Simple question about mono ceramic capacitors

I need to order some .1uF mono ceramic capacitors but there are so many, I don't know what to get. These are being used for HF suppression.

Where? Can you post the circuit diagram?

Figure out the operating voltage, then pick a cap with a max rating larger than that. All things being equal there's no technical penalty for using a 630V ceramic where a 100V model would have worked too; the former might be more expensive though.

Figure out how much space you have. Do you need a through-hole cap or an SMD one?

As for dielectrics, C0G/NP0 is better than X5R is better than X7R. C0G/NP0 is too expensive to be justified just for a bypass cap. Never use Y5V/Z5U caps if you can help it.

0.1u/100n is pretty large for an HF bypass cap; almost all capacitors you'll find in that value have a rather low self-resonance point (ie: the point where they stop acting like capacitors and start acting like inductors).

For more than you ever wanted to know: AVX have a free ceramic cap simulator, which shows you the performance of various caps at different voltages and frequencies. It's called SpiCap3, you can download it from their site.

abbey road d enfer said:

Where? Can you post the circuit diagram?

Click to expand...

+1.

JD 'context, context, context' B.

Here's the schematic:

http://www.dbxpro.com/product_downloads/Schematics/163X%20Schematic.pdf

The caps I need to replace are C3, C4, C8, C9, C10, C11, C18, C19.

If you wanted to know what else I did to the circuit, I replaced VCA 1 with a 2180A, OA1 with a LME49860 and OA2 with a LME49710. I also replaced all the power supply caps with Nichicon HE series and at least doubled their value and gave them a WIMA polyprop bypass, replaced the small electrolytics (C5, C13, C15) with WIMA MKS-2 polyester caps with a WIMA polyprop bypass, and replaced the mylar caps (C1, C2, C12, C17) with WIMA polyprops.

Just like JDB, I would recommend X5R. In any case, ceramic caps offer the lowest series inductance, which is the determining factor in power supply decoupling; then, X7R being mediocre and C0 too expensive, the choice is rather obvious.

X5R only goes to 50V and are not available in .1uF, at least from mouser. I need through-hole.

You are right about the C0G/NP0...$1.50 each!

Mouser Part No: 667-ECQ-V1104JM

Is it ok to replace a ceramic for a polyester?

X5R is better than X7R.

Click to expand...

Why? http://en.wikipedia.org/wiki/EIA_Class_2_dielectric

Samuel

jdbakker said:

0.1u/100n is pretty large for an HF bypass cap; almost all capacitors you'll find in that value have a rather low self-resonance point (ie: the point where they stop acting like capacitors and start acting like inductors).

Click to expand...

So, some 6 cent Vishay X7R ceramics will be ok for bypass caps? Should I stick with the original value of .01uF instead?

sws2h said:

jdbakker said:

0.1u/100n is pretty large for an HF bypass cap; almost all capacitors you'll find in that value have a rather low self-resonance point (ie: the point where they stop acting like capacitors and start acting like inductors).

Click to expand...

So, some 6 cent Vishay X7R ceramics will be ok for bypass caps? Should I stick with the original value of .01uF instead?

Click to expand...

I'm curious about JDB's comment also. Most bypass caps (on PSU boards as well as audio boards, close to opamp V+/- pins, etc) I've seen in projects around here are 0.1uF (100n) and no one's ever mentioned the low self resonance point before. Would .01 be better?

sws2h said:

X5R only goes to 50V and are not available in .1uF, at least from mouser. I need through-hole.

Click to expand...

You need only 15V!

mitsos said:

I'm curious about JDB's comment also. Most bypass caps (on PSU boards as well as audio boards, close to opamp V+/- pins, etc) I've seen in projects around here are 0.1uF (100n) and no one's ever mentioned the low self resonance point before. Would .01 be better?

Click to expand...

Me too, curious. A 100nF ceramic has its typical minimum impedance at 3-8MHz, as compared to a 10nF, which is at 10-20MHz. If you're concerned about RFI, smaller caps may be more effective than larger ones in a particular bandwidth. The choice of smaller caps may be done according to the environment (like a 20MHz clock), but I think that, for a unit that does not have digital signals in, larger caps are preferrable. And anyway, any less-than-perfect PCB routing decision can ruin the theory.

I use 0.1uf/100V X7R.  $0.04 for 2500 qty.  

X5R = -55C min, +85C max temp, 15% tolerance
X7R = -55C min, +125C max temp, 15% tolerance

Though you can buy some X7R caps with only 10% tolerance.

You can try connecting a scope to the output of your gear, and see the effect of with/without bypass caps and different values. You'd be amaze at how much noise bypass caps can remove. Conclusion: Use power supply bypass caps every time around each chip.

And regarding board design, position them as close as possible to the chip.

abbey road d enfer said:

mitsos said:

I'm curious about JDB's comment also. Most bypass caps (on PSU boards as well as audio boards, close to opamp V+/- pins, etc) I've seen in projects around here are 0.1uF (100n) and no one's ever mentioned the low self resonance point before. Would .01 be better?

Click to expand...

Me too, curious. A 100nF ceramic has its typical minimum impedance at 3-8MHz, as compared to a 10nF, which is at 10-20MHz. If you're concerned about RFI, smaller caps may be more effective than larger ones in a particular bandwidth. The choice of smaller caps may be done according to the environment (like a 20MHz clock), but I think that, for a unit that does not have digital signals in, larger caps are preferrable. And anyway, any less-than-perfect PCB routing decision can ruin the theory.

Click to expand...

No argument with any of that, and with 5532/072-class op-amps all is just fine with 100n caps.

Then again, it would appear that several people are using (much) faster op-amps in audio applications these days. These parts, like the LM6xxx/THS4xxx series or one of the other ADSL/video chips that are in fashion, do care about HF impedance. I've found in my own designs that a lower-valued cap can be more effective, I believe that some manufacturers' data sheets and app notes recommend likewise. One DIYish example would be the TPA6120A2 headphone amp, where IME some extra care towards the supply decoupling goes a long way. An additional concern is that better-behaved C0G/NP0 capacitors are affordable in 10n values, not so much for 100n, and the HF impedance upgrade in going from 100n X5R to 10n C0G can help quite a bit for the more temperamental parts.

Give a moment's thought to what the cap is supposed to achieve before blindly plonking in another 100n part is all I'm saying.

(You might well argue that a part fast enough to care this much about its power supply decoupling will certainly not want to live in a DIP package - let alone an adapter board with no local decoupling - and an upgrade through blind op-amp swapping will be a bit like Russian roulette; I couldn't possibly comment).

JD 'think before you plonk' B.
[stuff about X5R vs X7R to follow tomorrow, once I've had a chance to try and back remembered anecdotes and vague theories with harder evidence. Sleep first, though]