Filter cap choice for ultra low ripple

That is not strictly true. On the input of a switched cap PSU the ripple is current. That will transform into voltage by the impedance of the input supply, but the cap provides some of the current to reduce the voltage ripple. A lower impedance of the input supply only improves the situation, not worsen it.

On the output, the cap can be filtering the ripple from either the input ripple residue, or the output ripple if a switched converter. Again, a lower output impedance is better and a filtering cap will be taking advantage of that impedance.

Any active circuit that acts as a filter has to be effective at the frequency of interest. It is a common misconception that you can add a linear regulator to the output of a switched converter to get rid of the ripple. The frequency of ripple is often well outside the effective frequency range of the linear regulators.

On 09 Aug 2015 18:07:48 GMT, joe hey Gave us:

Dude, the pommy retard has been stalking and trolling me for years, so STFU about that which you are absolutely clueless please.

On 09 Aug 2015 18:19:15 GMT, joe hey Gave us:

Fuck you, you obtuse, know nothing about this group asswipe!

On 09 Aug 2015 18:19:15 GMT, joe hey Gave us:

I should 'sue' you with the penalty being that your ass ends up out in the middle of the desert buried up to your neck right next to an ant hill. Lets pour a bit of honey on top of your head as further 'suppers ready' enticement for them.

For someone who has not frequented the group, you sure are one mouthy little bitch.

On Sun, 9 Aug 2015 13:47:44 -0500, "Tim Williams" Gave us:

Isn't the term you were searching for 'PARD'?

On Sun, 09 Aug 2015 11:58:12 -0700, Jim Thompson Gave us:

Jim Thompson is an absolute idiot who should assume nothing, since he can't even put together what the fastest currently in use CPU is, despite google being at his mundane brained fingertips.

Absolutely. If the impedance of the source is increased, it has to be lowered again with more input capacitance. This is often helpful, though, to get rid of conducted emissions.

The output ripple is current too. The same issues apply. After the regulator (after the output capacitors), some impedance my help filter noise, though. Often a ferrite bead (and ceramic cap) work well.

That's generally true but there are wideband "regulators", if you're really worried about such things. Also, while the ripple may be outside a linear regulator's bandwidth, switching regulators tend to be noisy. A linear regulator often helps.

what a fopdoodle

Jamie

Troll question but anyway high capacity chemical batteries can offer very low noise. For example a 12v car battery. The leakage current in an electrolytic capacitor can be a source of noise. Probably a battery is better.

Output buffer amplifiers (transistor follower type) are insensitive to power supply ripple and noise. Other parts of a 'signal generator' may need any amount of current, or voltage regulation, or isolation from other stages. So, it's too complex a problem for 'choice of filter capacitor'.

In addition to the actual filter/regulation, you really need to watch out for the PCB layout. Make sure that the ripple current only flows in the loop of secondary, rectifier and (first) storage capacitor. This means that the rest of the circuit (both positive and ground) must be powered directly from the capacitor terminals.

Getting the regulator input from capacitor terminal and the ground return to somewhere on the PCB ground track between the rectifier terminal and ground capacitor terminal, now the ripple current is flowing through a common ground resistance, causing all kinds of hum problem.

With additional regulation and filter sections, take the signal ground PCB track at the output at the last filter stage, thus all ripple and filtered noise will be circulating in the secondary, rectifier, storage capacitor, regulator, additional filter loop and not pollute the actual signal ground.

You continue your history of under-specifying. Specs have numbers. "...as humanly possible" is not a number, or even a useful concept.

Bottom line is that you can't afford enough caps to solve a ripple problem. It's a system issue.

Back in the day, a bunch of us built TEK spectrum analyzers from a big box of rejected parts. One of my tasks was to build a current source to screen YIG oscillators. Boy, was that an eye-opener. Turns out that noise comes from everywhere at that level.

I'm surprised Phil Hobbs hasn't jumped in here with his beloved capacitance multipliers. But perhaps, since he's just been talking about them and you haven't taken notice of that, he assumes that you're just not a good listener?

In a nutshell: filter with whatever, then regulate your way to smoothness.

On Mon, 10 Aug 2015 18:21:15 +1000, Clifford Heath Gave us:

You obviously do not know how to read a thread.

Since the same charge needs to be transferred during the mains half cycle, using huge storage capacitors will result a smaller conduction angle and hence large peak currents. High peak current will cause many problems, including strong stray magnetic fields that can be coupled into various circuitry.

If low noise and not maximum efficiency is the goal, use moderate size storage capacitors with possibly additional series resistance and the peak currents remain small. Of course, this requires higher secondary voltages, but the increased peak rectified voltage can be limited with any conventional regulator.

You are an utter imbecile.

You really shouldn't flatter yourself, Nymbecile old girl. Do you really consider yourself worth stalking?

Well done Joe, you scored a 3'er.

Yes, he is very funny, almost cute. :)

joe