SMD capacitor questions

[Fancy Lime]

Hi there,

I've been looking into whether or not I dare to finally dive into SMD. Lot's of questions there.

I have never worked with tantal caps but with SMD that is probably the way to go for large polars. Is there some sort of design guide for safe design (avoiding surge currents....) with tantals?

Which ceramic types are ok to use in filters? Only NP0 / C0G?

Which type of ceramic should be used for larger capacitances?

Does anyone here use mixed SMD/TH? It seems that larger value film caps are difficult to get and I feel it may sometimes be easier to ass a few TH caps instead of circumnavigating the problem with parallelizing smaller caps or using polars.

Thanks,
Andy

[vigilante397]

SMD is the best

What are you dealing with that you have to worry about surge currents? Not a common thing to worry about with pedals. I use tantalums in most cases as even SMD aluminum electrolytics take up too much space. The only electrolytics I stock anymore are 470uF because I can't find a tantalum in that value, everything else I use tantalums.

With ceramics NP0/C0G is the goal, but they aren't always available in the value you need, I've found that X7R are good enough, but I wouldn't recommend anything below X5R.

For larger capacitances I still say stick with X5R and above, obvsiouly with NP0/C0G being the goal if they're available and you're willing to pay for them.

I only really mix SMD and TH if mojo is involved, or if there's a certain part (like BBD) that aren't available in SMD, though for my tube builds I use through-hole 250V filter caps as SMD caps that value at that voltage are pricey.

[nickcordle]

Been mixing SMD / TH ICs, 0805 resistors, TH greenies and electros for a while now.  So I've just been telling myself to think of the holes as vias as I'm routing.  The end results are fine but it's cumbersome and I'd like to get around it, particularly the filter greenies.  Wouldn't even need something really tiny, greenie or box film-sized but with pads instead of leads would be great.

Glancing around on Mouser, SMD film is $$$, NP0/C0G somewhat better but still $$ as the values get big, like above 22 nF, and then the values stop.  I assume that matches what you have found ...  So I guess to get the price down to where I'm accustomed, it's X7R.  Which is why I'm still coasting on my stash of greenies.

Sorry for a non-answer but consider it a thread bump, I too would be interested in more info or an explanation of the physics of why we can't have what we want.

[Fancy Lime]

@Nathan,

thanks, that is very valuable info for me! Had a look at your webpage. Nice designs and good looking boards, too.

What surge currents I worry about? That's the thing, I don't know. All I know that tantals can get destroyed by surge currents but I have no idea if we are talking pA, nA, µA, or mA for any specific Ta cap. I am assuming they come in various current ratings but was unable to find much useful info on that in datasheets. Surge voltage ratings are sometimes given.

So, what do you use for coupling caps? I sometimes find myself needing values in the 220nF to 4.7µF range. The latter, admittedly only for very specific designs when I try to keep the component count low by combining functions that maybe should not be combined (filtering and envelope followers and such). Much of this can probably be designed differently to suit the parts availability or be changed to polar caps, especially when total part count is not as much of a concern because the parts are tiny. But sometimes a non-polar 1µF  cap would be useful. What then? X7R straight in the audio path? TH film? I guess I will also have to invest some mental energy into thinking over the impedances of certain recurring filter arrangements.

@Nick
Yeah, what puzzles me most is the enormous price range and seemingly random pricing. I found ceramic SMD caps, at the same retailer, from the manufacturer, with identical specs in all but one category, with a 4-fold price difference. The only difference was the voltage rating (10V vs 25V) and the one with the higher rating was the cheaper one. All other parameters were identical. As soon as you look at other parameters being different, especially size, 10 fold differences are common. Again often with the "higher spec part" (smaller, higher voltage, better T-coefficient...) being the cheaper one. But not always.

How about cutting the leads on through-hole caps and bending them at right angles so that they are underneath the body? Not comfortable to solder but (possibly) better than having to drill holes.

Cheers,
Andy

[nickcordle]

How about cutting the leads on through-hole caps and bending them at right angles so that they are underneath the body?

The snoot lord in me cringes at the thought (which has occurred to me), but I've seen waaaaaaay dumber stuff than that in some very highly regarded gear, so ... maybe soon!  Hahaha

[Rob Strand]

So, what do you use for coupling caps? I sometimes find myself needing values in the 220nF to 4.7µF range.

For non-critical stuff you can use X7R.  If you want less distortion you can use larger SMD packages.

For small runs, or when you have other through-hole parts, you can still use through-holes like green-caps.   Many manufacturers of audio equipment still do that.

There are options nowadays to replace through-hole films,
https://www.avx.com/products/film-capacitors/smd-film-caps/

[vigilante397]

I use 0805 for everything I can get away with, it's my favorite size. I can get 0805 X7R caps in useful voltage ratings up to about 10uF, so I use X7R in the audio path all the time, and I have had zero complaints about the sound.

As for the tantalums, I have used them to filter as much as 100mA, though admittedly nothing more than that as my tube designs use either aluminum electrolytic (HV lines) or ceramic (heater supply lines) for filtering.

Wait a second, I use tantalums all the time at work in my day job. I just finished a design that uses tantalums as filters on a 20A output line, wasn't a problem (though admittedly these tantalums were $20 a piece). It's being tested by a third-party company right now and they haven't reported any problems yet either. For aircraft designs I've been told to use exclusively tantalums as aluminum electrolytics in harsh conditions are a no-no.

Had a look at your webpage. Nice designs and good looking boards, too.

And thank you, I always enjoy compliments

[Rob Strand]

I use 0805 for everything I can get away with, it's my favorite size. I can get 0805 X7R caps in useful voltage ratings up to about 10uF, so I use X7R in the audio path all the time, and I have had zero complaints about the sound.

Have a look at the "DC Bias Characteristic" graph, which is essentially capacitance vs applied voltage,
4u7 in an 0805 package,
https://product.tdk.com/en/search/capacitor/ceramic/mlcc/info?part_no=C2012X7R1C475K125AE

Effect of package size.

[vigilante397]

Interesting, I've never had any problem finding 10uF 16V in an 0805. I guess they're not TDK I'll take a look and see if I can find who makes them.

EDIT: Just looked through my last order, looks like they're Samsung. https://www.digikey.com/product-detail/en/samsung-electro-mechanics/CL21B106KOQNNNE/1276-2872-1-ND/3890958

[Rob Strand]

Interesting, I've never had any problem finding 10uF 16V in an 0805. I guess they're not TDK I'll take a look and see if I can find who makes them.

EDIT: Just looked through my last order, looks like they're Samsung. https://www.digikey.com/product-detail/en/samsung-electro-mechanics/CL21B106KOQNNNE/1276-2872-1-ND/3890958

IIRC the choices do thin out at 10uF.  I have found a few in the past as well.

More the problem is even at reasonable operating voltages the capacitance becomes non-linear and less than the nominal value.  If you have any DC bias the in-circuit capacitance can be quite a bit lower than the nominal value (10uF, 4.7uF).    The non-linearity promotes distortion.  What that graph shows is as you decrease the package size the voltage dependency gets worse.   For a given package size and cap value the effect is quite similar across all models and manufacturers - I've spent many hours troweling through datasheets.

That's why I recommend larger packages for larger value X7R's.   Films don't have this problem on any human-visible scale.

[PRR]

> What that graph shows ...

Some readers may benefit from seeing what the graph "should be".


As you say, plastic film, paper, and good electrolytics have "no" change with voltage. All the "Xxx" dielectrics do. When they were a big tan disk with 200V rating and 1V signals, that was OK (for some uses). When you ask "how small can it be?" you get these absurd errors, like being 66% down at 9V.