Filter cap choice for ultra low ripple

On Sun, 9 Aug 2015 14:27:50 +0000 (UTC), Cursitor Doom Gave us:

In general terms.... You're an idiot on par with SkyTard Belching.

Yeah, and you're a brainless troll with a long and very well-established track record.

Capacitance as large as humanly possible.

Roughly what frequency is the ripple at? Mains, 50kHz, 2MHz?

Polymer electrolytics are pretty good, as are ceramic caps. More stages are better than fewer, and you can always use the cap multiplier gambit. Regulators with few-uV RMS noise are available.

--sp

On Sun, 9 Aug 2015 15:13:22 +0000 (UTC), Cursitor Doom Gave us:

I am not a troll, and I have orders of magnitude more electronic knowledge than you do.

What's the matter, dumbfuck? Does the truth hurt?

Filtering ripple takes more than a simple "filter cap" or "filter cap choice", you retarded putz.

You sport about the same 'intelligence' about electronics as Donald Trump does on US history, or foreign policy or political posture. He is an absolute idiot, and you are right there in the same class.

You and "a high end signal generator" are two things that are mutually exclusive. So your grasp of what internals are needed in one are pretty far apart as well.

Particularly if you are so dumb (in electronics)that you need to come in here and ask such a question, which reveals just how ill equipped you are for completing any such design task.

And as far as my "track record" goes, it only puts the pinch on abject idiots such as yourself.

Go somewhere else and spread your stench.

In here you are transparent, and you asking such a low end, poorly worded revealing question makes you quite transparent, and obviously not equipped to tackle such a complex design as a "high end signal generator".

So yeah, punk... you are on par with the SkyTard, who has been invading several newsgroups with his utter stupidity.

And that is despite the fact that your question is on topic here. It simply exposes such how simple your skill level is. Quit while you are ahead., or your 'ultra low' profile of stupidity will get raised. ooops... too late.

Ripple is determined by two factors. One is the capacitance relative to the amount of ripple current flowing and the frequency. The other is the ESR of the cap relative to the peak current.

Ceramic caps have the lowest ESR of caps commonly used for PSU outputs. Electrolytic caps provide more capacitance at a lower price. Sometimes the best tradeoff is a combination, other times ceramic caps alone will do the job well enough. The only ways to know are to do the calculations or to build it both ways and measure the ripple. :)

Doesn't Pan offer filtering? In Agent I just killfile DecadentLoser so that I'm not exposed to his (or other loser's :-) rants.... keeps me calm and my blood pressure low >:-}

The only time I see his rants is when you feed the troll :-( ...Jim Thompson

If you can afford a volt or so of loss, a 2-stage cap multiplier is the ticket. You can get way over 120 dB rejection of SMPS frequencies.

Cheers

Phil Hobbs

You are an imbecile.

On Sun, 09 Aug 2015 08:50:41 -0700, Jim Thompson Gave us:

There was no rant, you absolutely retarded f*ck.

On Sun, 9 Aug 2015 09:03:07 -0700 (PDT), Phil Hobbs Gave us:

A "high end signal generator" would likely have a linear PS in it, not a noisy SMPS.

But you should know that, Phil.

On Sun, 09 Aug 2015 17:07:46 +0100, Pomegranate Bastard Gave us:

F*ck off and die, pommy bitch c*ck sucking retard.

Decadent Linux may be extraordinarily rude, but he is right that there is no answer to this question - we would need a lot more details to give useful answers. Of particular importance are the frequency of noise to be rejected, the type of currents and voltages involved, any unusual circumstances or restrictions (such as wanting to be as small as possible), and some information about the rest of the PSU design.

And a capacitor alone does not make a filter - you need it in combination with something else (an inductance or resistance), and usually in multiple stages with multiple filter components.

With that information, someone should be able to give you more helpful answers. (Not me, probably - it's not really my field.)

Back in the day, that would have been the case. In the last 20 years or so, we've had to adapt, though, and cap multipliers are a huge part of that.

Many of my protos are powered by random laptop bricks via cap multipliers.

Cheers

Phil Hobbs

On Sun, 09 Aug 2015 19:25:14 +0200, David Brown Gave us:

The term for the day is "envelope tracking".

Not a postal service term.

Jesus Christ, this guy really goes nuking around for even the slightest of reasons...

joe (quickly hiding in the cabinet)

(O.T.)

He reminds me of a quite (electronically) brilliant Finnish guy who got poisoned from his mercury amalgam fillings, became 'as mad as a hatter', tried to sue the (Finnish) government and the medical establishment, did probably some other things wrong too, and finally had to leave the country due to harassment/persecution and went to the USA with his electronics knowledge, where he continued ranting around where and whenever he got the opportunity and at the slightest provocation.

It seems that mercury poisoning does that to your brain/constitution.

joe

Generally one buys enough filter cap to get the ripple down to maybe 1 volt p-p, then adds a regulator to stabilize the voltage and greatly reduce ripple. A second regulator or c-multiplier can be used to further reduce noise and ripple, to power the more sensitive signal generator stages.

Best?

All things considered, same as always, electrolytic. Low cost, high value, moderate to low ESR, reasonable life (when not run hot).

For any other considerations, it's up to you. Cost? Operating frequency? Size? General *considerations*, yes, but no generalized *conclusion*. That's what us designers get paid for. ;-)

Even more generally, however: you seem to dismiss system-level considerations as obvious! But no, a signal generator is a system. The power supply only needs to be as well-filtered as the PSRR of the output stage, preamps and such require for the desired output noise level. And if there's a regulator stage in there as well, even less (primary) filtering is needed.

And you can locally filter and regulate stages that are unusually sensitive, or rebuild them in terms of ratios and ground-referenced voltages, to increase their PSRR.

There should be absolutely no excuse for excruciatingly clean DC, not since tubes have gone the way of the dodo. Complementary semiconductors (and good constant current characteristics) free us from the burden of B+ referenced signals and low PSRR. :-)

Tim