I always wondered how I could reverse engineer and figure out DC bias of a random MLCC I found on a board. I have my answer, I take this random capacitor and place it in a LC filter and do the same measurement as you did. Sweet!

I went to school for electronics when a 10uf ceramic was considered as rare as a tantalum unicorn. I'm learning a lot about these modern components lately and this info is really helpful, thanks! - The stuff they make now is pretty incredible but it has limitations, , as usual.

thanks. did not know the effect of voltage on capacitance of ceramic cap before

Awesome stuff !

Very clear demonstration.

Can't thank you enough for such nice explanation

Guess I learn something vital every day. No more ceramics in my audio signal chain

OK so you have highlighted a problem, but what is the solution to this problem???

This is only not true for C0G (NP0) capacitors, which are extremely stable and often perform better than film capacitors. Speaking of film capacitors, they have a frequency vs voltage graph. Beyond a certain frequency the permissible voltage drops rapidly.

It's good that you're highlighting this important, and perhaps surprising phenomenon. However, calling it "DC bias loss" makes no sense. How is this about "losing DC bias"? It's not. It's "capacitance reduction with increasing DC bias". And the surprise is that many capacitors are spec'd to possess a capacitance at their rated voltage of only 20% of their 0V capacitance (similar to the result you obtained). Given that many MLCCs are used as decoupling capacitors for power supply rails, and thus always operating with a significant bias, it's crucial that the circuit designer understand the C vs V curve for such a capacitor, and choose a cap with a voltage rating that's a substantial multiple of the actual voltage in the application -- or a cap with a larger capacitance, of course.