100s of ceramics in parallel for big capacitance

I would be very interested in seeing some manufacturer's data/graphs concerning voltage dependency of C0G/NP0 caps. Do you have a link or other reference?

Not so much.

Voltage Coefficient of Capacitors Voltage Coefficient of Capacitors Comparison & Solutions

Temperature and Voltage Variation of Ceramic Capacitors, or Why Your

For fun, I tested this with my handy eBay capacitance meter. I didn't want to put DC directly on the capacitance meter input so I put two very big NP electrolytic capacitors in series with the capacitor under test and some high value resistors in series with the power supply. It didn't take much to confirm the derration curves shown in the above articles. I wish I had known about that back in the 1970's when I was fighting the problem and didn't recognize the cause. (20-20 hindsight).

Which shows what I remembered. C0G/NP0 has nil voltage coefficient. Thanks for the link.

The problem wasn't as bad back then. I found some 70's dead-tree specs on capacitance voltage coefficients for various ceramic capacitors. There was some loss in capacitance at rated voltage (for X7R and similar dielectrics) but not to the same extent as today. Clearly manufacturers have tried to squeeze ever more capacitance into smaller and smaller packages; it hasn't all come from materials and processing improvements, some is from 'specmanship', rating capacitors for higher voltages than they would have in the earlier era.

I have an old HP 120 volt power supply, and I've connected it to various caramic caps, rated 6 volts and up. I've never damaged one, at

The nonlinear c-v curve of these caps should be useful for something, like parametric amplification. High-voltage edge-sharpening nonlinear transmission lines (shock lines) have been done with disrete ceramic caps and inductors, and with monolithic transmission lines made from strips of ceramic.

On Fri, 19 Sep 2014 15:28:56 +0000 (UTC), Frank Miles wrote:

True. Components were physically larger and therefore less susceptible to the effects of voltage variations in capacitance (C/V). However, I found a way to create the problem. The company, Intech Inc of Santa Clara, had two divisions. I was working for the marine radio division. In the same building was the modular products division, which made A/D, D/A, amps, and other plug in modules, mostly for military use. They used tiny MLCC caps in order to cram as much as possible into the standard potting shells. The tiny capacitors were irresistible, so I started using them in the radio products. They worked well for bypassing and coupling, but were a disaster when I tried to use them in filters and RC oscillators. They wouldn't be anywhere close to their calculated frequencies and appeared to be well out of tolerance. I had two lousy solutions. One was to use 50v parts in a 12v circuit. The other was to fudge the values until things worked and ignore the discrepancy with the calculations. Both methods worked, until the next lot of capacitors arrived, and I had to make changes on the production line or in test. Even selecting parts didn't help much as the company tended to select vendors of commodity parts by price, resulting in wide variations in C/V characteristics. For one radio, I purchased what was estimated to be the all the critical capacitors that would be necessary over the life of the product, to be sure that no other vendors parts would ever end up in it's place. Of course, marketing underestimated the market, and I had to go through yet another production fire drill when we ran out of parts. If I had known about the effect, my life would have been much easier.

I cannot find any references to loss of C vs V anywhere. Do you have some info on this?

Try Dissipation factor or ESR versus voltage:

Dammit, Jeff, why don't you read the posts completely and try to follow them?

My quest was to determine if C0G/NP0 really have a voltage coefficient and, it seems, is insignificant. I do know about the voltage coefficients of other classes of ceramics.

Hi John, If you are really interested, you could measure a few and report back. If you found anything above (say) 0.1% I'd be interested to know about it.

The 2nd graph on that page includes DF versus V results for a 100pf NP0 cap. It's not a complete answer, but should offer a clue. Looks like no change in DF over 0 to 50v.

Interesting that they hardly spec the X5R parts, just two graphs and virtually no numbers. ???

On Fri, 19 Sep 2014 16:35:27 -0400, rickman Gave us:

A lot more characteristic performance evaluation data usually get included with the High Voltage types.

Check some of those out.

See what folks like Philips have to say.

Ok, I see where I caused the confusion. When I said "loss of C vs V", I really meant reduction in capacitance versus applied voltage. That is, the voltage coefficient of capacitance.

I got the answer.