[Laika]

I'm building myself another Cmoy (for non portable use) that will only be powered by a wallwart (12v/.5w/reg) and will be in an enclosure where size isn't really a issue... My question is how big a value power caps can/should i go? My portable has 470's and I've seen a few ebay Cmoys with 1000's.... I'm guessing there is a cutoff where extra value isn't really useful?

 

[DKJones96]

If I'm not mistaken 470 is the point of diminishing returns. My only remaining cmoy actually uses 100uF and I've tried up to 3300 with no difference in sound over the 100s.

 

[Logistics]

But if size is still no issue then just make sure you use some sort of very high-quality cap, such as a Panasonic FM rather than the basic NHG's mentioned in the tutorial. What did you use last time?

 

[NelsonVandal]

How large caps you should use is depending on the impedance of your phones. The lower the impedance, the higher the cap value. It has to do with crosstalk in the lower (and mid) registers.

Capacitors in the audio pathway like the rails to ground caps in CMOY degrade the sound more than active components. Electrolytics are the worst. My suggestion is that you build it with an active ground channel like the PINT or Mini3. You need an extra opamp, and it will be a bit harder to build. If you want to take the easy way out you can use a TLE2426 as railsplitter (it's not as good as a good opamp, but I find it better than electrolytic capacitors).

 

[Laika]

Quote:

Originally Posted by Logistics /img/forum/go_quote.gif But if size is still no issue then just make sure you use some sort of very high-quality cap, such as a Panasonic FM rather than the basic NHG's mentioned in the tutorial. What did you use last time?

In my pocket Cmoy I used Lelon REA elec caps (470uf/16V) in the power supply. For the new build I was considering using the couple of same type but in a higher value and voltage... Unless its a bad idea?

 

[gurusan]

Quote:

Originally Posted by NelsonVandal /img/forum/go_quote.gif How large caps you should use is depending on the impedance of your phones. The lower the impedance, the higher the cap value. It has to do with crosstalk in the lower (and mid) registers. Capacitors in the audio pathway like the rails to ground caps in CMOY degrade the sound more than active components. Electrolytics are the worst. My suggestion is that you build it with an active ground channel like the PINT or Mini3. You need an extra opamp, and it will be a bit harder to build. If you want to take the easy way out you can use a TLE2426 as railsplitter (it's not as good as a good opamp, but I find it better than electrolytic capacitors).

He is talking about power caps.....

 

[NelsonVandal]

Quote:

Originally Posted by gurusan /img/forum/go_quote.gif He is talking about power caps.....

Yes and I'm also talking about power caps. They're very much in the audio path, sinking/sourcing the return currents. That's why 3 channel amps, amps with shunt regulators or any other methods of getting a low impedance active ground generally sound better. It's also much cheaper and smaller if you compare it to "HiFi capacitors".

 

[AudioCats]

agree, power caps should also be considered as "in the signal path".

 

[tomb]

I'm not so sure. With respect, his first response had to do with the impedance of the phones, implying RC filtering with the output caps. In a properly designed PS, the audio quality of power caps will have diminishing returns. This is because "audio-quality" caps are often not nearly as good in ESR and ripple as a good quality power cap. Again, in a properly designed PS, or even in a circuit with a reasonable PSRR (the CMoy is one), the de-coupling caps should be selected for ESR and ripple, not "audio quality."

Use the best cap for the specific purpose for which it's designed.

 

[kvant]

Quote:

Originally Posted by Laika /img/forum/go_quote.gif For the new build I was considering using the couple of same type but in a higher value and voltage... Unless its a bad idea?

Quote:

Originally Posted by NelsonVandal /img/forum/go_quote.gif How large caps you should use is depending on the impedance of your phones. The lower the impedance, the higher the cap value. It has to do with crosstalk in the lower (and mid) registers.

If you are talking about the "classic" CMoy (with virtual-ground circuit being just a resistor divider with resistors bypassed by discussed power supply capacitors), then it is as NelsonVandal says: it depends on the headphones you plan to use with your amp. If I didn't mess up the math, the impedance of the virtual ground node is

|Z|=R/2 1/sqrt ( 1 + ( 2 pi f R C )^2 )

where R is the resistance of the resistors forming the divider, C is the capacitance of bypassing capacitors and f is frequency of the signal. If I use values from the tangent's schematic (R=4.7k and C=220uF), I get |Z|=18 ohm at 20Hz, with C=470uF it would be |Z|=8.5 ohm. With 300 ohm phones this is not a big deal, but with 32 ohm phones it means quite a bit of crosstalk at low frequencies and, perhaps more importantly, also attenuation of these low frequencies.

There are other reasons why this CMoy plan is not quite an optimal solution for low-impedance phones... but if I was building non-portable CMoy following the "classic" schematic, I would put in there 1000uF, maybe even more. As for the "audio quality" - no comment.

 

[NelsonVandal]

Thank you kvant for the EE perspective. I've heard and measured this effect, but I wasn't familiar with the formula. A low impedance point for the return currents is IMO essential for a natural stereo image.

I think one must keep power supply and how the amp handles the return currents apart when discussing the effects of "power supply" caps.

I still think an active ground channel is preferable. It's cheaper and smaller this way. The crosstalk is very much better and the distortion is lower. There's a potential to get the "right blend" by combining two different (op)amps. The drawback is an increased power consumtion.

 

[Laika]

Quote:

Originally Posted by NelsonVandal /img/forum/go_quote.gif How large caps you should use is depending on the impedance of your phones. The lower the impedance, the higher the cap value. It has to do with crosstalk in the lower (and mid) registers. Capacitors in the audio pathway like the rails to ground caps in CMOY degrade the sound more than active components. Electrolytics are the worst. My suggestion is that you build it with an active ground channel like the PINT or Mini3. You need an extra opamp, and it will be a bit harder to build. If you want to take the easy way out you can use a TLE2426 as railsplitter (it's not as good as a good opamp, but I find it better than electrolytic capacitors).

Quote:

Originally Posted by kvant /img/forum/go_quote.gif If you are talking about the "classic" CMoy (with virtual-ground circuit being just a resistor divider with resistors bypassed by discussed power supply capacitors), then it is as NelsonVandal says: it depends on the headphones you plan to use with your amp. If I didn't mess up the math, the impedance of the virtual ground node is |Z|=R/2 1/sqrt ( 1 + ( 2 pi f R C )^2 ) where R is the resistance of the resistors forming the divider, C is the capacitance of bypassing capacitors and f is frequency of the signal. If I use values from the tangent's schematic (R=4.7k and C=220uF), I get |Z|=18 ohm at 20Hz, with C=470uF it would be |Z|=8.5 ohm. With 300 ohm phones this is not a big deal, but with 32 ohm phones it means quite a bit of crosstalk at low frequencies and, perhaps more importantly, also attenuation of these low frequencies. There are other reasons why this CMoy plan is not quite an optimal solution for low-impedance phones... but if I was building non-portable CMoy following the "classic" schematic, I would put in there 1000uF, maybe even more. As for the "audio quality" - no comment.

I'm using SR80i's (32 ohm) headphones ... is there an optimal power cap value for the cmoy with these?

I intend to make the power pcb on my new build separate so I can easily replace it when I get my head around the other power supply options... Using an opamp looks interesting, I've been reading the tangent page on virtual ground circuits and it mentions using a cheap 741 opamp in a push, are LM833/TL072 any better or even suitable?

 

[Laika]

What cmoy power cap value would be recommended for use with SR80i's headphones (32 ohm)?? (standard cmoy power circuit)

I hope this doesn't end up a double post... I tried posting a similar post a few days ago and it said it required mod approval... tis yet to be seen.

 

[00940]

With a non portable cmoy, you're not concerned about power consumption. Why only considering the capacitors of the rail splitter ? Heavily reduce the value of the resistors too. This will greatly improve the accuracy of the rail splitter with low impedance cans.

I'd use something like 1k resistors with 1000uF/16V caps.

 

[chosborne]

I've used a 1000uF -> 50R -> 470uF RC filter with extremely good success when using a wall-wart. Just don't forget to decouple the power pins near the opamp with ~10uF caps too.