capacitor gets hot in a HF-HV resonant circuit

On Mon, 26 Oct 2015 17:05:51 +0100, blisca Gave us:

Flip the transformer. A one turn primary and stepped up into a two turn secondary.

What you described is a step DOWN. For that to generate 100V, the primary has to generate 200.

So maybe you got it ass backwards.

At a guess it's the dissipation factor in the cap (ESR). New NPO ceramic's are pretty good, but basically about the same as polypropelene. (Some better, some worse.) You could try a series parallel combo of a bunch of caps, keeping C the same but with a lower overall ESR.

George H.

Thanks George my doubt is that it is not an ESR matter,but dielectric absorption. Could one say that are 2 ways to describe the same phenomenon?

My doubt about it comes frome the fact that using 2 paralleled 1/2 value caps results at oscilloscope in a ringing of the same amplitude. I expected that lowering the series R would lower the wasted power,leading to a ringing more wide in amplitude and slower in decreasing

Diego

I'm confused. Your "few tenths of nF" (say, 300pF?) at 100Vpp, 150kHz shouldn't have anything near this current / dissipation with good caps like polypropylene or NP0/C0G. Do you really mean uF? And is the 48degC a final temperature, or a temperature rise?

polyester caps will have large losses at 150kHz, and as they heat up it will only get far worse, not to mention the capacitance will drift:

polypropylene may be your best bet, if the units you are are able to handle the current. Metallized film only caps will have a higher ESR and are more likely to self-heal or just burn up. More caps in parallel can help too in dropping ESR if your dielectric is suitable in the first place.

es

he

No, dielectric absorption is like memory in the cap, it remembers it's last charge state. (It tends to "recharge a bit... at least that's my understan ding I've never really bumped my head into DA.. DF on the other hand is an old friend.)

g Hmm it's hard to know what that will do... you could also put 2 higher values in series.. in both cases I would guess about the same ESR. But try four of the same cap, in series and parallel p air.

Oh $hit, scratch that. It's still the same. (Idiot... talking to myself.)

George H.

Got a schematic?

Several people, like ATC and Dilabs, sell low ESR capacitors, designed to not unsolder themselves at high RF currents.

You might parallel a few caps and heat sink their end caps with PCB copper pours. That would help a lot.

What are their dissipation factors (DF) at 150kHz? If not in the datasheet ask the manufacturer to furnish that data in writing. If they can't then do not use their capacitors. If the DF is more than a small fraction of a percent they are not suitable for a resonant or power filter application.

Try ceramic NP0, my experience with those is quite good. For example, these are 0.1% in the dissipation factor or DF and cost around $0.50 in quantities:

If that's still not enough split them up into many smaller parallel caps. At 100Vpp you may have to do that because this is a very high amplitude.

As one SW engineer put it, we are here to serve :-)

Sorry for generate confusion This was only due to my lack of masterhood of English language. It should be "tens of nF" ,in practice 47 nf with 2 turns

thanks again, and forgive me for wasting your time with that error

You're seeing a combination of dielectric losses and ohmic heating of the capacitor plates, Both can be substantially reduced by using a group of metal foil (not metalized film) polypropylene capacitors in parallel to reduce the effective ESR of the capacitors. This construction is often used on high di/dt polypropylene snubber capacitors.

And lets not forget Johanson. Kemet has some good NP0 caps as well. So does TDK but their datasheets aren't very good and that is why I often cannot spec them in (datasheets must back up assumed or simulated performance).

Yes. But best if they don't dissipate much in the first place because that can make a deliberate ringing event less predictable from unit to unit or between production lots.

or look at what people are doing for induction heaters

-Lasse

They use surprisingly cheap looking polypropylene caps that are engineered to be just good enough. No question, they're custom or were designed for induction cooktops. My "1800 watt" cooktop hasn't died yet but the first tear down showed what is just two (possibly hybrid foil/metallized) caps. I was surprised at how cheap everything inside looked and still amazed it works reliably and without catching on fire, or having a power factor of

That seems like a non-problem unless you have some kind of serious power constraint/limit/battery-life issue.

es

he

The ringing amplitude reflects the energy stored in the capacitor-inductor tank circuit.

The rate of decay of the ringing reflects the losses - heat generation - in the series resistances of the capacitance and the inductance.

You've got to have a very heavily damped system for the resistive losses to have an immediately visible effect on the initial amplitude of the ringing .

USA automotive engineers most likely :)

Jamie

Il 26/10/2015 18:28, John Larkin ha scritto: > On Mon, 26 Oct 2015 17:05:51 +0100, blisca wrote: > >> Hi >> >> i was asked to help some friends in a task that is unusual for me: >> I have to generate a 100 Vpp -150 KHz ringing on a single turn of a >> transformer secondary. >> >> >> I generate an adjustable PWM with a microcontroller,then a driver drives >> a N MOSFET connected to VDD via a couple of turns(the primary of the >> transformer),paralleled by a capacitor** >> With a pulse width of about 7us i have the necessary overvoltage and the >> desired ringing,but the resonant capacitor gets hot. >> >> >> >> >> >> **In series to the MOSFET drain i put an ultrafast diode,so to allow the >> ringing going below zero volts without being limited by the intrinsic >> MOSFET diode >

Once more,thanks to everyone for your time,your precious help and your tips.

Yes John,i used exactly this basic schematic

The choice of the MOS BUZ41A was just because i have a bunch of these MOSFET in the drawer.

The green trace in this pic is the current,the yellow trace is the gate pulse(is very steep,maybe was taken before the MOS driver,i'm not shure).The waveforms are similar even for a longer drive pulse and a ringing of 130V.

The initial narrow green pulse is the initial charge of the capacitor,when its lower armor is suddenly pulled down to ground from

I don't think this pulse is determinant to warm the capacitor,because if i drive the MOS with a wider gate pulse,the capacitor warms more, despite the pulse remains the same

I think that the apparently linear ramp shows that the transformer is not saturating

The purple trace is the output on the single turn secondary on a 10M probe. In the second pic the probe was erroneously connected reversed,normally i use it in a way that the output is in phase with the drain voltage

Yeah, they're forced to be cheap but come up with some really clever, refined designs, packaging and processes. But of course, some cars still have electrical fires when parked in the garage and turned off.