X&R ceramic capacitors for DC Filtering

[mrinalmani]

I need to install a capacitor in series with the primary winding of a high frequency transformer working at around 100 KHz.
The purpose is clear, to prevent core saturation due to flux walking.
The calculated value comes out to be around 10 to 15uF and 16V.

NP0 capacitors are expensive for the values given above. I am thinking to use X7R instead.
However, some vendors prohibit the use of X7R in AC configuration, while others provide softwares which show the maximum AC current allowed through a capacitor for a given capacitance and case size. The only restriction imposed is due to the temperature rise.
Does anyone have experience with X7R capacitors used in AC application. Are they safe for AC?
Please share our experiences...
Thanks!

 

[FvM]

I would rely on Murata. They allow AC without DC bias, Vpp up to capacitor rated voltage. At high frequency current rating is the effective limit.

 

[D.A.(Tony)Stewart]

The key to your application is ripple current at switch rate and the impedance of ESR must be much less than the Capacitance impedance of 0.1 Ohm at this frequency.
How much ( Wechselstrom) ripple current do you need?

Hopefully not 16V/0.1Ohm? not a chance
hard to find 1 mOhm in reasonable budgets.

But for $7 this one comes close https://www.digikey.com/product-detail/en/B32794D2156K/495-4136-ND/2182154 to possibly working. with an SRF of >200kHz @ 15uF 3 mOhm
13 Arms MAX @60'C

I think Epcos design is best with self healing so it wont melt into a pool of plastic.

 

[mrinalmani]

Current requirement is nearly 100A.
A parallel combination may help.
I am attaching the temperature vs ripple current graph of a Kemet 4.7uF, 16V, X7R, 1210 Case
The ESR is nearly 6mOhm. The capacitor costs around 0.3$ @ 1K.

Is the ac capability vendor specific or the very X7R dielectric should be avoided for ac. (Perhaps due to internal corona)

FvM
Thanks, but even with Murata, what can be said about the reliability of X7R in AC. Have you seen it being used in such applications with acceptable failure rates?

 

[FvM]

As long as you don't know a specific failure mechanism in pure AC operation, it's pointless to guess about reliability. Can you refer to vendor literature that discuss a respective issue for X7R or X5R?

Regarding current rating in parallel circuit, you should calculate possible ESR mismatch and stay below a maximum rating. I have used X7R in ripple filter of 4-quadrant power amplifiers which can be operated as current source into a DC short, in other words with no capacitor DC bias. I didn't become aware of any problems. The actual AC current in this application is however considerable below thermal limits.

My feeling is that you won't get problems if the capacitors are used with 20 or 30 K temperature rise. But I would avoid low voltage AC coupling in the said application and use DC current control if necessary. Which also means that I would primarly check if there's an actual possibility of "flux walk" and transformer saturation. The very most of inverter circuits are self balancing with quite a little output dead time.

 

[D.A.(Tony)Stewart]

A 1210 ceramic part is rated for 500mW max at 70'C.

The 6 mΩ X7R cap in same size at 100Arms will dissipate 60Watts and fuse open in milliseconds. Thus you need 120 parts in an array of 12S10P to run with a 50'C rise above 20'C!! And would cost 120x0.3$=$42

The 3 mΩ Polypropylene cap would dissipate 30W at 100Arms and melt as that is 770% of the 13 Amp rating.

P=I^2 R
Back to the drawing board with your design.

What is the goal? In, Out?

 

[FvM]

The 6 mΩ X7R cap in same size at 100Arms will dissipate 60Watts and fuse open in milliseconds. Thus you need 120 parts in an array of 12S10P to run with a 50'C rise above 20'C!!

??? Please observe the square in P = I²R.

 

[mrinalmani]

Oh! I think there's a slight miscalculation!
Wit 120 capacitors, the equivalent ESR would be 6/120 mOhm and the resulting power loss would be 0.5W distributed across those 120 chips!!! (Nearly 5mW per cap)
I think 7 or 8 in parallel should be just enough with good thermal management.

FVM
Thanks! I dont suppose I know the reason why AC was not recommended for X7R. You're right... about the current mode control, but I was just wondering whether the capacitor method would be practically feasible. I think I'll prototype a board with a few 1210 X7R in parallel and post the results.
It is also surprising to know that most transformers are self balanced. Maybe probably due to low operating frequency.. 25KHz to 50KHz or so and also due to higher DC resistance of primary and the MOSFET.
In my case the net series resistance of MOSFETs + Primary is only 3mOhm. Even a 20ns mismatch would result in nearly 10A of bleeder current.