Page 1 of 4

SMT passives for audio

Posted: Sat Mar 05, 2011 5:19 pm
by SUBass
I've really gotten interested in working with SMT as I don't have to drill out a PCB and can "cook" the board on the hot-plate in a matter of minutes and be done.

That said...What are people using for SMT parts for audio circuits? Is there a particular size/brand/line of resistors that you like to use? What about caps? There's obviously many different tasks for caps and different technologies are better suited to different locations. Sooo...

What are your choices for AC coupling? What about PS bypassing and filtering/reservoir?

Thanks!

James

Re: SMT passives for audio

Posted: Sat Mar 05, 2011 6:43 pm
by JR.
The parts for SMT and TH (thru hole) PCB are generally made by the same folks so similar.

About the only major difference I find is that some of the film capacitor types I liked for audio are not robust enough to handle the temperatures of IR reflow and SMT processes in general. I tried and had problems with value drift and outright failures.

For modest values NPO or COG ceramic SMT caps are more than adequate for audio linearity.

JR

Re: SMT passives for audio

Posted: Mon Mar 07, 2011 12:25 pm
by mediatechnology
Thanks for joining us!

There are some pretty significant differences in SMT resistors which, due to signal-induced self-heating, have better or worse THD-3. On a recent demo board THAT used Susumu because they tested better. Those are available from Digi-Key. Dave Hill of Crane Song shared some of his results and has done quite a bit of testing of SMT resistors. His measurements also confirmed that the Susumu were the best performers of the half-dozen or so brands he tested. IIRC the Panasonic were pretty good but not as clean as the Susumu.

I also just saw an article about acoustic noise generated by SMT ceramic caps. I'll try to remember where I found that.

Re: SMT passives for audio

Posted: Mon Mar 07, 2011 1:26 pm
by emrr
Good knowledge pool to build, thanks!

Re: SMT passives for audio

Posted: Tue Mar 08, 2011 7:23 am
by SUBass
About those Susumu resistors...Digikey has about a dozen series of them in three different compositions. Thin film seems to be the only practical composition as the others have too small of values. Unfortunately that still leaves a few series. Off to the data sheets I guess. Another comment....I seem to recall reading something that the physically larger components are better behaved. Like given two resistors of the same value and composition, the larger of the two would be better behaved.

Anyone have any experience with Niobium Oxide caps? They're supposedly good for audio coupling but only come in really low voltages. No single-supply circuits for those guys.

Re: SMT passives for audio

Posted: Tue Mar 08, 2011 7:29 am
by mediatechnology
I seem to recall reading something that the physically larger components are better behaved. Like given two resistors of the same value and composition, the larger of the two would be better behaved.
We up-sized resistors by one size on a later spin of the board and changed vendors to reduce the self-heating THD. This was in an output buffer and the values were fairly low, 1-2 K or so IIRC. I'll dig in my old e-mails and see if I can get you the Susumu series.

Re: SMT passives for audio

Posted: Tue Mar 08, 2011 8:49 am
by mediatechnology
Here's the Susumu link:

http://www.susumu.co.jp/english/pdf/products-j01-04.pdf

Availability of these resistors with "comfort sound" through distribution was an issue:

http://www.susumu.co.jp/english/pdf/products-j01-05.pdf

Re: SMT passives for audio

Posted: Tue Mar 08, 2011 9:00 am
by JR.
SUBass wrote:About those Susumu resistors...Digikey has about a dozen series of them in three different compositions. Thin film seems to be the only practical composition as the others have too small of values. Unfortunately that still leaves a few series. Off to the data sheets I guess. Another comment....I seem to recall reading something that the physically larger components are better behaved. Like given two resistors of the same value and composition, the larger of the two would be better behaved.

Anyone have any experience with Niobium Oxide caps? They're supposedly good for audio coupling but only come in really low voltages. No single-supply circuits for those guys.
It is not simple to make broad generalizations. For passive components like resistors that dissipate power, physically larger packages mean more surface area and thermal mass, so hopefully lower and less fluctuation in operating temperatures. For components like capacitors, size is often reflective of technology. Modern electrolytic caps are often smaller than the old ones they replace with generally better performance (there are exceptions). Film caps are generally larger than electrolytic caps for a given capacitance/voltage product, etc.

Within a resistor there is a constant residual noise mechanisms related to temperature and value (Johnson noise), and another modulation noise mechanism related to the granularity or smoothness of the resistance element wrt current flow (excess noise). This excess noise for audio applications is generally masked by signal since it is several tens of dB below the nominal signal levels. There are subtle differences in excess noise between different resistor types. For modern solid state design, one might want to use a low excess noise technology resistor for something like a phantom supply resistor that sits there with DC across it, but again this noise is generally well below other noise sources around such circuitry. Most use precision resistors there for tracking and get low excess noise as a serendipitous unintended consequence.

JR

Re: SMT passives for audio

Posted: Tue Mar 08, 2011 9:44 am
by mediatechnology
We went from 0603 to 0805 resistors for the lower values. In larger values there's unlikely to be a difference but in the low values where the signal current was a significant percentage of the power dissipation the tempco caused modulation.

I agree with John that with caps bigger isn't necessarily better.

In an e-mail with bcarso some time ago he mentioned that Audio Precision had issues with SMT resistors in attenuators. Eventually they went back to through hole for certain values.

In mic preamp applications the phantom pullup resistors are not trivial nor are the series current limiters. We went with TH there due to supply issues. The pullups were noisy in SMT (what we could actually get) and the SMT in MELF for the series limiters were also hard to source. Those see high peak currents.

But, in a 100K SMT I'll doubt that you'll be able to measure much difference in THD from an 0603 to 0805. So the answer to "is bigger better" is "it depends."

This comparison is between the Susumu and TH:

Image

Re: SMT passives for audio

Posted: Tue Mar 08, 2011 3:26 pm
by JR.
Am I correct to read those plots as susumus were "as good' as metal film TH?

JR