film cap test

John Larkin wrote: ============== >

** But kept secret about so you could troll everyone. Just like you have been doing for 25 years here.

Peeeeeeuukkkeeeeee !!!!!!!!!!!

Surface-mount film caps would have lower inductance, but I've always found them to be unreliable. They delaminate and wash water gets between the layers.

<snip>

So what's wrong with John Larkin's board-cleaning procedures? The industry as a whole presumably finds them reliable enough.

No they don't. They wish they could, but not so far.

SMT parts are stacked film, without any environmental seal on the sheared edges.

They need to be sealed to survive, so conformal coatings or encapsulation of the final assy is required.

RL

The post that started this thread said what I'm doing, namely building an LC filter in a switching regulator.

No secret there.

JL is a trolling s*****ad

=============================== >

** Like hell it did.

" We want some biggish film caps to use in some LC filters for a couple of switching power things "

** Reveals NOTHING. Exactly like ALL your bullshit posts. Wot a fake.

...... Phil

I've never used a SMT film cap on account of all the horror stories. I recall someone here posting a Digikey product page for one, with a photo that clearly showed delamination before it even got onto a board!

[Just checked--it was you, in a 2015 thread titled "delaminated caps", but the Dropbox links are (inevitably) dead.]

Cheers

Phil Hobbs

The problem was pretty clear. If the issue interested you, and you didn't understand what I'm doing, you could have asked.

The filter is actually interesting. We have a 60-volt 250 KHz 2-fet half-bridge and it needs to be lowpass filtered to make a clean 0 to

36 volt programmable DC power supply. We'll digitize the final output voltage and current and close the loops in an FPGA, which then makes the PWM into the fets. So the filter needs to really kill the 250 KHz ripple (without frying itself) but have low phase shift and low Q so it doesn't wreck the control loop. And it has to be made out of parts that we have or can actually buy now.

Turns out that, among other things, the Rds-on of the fets is an important part of the filter response.

We sim'd this with the LTC4444 gate drivers but prefer to use the TI UCC27712. We just tested that and it's excellent. Ignoring the data sheet, we just connect directly to the fet gates.

Here's my little half-bridge test board, before massive hacks:

It's regulating 200 watts and barely getting warm.

I don't know why people still sell these horrors. I've never seen one on a piece of production gear.

Here's a test of the c-v of a ceramic cap. We'll run at 60 volts as a supply bypass, so we'll get about 1.2 uF, which will be OK if we put a few in parallel with one big electro. That's smaller and less ESL than a giant film cap.

We still need the big films in the supply output filter.

Nothing. This is just another example of your anti-Larkin predujice showing through. It's time someone pointed it out to you.

Ceramic caps' C(V) curves are all over the map. The drudgery required to find a part that 'achieves' 70% of its rated capacitance at *gasp* 30% of its rated voltage has got old, old, old, old, and old. (Did I mention old?)

At this point, when we find a reasonably acceptable one, we hold our noses and do a lifetime buy.

On the plus side, the good manufacturers have websites where you can get reasonably accurate C(V) curves, but the amount of assing around is assssstrononmical.

Cheerssss

Phil Hobbsssss

It was a perfectly sensible question. And it has got a perfectly sensible answer

As legg pointed out

So it wasn't John Larkin's board cleaning procedures that were inadequate, but rather his failure to apply a conformal coating to what he clearly now knows to be a vulnerable part.

I ended up getting educated because I posted a sensible question. If you want to see a posting sensible questions that gets a sensible answer as an "exhibition of anti-Larkin prejudice" you've got your own kind of problem, and it might be that time someone pointed it out to you.

He reflexively declares me to be wrong about everything, because his emotions totally dominate his ability to reason. Or to use google.

Ignore him.

Legg pointed out that you should have used a conformable coating on your stacked film capacitors if you were going to put them through a water wash.

The reasonable point I was making was that you must have got something wrong if you found them to be unreliable. I don't make "reflexive declarations" and I certainly don't claim that you are wrong about everything - though your ideas about anthropogenic global warming are remarkably consistently wrong.

Blanket advice, of the kind you were objecting to. If you bothered to read what I actually post, you might produce less reflexive responses, but since what you seem to be looking for is uncritical flattery you are probably right to skip my comments.

That seems odd; you don't want to conformal coat before soldering, and you don't want to solder, then conformal coat before washing, do you?

I suppose you could install wire-wrap posts, clean the board, but conformal coat (the capacitors), then rely on wire-wrap attachment of the capacitors to insulation-displace the coating... otherwise, the conformal coating that matters has to be selectively applied before putting the parts into a reel. One hopes the capacitor manufacture process has been designed with this issue in mind.

Sounds labor intensive.

Just use radial leaded film caps.

When I wanted to see the innards of my brown 6.8u film cap. I put it on an anvil and hit it with a hammer. Nothing happened. The way to get inside was to put it in a big vise, lengthwise, and squeeze it until the shell shattered.

There seems to be an unofficial concensus that hi-K ceramic caps are are rated for the voltage that makes 10% of their rated capacitance. They can generally stand several times that voltage.

Yeah, through-hole film caps rock. They're really the only technology for applications needing good linearity at higher CV values than you can get in C0G. One typical use for us is filtering the bias supplies for avalanche photodiodes and SiPMs, but they're also good for slow analog ramps.

Cheers

Phil Hobbs