it sees 22K potentiometer as the load and more than sufficiently high but the previous interaction between the potentiometer and R2 is eliminated.

Would this not result in a scratchy volume pot as the opamp is using it as a DC path to ground?

Cheers