Quality audio SMT caps?

[Araho]

Hey all!

Anybody know of any surface mount capacitors that have good audio performance? I did some googling&reading, but most of the posts are from 2009 or older, and everyone in those forums are basically throwing tons of shit about the quality, thermal noise etc of SMT caps.

So I was hoping someone here knew about any higher quality capacitors that don't cost a shirt, three fingers and an eyeball. Anyone?

Araho

[madires]

Are you asking about caps in the signal path like coupling caps? Please tell us a little bit more about your circuit or what you are trying to accomplish.

[strobot]

I have used Cornell Dubilier FCA1210C105M-G2 acrylic SMT film caps with good experience for audio AC coupling.

SMT film caps are expensive.

[mtdoc]

Just use any old SMT cap and coat it with some of THIS. Or perhaps cover it with a small piece of mahogany....(sorry - couldn't resist.. )

[Araho]

@Madires: Yes, I'm talking about caps in the signal path for AC coupling

Basically it's the input of an ADC that requires a 47uF cap in the signal path, and I would like to keep most parts SMT.

@Strobot: Thanks, I'll check them out!

@Mtdoc: Hehe, I know, audio sound quality/difference is often placebo. But there definitely is a quality difference in thermal drift, thermal noise etc in passive components, and it seems like a good idea to consider how it affects performance/sound

[free_electron]

panasonic filmcaps for SMT. Problem is that filmcaps are problematic in reflow ovens. Wave soldering filmcaps is not a problem as the part itself doesnt really heat that much. in a reflow oven the part goes through the hot zone. That is why you see a lot of thru hole filmcaps still being used on boards that are otherwise almost all SMT.

[strobot]

...47uF cap in the signal path, and I would like to keep most parts SMT.

I have never seen any SMT film caps >1uF.  To spec a 47uF film cap in the signal path is extremely uncommon and would be very expensive.  You should look in to tantalums or electrolytics.

[envisionelec]

I used Niobium Oxides for coupling in a high end audio design I did a few years ago. Their performance was excellent.

[madires]

Assuming 10Hz for the -3dB bottom frequency limit the ADC would have an input impedance of around 338 Ohms if it requires a 47µF cap. Is the ADC's input impedance really that low?

[TMM]

Assuming 10Hz for the -3dB bottom frequency limit the ADC would have an input impedance of around 338 Ohms if it requires a 47µF cap. Is the ADC's input impedance really that low?

This.

The largest coupling cap i've needed to use was 20uF (two 10uF PET caps in parallel, through hole) and that was for an audio measurement system that needed negligible roll off in response down to 20Hz.

Unless it is for a theater subwoofer or a measurement system you can probably get away with a much smaller coupling capacitor.

[c4757p]

Assuming 10Hz for the -3dB bottom frequency limit the ADC would have an input impedance of around 338 Ohms if it requires a 47µF cap. Is the ADC's input impedance really that low?

I don't know what this is for, but I've seen a lot of hobby-type audio schematics floating around that specify improbably large capacitors. I recall recently seeing a schematic for a guitar distortion pedal that wanted 1uF to couple into a stage with an input impedance of some 500k...

[tszaboo]

@Madires: Yes, I'm talking about caps in the signal path for AC coupling

Basically it's the input of an ADC that requires a 47uF cap in the signal path, and I would like to keep most parts SMT.

That is not AC coupling, that is decoupling. 47uF is not that uncommon for high bit count ADCs, but they specify ceramics for that.

[Araho]

This picture is cut directly from the datasheet of the ADAU1701 SigmaDSP chip. It has a stereo DeltaSigma-ADC on these inputs. This is a current-input ADC, so it can accept different voltages on the input simply by scaling the input resistor. The full-scale input is 100uA rms.



This doesn't look like decoupling to me, NANDBlog?

@Envisionelec: I'll check out some Niobium Oxides later today, thanks As far as I've seen yet, they're not too expensive and can be found in reasonably large capacitances.

[larry42]

I've never heard of thermal noise being a problem with caps in audio circuits.

Capacitor non-linearites however, are well known.

Here is an app note about it for other readers that doubt the effect:
http://www.maximintegrated.com/app-notes/index.mvp/id/3171

I guess I would look at a plastic film based or aluminium electrolytic (back to back to prevent polarity issues)

[Araho]

Thanks Larry, that article was really interesting!

[krivx]

Assuming 10Hz for the -3dB bottom frequency limit the ADC would have an input impedance of around 338 Ohms if it requires a 47µF cap. Is the ADC's input impedance really that low?

I don't know what this is for, but I've seen a lot of hobby-type audio schematics floating around that specify improbably large capacitors. I recall recently seeing a schematic for a guitar distortion pedal that wanted 1uF to couple into a stage with an input impedance of some 500k...

It's common for people to modify existing designs for bass guitar (or for more low-frequency performance) by experimenting with larger coupling capacitors. Someone may have just increased a coupling cap by a factor of 10 or 50 and liked the result.

[larry42]

Thanks Larry, that article was really interesting!

Analog Devices and TI have a couple of similar notes. I would avoid most ceramics in the coupling path, if size is not an issue. Many ceramic are piezoelectric and I've literally heard 1206 sized caps "sing" (i.e. mechanically resonate) at ~8kHz when hit with a sine wave of a couple of volts.

Of course other materials have other disadvantages (cost, size, ESR, ESL, ripple current limitation) - tradeoffs

[grenert]

Cornell Dubilier make acrylic SMD caps (FCA series) that I've had good luck with in audio projects.  However, I doubt they go up to 47uF.  That is really too large for a surface mount film cap. 
I agree with others that you need to ask if you really do need such a high value cap.
For large-ish values under a microfarad, I've been very happy with NP0 ceramics (and only NP0).  They are really distinct from other ceramic caps in many ways.

EDIT: Silly me, I see that Strobot already mentioned these.  Well, to add some novel content, here are datasheets of four SMD film cap series (Panasonic ECPU, ECWU, ECHU; CDE FCA) and a TDK NP0/C0G cap.

[ampdoctor]

Assuming 10Hz for the -3dB bottom frequency limit the ADC would have an input impedance of around 338 Ohms if it requires a 47µF cap. Is the ADC's input impedance really that low?

I don't know what this is for, but I've seen a lot of hobby-type audio schematics floating around that specify improbably large capacitors. I recall recently seeing a schematic for a guitar distortion pedal that wanted 1uF to couple into a stage with an input impedance of some 500k...

It's common for people to modify existing designs for bass guitar (or for more low-frequency performance) by experimenting with larger coupling capacitors. Someone may have just increased a coupling cap by a factor of 10 or 50 and liked the result.

You've got to be careful when comparing instrument amplifiers designed for sound production to home or pro audio equipment designed for sound RE-production. With the former you're actually creating the sound in an artistic and musical way, while the latter you want to hear exactly what the artist is trying to hear the sound exactly the way the artist intended it to be heard. Two very different worlds.

In guitar, bass, and various effects pedals on average the last thing you want is low distortion and flat frequency response. You actually WANT to limit the bandwidth, create mid-band dips and peaks, increase some harmonics while reducing others, introduce feedback loops that will cause circuit instability under certain conditions, etc. It's almost anti-engineering and is incredibly challenging to do well. As an example Larry pointed out that ceramics can be quite microphonic, are are often times a death sentence in instrument amplifiers. But sometimes you want exactly that! There's a whole lot of cut and try and not necessarily a lot of "formal engineering" going on.