Capacitor choice

[Pierre]

Hello all.
What's the best choice for DC coupling input capacitors?

SMD tantalum (2.2uF/16v) (polarised)
or
0805 ceramic SMD (2.2uF/16V) (non-polarised)

And for the bootstrap capacitor? At the same value, what is best: a 10uF/16v ceramic or a 10uF/16V tantalum in parallel with a 0.1uF ceramic?

Thanks!

 

[IVX]

i think SMD tantalum (2.2uF/16v) (polarised) or 0805 ceramic SMD (2.2uF/16V) (non-polarised) both isn't so good choice for the input cap. Non-polar aluminium cap 2.2/16 will pretty good and cheap solution (BG or Elna even), or polyester film cap. Ceramic, as bootstrap cap, is the best.

 

[sx881663]

Cap choice

As poor as SM ceramic caps are for coupling an audio signal they still are better than the best electrolytic and a nonpolarized electrolytic is one of the worst. (No bias voltage.) The difference is SM ceramic caps can take a great deal of time to break in so may sound worse at first. Go back and read all the posts about the sound of new Sonic Impact Tripath amps and how they finally settle in. The best choice is to jumper the on board caps out completely and put good film caps off the board. Depending on the circuit configuration further changes may be necessary for this to work properly.
Roger

 

[Pierre]

Now that you mention it, there has been a lot of discussion about "capacitor break in", or how do the sound change as they get used. I am curious to hear any scientifical explanation for that phenomena.

Although I can't think of any, there must be, as very serious people in the forum have talked about it!

Another question is... Is it possible that an amplifier with -not very good- coupling caps (for example, tantalum SMD), measure 0.03% THD at, say, 1 KHz, when that same amplifier would really produce much less distortion without it? Or are we really talking about the difference between a theoretical 0.000002% to 0.000001% with/without that cap?

Thanks!

 

[sx881663]

Cap quality

It sure would make things simple if we could just hook a meter up and say whether a part is good or bad sounding. All testing for components is basically static, meaning one voltage or frequency. This will not give any better correlation with Sonics than THD testing does. It can only be an indicator. Interestingly enough the small value caps used for coupling do have a strong correlation in that the better they test the better they sound. The quality polypropylene film caps discussed here test the best of all with the lowest ESR, lowest dissipation factor, lowest dielectric absorption and lowest microphonics. There are caps that might have one parameter better but not all.
Most of what makes one part sound more transparent than another is its ability to pass very low level information that is buried in with much higher level signals. This can easily be -90bd down and still be clearly heard when present. The existing distortion tests just can’t dig this kind of stuff out so it is left to our ears to do so. Our ear/brain combination is far more complex than any computer based instrument but like a computer it needs to be programmed. We call this “training” with the end result being golden ears! Just like a piece of test equipment we also must have a solid reference of “zero” distortion to compare to. This is known as live acoustic music!
In chasing after all the nuts and bolts lets never forget that our goal is to get as close as we can to that true reference! This means taking some time out occasionally and going to a real acoustic concert or chamber group or even just the church choir. One really does need to do this occasionally to realize, again, how bad our systems really are.
Roger

 

[IVX]

I can't do THD measurement less then loopback EMU1212M =.0001%@1khz, aluminium caps manufacturers claim at the rated ripple current -120db of THD, film cap probably better else, but audiophile folks is really hear sound differences between those caps (THD isn't reason, i suspect).

 

[sx881663]

Testing amps and stuff

I suspect the sonic culprit is mostly noise. This is generated by all the parts and can be modulated along with the signal hiding low level information. In a typical THD test this would look like a typical noise floor and not draw your attention.
Bottom line is to strive for the lowest possible noise of all types. Do this in selecting all components and in doing wiring layouts. Be particularly careful of grounding and keep high currents out of the reference/signal grounds.
Roger

 

[mzzj]

Now that you mention it, there has been a lot of discussion about "capacitor break in", or how do the sound change as they get used. I am curious to hear any scientifical explanation for that phenomena.

Although I can't think of any, there must be, as very serious people in the forum have talked about it!

Another question is... Is it possible that an amplifier with -not very good- coupling caps (for example, tantalum SMD), measure 0.03% THD at, say, 1 KHz, when that same amplifier would really produce much less distortion without it? Or are we really talking about the difference between a theoretical 0.000002% to 0.000001% with/without that cap?

Thanks!

D. Self for example has reported some THD measurements of electrolytics, LF distortion of cap can be 10 times worse than good quality power amplifier. Key to make it neglible is to ensure that there is no AC signal voltage across cap. This makes their use for filters impossible, but use for dc decoupling is possible if you choose enough big value so that LP -3db freq is something like below 2 hz. IF you want to limit your freq response to 20 or 16hz, DONT use electrolytics for that. Same applies to big value ceramics, their cap value is hugely voltage dependant (NP0 ceramics are ok but not availlable in big values).