X7R ceramic capacitors aren't supposed to loose this much capacitance surely?

Hello,

We are using a high voltage linear regulator as in the attached mains connected schematic. Its an LR8 linear regulator.

We simply wanted to know how much current was being drawn from its output by our circuit….nothing more.

Therefore we simply noted the depth of the voltage trough at its input rail. As you know, if the input capacitance is known, it is then possible to derive the current drawn from the linear regulator simply by doing this. The voltage trough went down to 134V, and since the input capacitance was 2 paralleled 100nF ceramic capacitors (C1210C104KBRACTU), this means that the current drawn from the linear regulator is 7mA…..

…..However, it turns out that this is not the case. The current being drawn is actually only 3mA. (we found this out by later measuring it directly with a sense resistor)

This must mean that the 100nF capacitors (1210 size) are suffering a loss of capacitance as the voltage rises on them…however, they are 630V rated , X7R ceramic capacitors, and the peak of the voltage on them only goes up to 340V.

So howcome these 630V rated ceramic X7R ceramic capacitors are suffering such severe loss of capacitance?

The 200nF's worth of ceramic capacitance has behaved as if it were only 87nF.

An LTspice simulation of the situation is also shown.

C1210C104KBRACTU 100nF ceramic capacitor datasheet:
www.farnell.com/datasheets/2316000.pdf?_ga=2.58930917.935961512.1522260916-287702491.1506211291&_gac=1.204604068.1521312230.EAIaIQobChMI2Yrg04H02QIVrLDtCh19UwCDEAAYAiAAEgJOTPD_BwE

LR8 linear regulator datasheet:
https://www.google.co.uk/search?sou...1j46i131k1j0i10k1j0i22i10i30k1.80.08upV730r1Q

You are correct, it should not lose that much capacitance based on voltage derating.
But , how about initial tolerance ? And temperature ?

That capacitance loss for the DC bias sounds right to me.

This datasheet here contains DC bias information but is not for your exact capacitor voltage rating but it is same package and also a high voltage X7R capacitor. At 50% rated voltage you get about 56% the rated capacitance.

**broken link removed**

You are operating at 53% rated voltage and are getting 43.5% capacitance. Considering that your cap is higher voltage in the same package compared to the one in the datasheet, your would expect more capacitance loss for the same % of DC bias compared to the capacitor in the datasheet, which is exactly what you see.

I got this from:
https://ds.murata.co.jp/simsurfing/index.html?lcid=en-us

Browse it to look at a bunch of random capacitors and look at influence different combinations of voltage rating, package, and capacitance on DC bias capacitance loss. Their "normal" MLCCs is starts with GRM just to trim down on the search results a bit. X7R caps aren't great past 30% DC bias but X5R is even worse unless you go with enormously oversized SMD packages for the capacitance value or voltage rating.

If you need accurate capacitance, do not use above 10-20% of the rated voltage. Capacitance drops like a rock beyond that (-30% or more)

https://www.murata.com/support/faqs/products/capacitor/mlcc/char/0005
the above link explains this phenomenon. it appears this doen't happen to C0G type caps.