Capacitor soakage

Did you read Bob Pease's article? A bunch of RC's in parallel with an ideal cap should do it.

--sp

The actual performance of examples will vary from one to another, from batch to batch and even with the cap's particular history.

Mad to assume they are all the same.

Leakage is just as variable too.

Try polystyrene is you want to minimise that.

..... Phil

Yes I am aware of the variability but this is an extreme application where individual systems are hand adjusted to get the desired response.

Yes I have read the articles by Bob Pease but I need to be able to measure and determine the equivalent circuit for a given capacitor so that compensation can be applied. Each unit is hand adjusted. Scott Hamilton.

Yikes, that sounds like nailing jello to a tree. Do you have any reason to suppose that all the internal time constants would be temperature-independent? Cuz otherwise that's a nasty calibration table to create.

Cheers

Phil Hobbs

Is there any reason to believe it will be linear with the voltage step? (Sounds like a research project.)

I thought I read once that teflon caps had low soakage. Maybe they are better than polypropylene?

George h.

Film caps are all pretty good--none of the common polymers is ferroelectric, so they don't saturate like titanates and so forth.

If I were doing it, I'd probably use a dynamic signal analyzer to measure the transfer function of an RC lowpass and let it run over the weekend.

Cheers

Phil Hobbs

I'd time-domain it. Charge overnight, short briefly, log the open-circuit voltage for a day or so. You'd need a low-bias-current DVM or follower opamp.

There could be a huge range of internal tau's.

Someone here once posted long-term (months and years) leakage of some charged film caps.

I'd sort of expect leakage and tempco to make the analysis harder that way, though.

Cheers

Phil Hobbs

My dim recollection is that teflon is not so good. It wouldn't be surprising - fluorine having such strong electronegativity would likely have a "polarizing" effect that could be subject to relaxation responses.

Mylar is definitely bad; polycarbonates not great either. Polypropylene is good, perhaps nearly as good as the ever-meltable polystyrene.

The ceramics are mostly mediocre (or worse), though some of the newer C0Gs are close to polypropylene.

Ah yes... got me to look it up in AoE3(p301): mostly what I've been sayin', except for the PTFE! This may be another case like the C0Gs - i.e. dependent on manufacturer/unknown parameters.

Film cap charge soak does vary with dielectric - polyester gives the worst charge soak, and polycarbonate isn't much better. Polypropylene is roughly an order of magnitude better. Telflon is reputed to be better still - but hard to get hold and expensive.

Polystyrene is supposed to be good, but offers a lot less capacitance per unit volume.

Well, yes, but wouldn't it be better to shell out the cash for something that has minimal problems _first_, and _then_ go and hand-optimize it?

And -- could you do better with a Really Good ADC attached to a Really Good voltage reference and a freaking plain-old-garden-variety microprocessor, that's talking to a Really Good DAC?

Because I suspect that, within the confines of your Really Good Cap and the whole compensation network to make it Really Better, you could fit all the stuff that I'm talking about and have room left over for a small video game.

Would it not be easier to DAQ everything and implement the filter with a sufficient number of DSP bits?

Even with high dynamic range (>20 bits), if the bandwidth isn't terrifying (sub kHz?), it starts to look a whole lot better...

Tim

Some of the newer dielectrics that can withstand reflow soldering temperatures are not too bad. PPS (Polyphenylene sulfide), for example.

--sp

did you check the Wiki on DA ?

There is a link there on modelling DA.

Looks like what you are after.

..... Phil

My copy of AoE3 is at work, I've never used teflon so I have no idea in practice. (The one time I went looking for teflon caps, they seemed to be part of audio-foolery and over priced.) Yeah, there are some "good" C0G's? I seem to recall the B. Pease plot with some good and bad ceramics. All the new cog/npo that I've used have been fine wrt DF, loss in some LC resonant circuit. That might, or might not, relate to DA.

George H.

• I thought i read that Teflon(TM) was rather good at holding a "reserve" charge after discharging, similar to glass (viz: CRT). An air or vacuum capacitor may be near the best WRT "soakage".

Check out the figure on page 301 of AoE3. We need Win to tell us which COG cap is brand X and which brand Y...

George H.

Leakage is just one more component of the circuit model, the 1st order time constant. But a good film cap apparently leaks a fraction of a per cent per year.

Capacitance tempco will change the voltage, so I guess it should be tested at constant temperature. But then the OP will have to run the actual thing at a constant temp if he expects to compensate to the measured accuracy.