Where PP SMD caps have disappeared?

[alj]

I was just about to order some Panasonic SMD PP caps for my audio project and noticed that ALL SMD polypropylene caps have disappeared from both mouser and digikey and everywhere I looked.

What is going on?

[capt bullshot]

Maybe there wasn't enough demand for such caps? I haven't seen SMT film caps often (if at all). And maybe it's better in general to use film caps as THT devices?
I once had a project that used SMT film caps, though it worked fine I always was in doubt if these caps would get damaged by the reflow process.

[alj]

Weird. For years reflow was not a problem... then polystyrene SMD have disappeared and now PP.

Not sure what to use for audio now. C0G? Or switch back to bulky through the hole ones....

[Kjelt]

Wima?

https://www.wima.de/en/our-product-range/smd-capacitors/smd-pps/

[alj]

I never used PPS primarily because dissipation factor rapidly grows at around 10 khz.
You still can get Panasonic PPS at mouser, although they are more expensive than PP used to be.
that could be alternative to investigate (together with c0g).  But PP is already proven and cheap material for audio caps.

[T3sl4co1l]

Stacked film were always troublesome, I think.  They're an implausible bit of materials-science: the film melts well before soldering temperature is reached.  The stack inevitably flakes open a bit, allowing flux in -- the leakage isn't nearly as good as an encapsulated THT part, and they're impossible to wash.

What's wrong with THT?  Just prefer automated assembly of chips?

Audio what?  Coupling caps?  Those really don't care, they can be X7R.  Microphony is a big drawback, in which case consider electrolytic, tantalum or polymer.  Electrolytic is especially appropriate for coupling: the modest ESR and loose capacitance tolerance don't matter in the slightest.

X7R would be undesirable for filters, where the distortion would be measurable.  (Notice I didn't make note of distortion in regards to coupling.  It's just not an issue.)  Smaller values should be reasonable there, though, in which case C0G would be economical.

Or you can always go the industry route and pull everything into the digital domain.  DSPs are cheap and plentiful!  Probably a lot slower to program than an analog filter though.

Tim

[dmills]

Electrolytic is especially appropriate for coupling: the modest ESR and loose capacitance tolerance don't matter in the slightest.

True, but your marketing department will hate you!
The art of the coupling cap is just to make it stupidly high value, so there is bugger all signal voltage developed across it, elcos are good for this.

For filters, I concur that C0G are (finally) available in useful values, and are pretty much blameless.

About the only place that film males sense is if you really must do the passive crossover thing for loudspeakers, where the cap is almost as much of a pain in the arse as the inductor, better to cross over at line level where sane components are available.

Regards, Dan.

[T3sl4co1l]

True, but your marketing department will hate you!

That's why I don't make audio products.

Tim

[alj]

X7R will not work for sure. 2% dissipation factor. Comparing to 0.15% c0g.   and... 0.02% !!!! of PP

Yes usually small values are used in filters (such as RIAA corrector) or another example - TDA1541A has 16 external decoupling 47n caps - ceramic is really bad there and not even sure why - dissipation factor or resonance?

What is wrong with C0G? Any ceramic cap is piezo resonator.  And on high values closer to 1uF resonance frequency is in audible range - you can even see it on data sheet. Also above .1 uF value the price is going up pretty quckly and above 1uF it is unobtanium.

Decoupling is not a problem - large TTH PP can be used or good electrolytes or even no decoupling at all (servo).

What is wrong with using TTH? Size!   When you put 15 decoupling caps around  TDA1541A as close to legs as possible, you just solder SMD ones right under the chip.  15 tth caps need to be around chip pushing other critical components further away on both sides of PCB board.

[alj]

You mean Epcos? I find that Nichicons are as decent as Epcos and they cost less.

yep you need to pick cap at least 10 times bigger than calculated to avoid hysteresis, and measurable distortion grows on caps rated lower than 50V so they have to be 50V for max performance which means they all going to be pretty big.

[dmills]

Type I dielectrics (Which C0G are) are not piezoelectric, that is type II, I don't see anyone suggesting type II in a filter network (Because that would be crazy, especially as they tend to have a voltage coefficient as well).

God, a '1541 haven't used one of those in the best part of 20 years, but I remember them as needing a stupid lot of passives compared to a modern part.

47nF, 25V, 0805, C0G, try a Kemet C0805X473K3GECAUTO7210 for less then 10c each in quantity 100 as an example of a modern production part?  Whats not to like? Cheaper then film, survives reflow without needing wanky reflow profiles and has lower ESL then a thru hole part will manage (And this one is even soft term so likely to survive hand soldering).

Regards, Dan.

[alj]

Not trying to argue but I found this graph
http://www.cde.com/resources/catalogs/ceramperf.pdf
Which shows that c0g is even more resonant than X7R (by looking at impedance vs frequency for equal values). Maybe lower dissipation increases q?   


Yes modern DACs are easier to use. And I tried them.  They just sound boring. Again I do not know why (only guesses) and it is subjective but other people besides me had similar experience.

[TimNJ]

Just use C0G and call it a day. 

[T3sl4co1l]

X7R will not work for sure. 2% dissipation factor. Comparing to 0.15% c0g.   and... 0.02% !!!! of PP

Are you making high Q or passive resonant filters or oscillators?

Yes usually small values are used in filters (such as RIAA corrector) or another example - TDA1541A has 16 external decoupling 47n caps - ceramic is really bad there and not even sure why - dissipation factor or resonance?

What's so bad about ceramic decoupling?

What is wrong with C0G? Any ceramic cap is piezo resonator.

Any type 2 dielectric is a poor piezo transducer.  Not usually much of a resonator.

C0G is a completely different material, and does not exhibit piezoelectricity.  It does exhibit electrostriction, just as any other capacitor (including PP) does.

And on high values closer to 1uF resonance frequency is in audible range - you can even see it on data sheet. Also above .1 uF value the price is going up pretty quckly and above 1uF it is unobtanium.

What resonance frequency?

Also, if you're referring to the Z(F) curve, that's not even what you'll get in circuit.  That's the minimum possible case, with the capacitor placed in a fixture which is nulled for ESL of connecting traces.  The real trace length in a typical application will be several times the body length.

But you can rest assured even that doesn't matter at audio frequencies.

What is wrong with using TTH? Size!   When you put 15 decoupling caps around  TDA1541A as close to legs as possible, you just solder SMD ones right under the chip.  15 tth caps need to be around chip pushing other critical components further away on both sides of PCB board.

Audiophool paranoia or small size?  Which will win? STAY TUNED TO FIND OUT!

Tim

[dmills]

Of course lower dissipation increases Q, that is kind of basic, you add series resistance to lower Q, so a less dissipative cap will inherently have higher Q.

Those curves by the way are in units of MHz, you would struggle to get down into the audio band with these parts and reasonable trace inductance (very low ESR electrolytic caps in power supplies on the other hand can cause problems this way in audio circuits, better to use more lossy varieties).

73 Dan.

[alj]

I posted link - resonant frequencies are here  http://www.cde.com/resources/catalogs/ceramperf.pdf 

Whats wrong with ceramics? Like I said "i'm not sure why" but the difference is clearly audible when i replaced them with PS in my previous TDA1541A project (yes there were times when I could get SMD Polystyrene). They were no c0gs though. However i'm afraid to go that direction now because if c0g does not work out for some reason i will have to redo PCB for TTH ones due to nonavailability of SMD PP which already proven to be fine but have suddenly became unobtainable.


Small size goes fine with "audiophool" paranoia :-) Why chose when both can win?

[alj]

Dmills DOH did not notice it is Mhz. Fair enough! Ok we can assume c0g will  work fine in analog filters then.

However on TDA1541A decoupling caps work on DEM clock frequency which in my case will be division of 11Mhz main click (either by 2 or by 4).

[floobydust]

I think SMD C0G ceramics could perform worse, despite being technically superior- they aggravate side-effects.
TDA1541 is a big DIP-28 though-hole part (?) and I would (in it's day) expect several nH including the boxed film cap's lead and the PCB trace and DIP package inductance. A boxed MKP 0.047uF is around 7MHz SRF, 0.1uF is around 6MHz.
It might be desirable to have that high ESL on those DAC decoupling capacitors.

Don't be afraid to add a few ohms of ESR to those caps and experiment. Something to lower the Q and lessen ringing. It's not always about DA, and I find high-Q decoupling circuits can perform worse depending on the PCB layout and resulting self-resonant frequency.

[dmills]

High Q decouplers can indeed be a mixed blessing, and Q killing resistors can be very much a thing, but it depends on context and sometimes (especially with old, slow, mixed signal parts like that) the only way to know is to experiment.

The C0G seems to have a SRF in the 50MHz or so region, so unlike the boxed cap will at least still be a capacitor for the first few harmonics of the clock!

There is good discussion of resonances in power networks (and a recipe for a useful broadband power filter that actually behaves up to GHz frequencies) in "High speed signal propagation", well worth the time.

Regards, Dan.

[alj]

So, for 47n cap on 176kHz dem clock frequency if i add 0.1 Ohm in series the q factor will be about 188. Does that sound acceptable to kill ringing?

[T3sl4co1l]

Why would it be ringing at the clock frequency..?  I'm not sure what you're after.

Tim

[alj]

They wont but the idea was to keep reasonable q factor at the frequency they will be used on and on higher frequency (where they will resonate) q would be low enough to not cause any troubles. But i think I'm overthinking this and i'l try without any resistors first.

[floobydust]

It's worth looking at the DAC datasheet, it seems to have resistors already built-in.