MLCC Capacitors and my experiments of DC bias on Capacitance

[SteveyG]

I took a little look at how Class II Ceramic Capacitors change in capacitance with applied DC which some of you may find useful.

The video describes more, but essentially the capacitance decreases with increased electric field in the dielectric material. At the maximum working voltage of the capacitor, the capacitance may only be 20% of the originally specified value so something to watch out for in your design if you are not already aware of this effect.

C0G, NP0 (Class I) capacitors are not effected in the same way by DC bias.

I hope this helps someone

[laneboysrc]

Thanks for this informative video!

While I have seen already a few  other videos discussing this topic, yours is certainly standing out because it is very well explained and demonstrated.

Thanks for also measuring other capacitor types and showing their different characteristics 

cheers, Werner

[Siwastaja]

Thanks for spreading the word.

Most know the phenomenon as of now, but the details do matter, and #1 giveaway is: never assume any characteristic based on dielectric name; look for the actual curves. If the curves don't exist, and if the capacitance is of any importance, do your own measurements. For example, widely touted assumptions that X7R only go down to 40-50% with full rated DC voltage, and only Y5V are worse than that, are wrong. Some X7R parts are as poor as some Y5V parts.

[The Electrician]

TDK has a couple of videos on this topic:

[xavier60]

They can also lose capacitance with age,
https://www.johansondielectrics.com/ceramic-capacitor-aging-made-simple

[tautech]

Also discussed in some detail in the video in this thread:
https://www.eevblog.com/forum/chat/capacitors-explained-by-james-lewis-of-kemet/

[HwAoRrDk]

I find it a shame that capacitor manufacturers by default do not put these characteristic curves in their datasheets. It would be convenient if they did. Instead you have to go looking on the some obscure part of their websites, search by the exact part number, download a separate characteristic sheet, etc.

All the biggest manufacturers (Murata, Samsung, Kemet, etc) do make this info available, but I have yet to see it provided by the cheapest far-eastern manufacturers (such as you might find on LCSC, for example). If using these parts, you just have to assume the characteristics roughly match those of equivalent name-brand parts.

[Gyro]

They can also lose capacitance with age,
https://www.johansondielectrics.com/ceramic-capacitor-aging-made-simple

It's a shame that X7R and Y5V capacitance values aren't more stable with temperature and yes, particularly time. They would be handy for low cost, low frequency VCOs - outside the convenient range of varicap diodes (although some large junction diodes can provide a reasonable higher capacitance range).


P.S. I like your test setup SteveG, it demonstrates the effect well.

[splin]

TDK has a couple of videos on this topic:

Interesting; the second video, "Measuring  Capacitance accurately" shows a U1272A measuring a 100uF (93uF as measured with a precision bench LCR meter) MLCC as 66uF - approx 30% error despite having a spec of 1% + 2 digits.

This was attributed to the meter using too low a current for measuring larger,  >10uF capacitors. The discrepancy, 30% versus 1% specification is enormous so what is going wrong?  How could Keysight have got it so wrong, this isn't a $5 noname meter? Perhaps this error is specific to MLCCs, but that seems a far stretch. Something doesn't seem to add up here.

[The Electrician]

TDK has a couple of videos on this topic:

Interesting; the second video, "Measuring  Capacitance accurately" shows a U1272A measuring a 100uF (93uF as measured with a precision bench LCR meter) MLCC as 66uF - approx 30% error despite having a spec of 1% + 2 digits.

This was attributed to the meter using too low a current for measuring larger,  >10uF capacitors. The discrepancy, 30% versus 1% specification is enormous so what is going wrong?  How could Keysight have got it so wrong, this isn't a $5 noname meter? Perhaps this error is specific to MLCCs, but that seems a far stretch. Something doesn't seem to add up here.

At 3:00 the EIA standard measuring condition for 100 uF capacitors is given as: apply .5 VAC at 120 Hz

At 4:00 the bench meter shows that the applied voltage is 499.926 mV (close enough to .5 VAC).  To maintain that voltage across the 100 uF cap at 120 Hz requires a current of 35.1471 mA.  Turning on the meter's ALC function reduces the driving impedance of the meter so that it can supply that much current.

When the ALC function of the meter was turned off at 4:35, the meter's output impedance reverted to the nominal 100 ohms, and at that impedance, the meter could only keep the voltage across the cap at 89.0973 mV with a current of 4.9 mA.  That's what he meant when he said that the U1272A couldn't supply enough current (to maintain the voltage across the cap at .5 VAC).  This is not the required standard condition used by capacitor manufacturers, so you don't get the capacitance specified by the manufacturer.

If a 100 uF capacitor of a type that doesn't have a large voltage coefficient, such as a film cap, were being measured, the voltage applied to the cap wouldn't matter; you would get the same capacitance for any applied voltage--the U1272A wouldn't have a problem.

The lesson is, if you're testing wonky ceramic caps, you have to use the same measurement conditions as the manufacturer if you want to get the manufacturer's specified capacitance.  For large MLCCs, the U1272A (and probably most other hand-helds) can't do the job.