[eric343]

Am I correct in assuming that one normally uses tantalum or ceramic capacitors for power supply bypassing?

 

[guzzler]

what are you bypassing?? or a better question might be, what do you mean by bypass?? it's a word that many people have many different understandings of...

rectifier diodes: ceramic or suppression caps
regulator: aluminium elcos or tantalum
opamps: metallised polyester, polycarbonate, polypropylene etc... pretty much anything fast...

wouldn't recommend ceramics in low noise sections (eg on the opamp supply rails) as they are noisy and microphonic

g

 

[aos]

It is often suggested in datasheets to use tantalum / ceramic combinations to bypass opamps and many other types of chips. If you're working at very high frequencies, ceramics (including special types) are your only option.

But here you'll most often hear suggestions to use combinations of film capacitors and aluminum electrolytics (typically specialty caps like Elna or Black Gate). Or even "triplets" - three capacitors differing about 2 orders of magnitude each (e.g. 1nF, 100nF and 100uF - I am not certain what's the rule here). I personally use specialty polymer capacitors (like OS-CON) which you should also bypass with a film cap. I can't say I ever notice a difference, I just try to approach this from engineering perspective.

 

[tangent]

It depends on what you're after. Do you need high linearity? Then you need film or NP0 ceramic. Do you care more about low ESR and high frequency than linearity? Then ceramic is good. Do you need bulk capacitance more than low ESR? Use electrolytics. Do you need a balance here, with low ESR but a significant amount of capacitance? Use dipped tantalums.

This is only scratching the surface. I've ignored microphonics, parasitics, voltage tolerance, and the entire leaded vs. SMT thing.

The "triple" issue aos brings up has to do with broadband bypass. Most caps have a kind of 'V' characteristic in their ESR vs. frequency graph. Down at the point of the V is where that cap passes signals with the least resistance, and ESR goes up from there with both higher and lower frequency signals. By paralleling several different types and values, you can overlap a few of these 'V's, which gives lower overall ESR partly because now different frequencies can take different paths through the cap array, and because ESR divides like regular resistance with caps in parallel.

There's a good app note out there about this. "How (Not) to Decouple High-Speed Operational Amplifiers" by Bruce Carter of TI. It applies to circuits besides op-amp ones, of course.

 

[rickcr42]

RF bypass , ceramic works for me

since it is filtering across a resistor and not "strictly' in the signal path (shunting actually, either across the feedback resistor or the input load setting resistor or both) it , at least to my ears , has no detrimental effects

 

[Voodoochile]

Small polyesters or metallized polypropylenes are often fast enough to do the job as well. Testing a few is a good way to determine what will fill the bill.

The very fast ceramics are sometimes more than what you need. If a slower quieter cap will handle it- all the better. As Rick notes, they are not passing a signal directly. It seems for audio OPA bypass purposes that a ceramic is often overkill for the operating parameters. (Versus telecom apps, for example)

 

[rickcr42]

Voodoochile

cool thing is

if you use ceramics for RF filtration they have no bearing on the sound

in a low gain circuit (say X 1- X10) a ferrite on the cord is not needed if you connect a ceramic directly across the (if non inverting opamp circuit) the input impedence determining resistor , which determines what freqs get IN and an "output to invert input" resistor (value determined by R value and desired cutoff , a classic single pole low pass filter when in the loop) which determines the freq cutoff of the feedback loop



another note about caps in general

they all have a "voice" and can change the voicing of the overall circuit

Active stage Too Bright ?

use a known "soft" coupling cap

Too damn soft ?

you know the answer


PLUS !

the power supply is "in circuit"

piss poor power (AC line not batt) with a GREAT design means piss poor sound or at best "something is missing"

GREAT psu with average active stage (within limits) will whip ass in comparison

solid psu and solid design

well.....................

also

always a good idea to mix brands or you will always have a device that voices to the sound character of the inline caps

say for argument-solen in,audiocap out

ying/yang

and for some reason with film caps,and i have no idea why , the higher voltage rated caps sound better

i.e. -600V sounds better than 200 V

if you have room go for the higher V rated cap

OK ,'nuff said by me on this

 

[Voodoochile]

I was actually thinking more of the use of film caps for bypassing the electrolytics and their effect on transients... but that is great info on RF filtration. Thanks for the detailed schooling.

You can't help but pick up more bits of information around here- I love it.

Quote:

Originally posted by rickcr42 and for some reason with film caps,and i have no idea why , the higher voltage rated caps sound better i.e. -600V sounds better than 200 V

That's very interesting!

 

[aos]

Quote:

and for some reason with film caps,and i have no idea why , the higher voltage rated caps sound better

That should be mentioned in the "picking capacitors" article I think. Film capacitors with higher voltage rating also have lower loss.

And rick, I could not agree more - I firmly believe PSU has a major impact on sound.

Ceramics are microphonic, if you look at datasheets of some ultralow noise voltage regulators you'll find even graphs showing what happens when you use ceramics as noise reduction caps and then tap them with a pencil - more noise than the regulator makes. I use PPS film in those positions instead.

Quote:

The "triple" issue aos brings up has to do with broadband bypass. Most caps have a kind of 'V' characteristic in their ESR

I am well aware of this issue. What I do not know is what is the actual triplet rule (determining cap value relations to each other), as it's a rule of the thumb thing that someone put forward and is sometimes quoted by DIYers. Of course it's not as easy to do this the proper way, which is what datasheets point to.

 

[tangent]

It was my impression that the one or two order of magnitude rule had no real foundation in theory. It's just a convenient rule of thumb, with the idea that it gets you a broad selection of ESR curves. If the values are too close together, the ESR curves are too similar, so the results aren't as good. If the values are too far apart, you may have an ESR hump where you need a dip. I know you know this, too, but I thought I'd point these things out to those just reading the thread for edification.

 

[Voodoochile]

Quote:

Originally posted by aos And rick, I could not agree more - I firmly believe PSU has a major impact on sound.

I'll third that.

 

[MegaMeee]

n00b question: i was gonna use wima polyester caps in my ra'1 clone, but the original i think uses tantalums. could this be significant, should i get some tantalums instead?

 

[eric343]

Buy both. In all likelyhood the Wimas will sound better; however, the remarkably good sound quality that the RA-1 apparently gets out of a Cmoy-type circuit suggests that it has been very, very carefully tweaked and voiced, so the tantalums in this case may contribute positively to the sound.

 

[Voodoochile]

In the case of tantalums, note they are polarized like electrolytics, and must be installed accordingly. You really have to squint to see the markings.

They don't last very long when installed backwards

.
Don't ask how I know that tidbit, please!

 

[ppl]

IMHO Power Supply Bypass like op-Amps and Buffers I like Polypropylene Film or Better yet Foil. The Foil Types are most often too Big to fit next to an IC so i would use the Film types like Wima Box types. These caps should be conected Directly to the Op-Amp Supply Pins as close as possible.

When more than one capacitor type is parralelled in order to get a more evean Coverage then Caution is advised, WARNING! Capacitors of Different types can and do create resonent Traps within the Circuit. so I do not recomend that the Novice parralell random disimiler capacitors and insted use a larger Film type to coverall the Frequency ranges of interest.

Regarding Small capacitors for frequency compensation then Polystyrene is best but hard to obtain, Use Silver Mica. Ceramics are a No.. No in any audio circuit including Power supply and including any power conditioners, Lots of Data is available on the Nasties of these caps with Microphonics being just one of them.

Tantalum has been used as Supply baypassing for Decades these are Nasty also. If you put more than there Max working voltage even for a mSec thay will temp short circuit. sound wize that add grain and grit and also should be avoided.