PWM DAC with electrolytik cap

[Genuigr]

Hey Guys,

I'll just leave this short question here:
Can I use an electrolytik Cap for a RC-Low-Pass in a PWM DAC?

That's it,

Fabian

[Rerouter]

it would depend on your frequency, electrolytics generally have moderatly high esr, of a few ohms, and this will offset your corner frequency,

[SeanB]

Yes, only if you are not worried about frequency stability, output voltage and phase stability or anything remotely temperature stable.

[Genuigr]

I want to use it at 10kHz with R=10k and C=10µF.
The cap is so high because an accurate value is more important to me than faster reaction.
I use the signal for 2 seperate op amps, just some comparating.

[SeanB]

10uF is in the range of a foil capacitor value wise, use that instead. Electrolytic is not good for a filter, as it drifts with age, voltage, how it feels. That is both in capacitance, leakage and it has a terrible dielectric soak. You would go to minimum signal and still find the output has a 100mV of signal there 2 minutes later.

[codeboy2k]

... it drifts with age, voltage, how it feels.

lol @ how it feels

[SeanB]

I had a piece of industrial logic that used 555 timers and electrolytics as timing. With a 10 minute delay on them I could never get a repeatable delay that was under 2 minutes either way. After one of the cascaded timers died I tossed it and used a spare 386 computer and a EGA card ( for the built in parallel printer port that had nice latched outputs and a buffer afterwards) to drive the outputs this controlled. Whipped up a quick basic program and refined it a little to get the resolution down and left it running. Ran for years.

[AndyC_772]

I may be a bit late here, but anyway....

Using an electrolytic cap for this application is a bad idea. Electrolytic caps have high ESR, so you'll always get some of your PWM signal appearing superimposed on the output of your filter. The output will never be quiet, stable or accurate - it'll always have this noise on it.

To see why, think about the equivalent circuit of the cap: it's like an ideal capacitor with a small resistor in series, and for an ordinary electrolytic this can easily be 1 Ohm or thereabouts. (There are plenty of other parasitic elements too, but they're less likely to be important).

The output of your filter will be the voltage on the 'ideal' capacitor plus the voltage across the ESR resistor. If your series resistor is 10k, and the ESR of the cap is 1 Ohm, you'll see 1/10001 of the PWM signal superimposed on your filter's output, which is (say) 33mV if your PWM signal is a 3.3V CMOS signal. That may be fine, of course - only you know that.

There will be series inductance too, which may become significant if your PWM signal is high frequency.

However, why not use a ceramic 10uF cap which will have much lower ESR and ESL? It could be 10mOhm or less, giving a much quieter output.

[poptones]

Why all this theoretical talk about ideal impedances? Every power supply in use today in our computers has electrolytic caps in it and this esr and such doesn't particularly hamper them even at high switching frequencies - that's why we bypass electrolytic caps with low value film caps! You can get low esr electrolytics, you can also get low da and df electros. Depending on the application, these parameters can be of as much detriment as the esr. What is the application? If it's non critical - you're not trying to measure signals in the tens of microvolts or dealing with surge currents of amperes - then a cheap electro would probably work just fine.