Modification of CRC filtering

[DSP_Geek]

Part of the problem with electrolytic caps is the equivalent series inductance (ESL), which as I understand it can reach close to 1 microhenry for large models. Placing a smaller film cap in parallel isn't such a good idea either, because if the electro's series resistance is low enough then the film cap will create a resonant circuit with the electrolytic's ESL.

High ESL is a bit of an annoyance since rectifiers create crud all over the spectrum, and it decreases filtering effectiveness in the high frequencies not far from where amplifiers start losing power supply rejection ratio (PSRR).

The Pass people turned me onto CRC filtering, which reduces ripple rather nicely, and as it turns out I have two resistors in series on each leg, which led me to an idea: why not try a CRCRC network, albeit with the center capacitor a 10 or 20 uF film type? It would be isolated from the behemoths on each side by the Rs, thus killing the Q of the potential resonant circuit.

I haven't tried it yet, but so far the math looks good. Comments?


Cheers,
Francois.

[anatech]

Hi Francois,
You can put snubbers across the rectifiers and transformer secondary while you are at it. I would tend to use a film cap to decouple later (the last "C") and local at each stage.

Now, carlosfm can tell you about snubbers. In essence, you want to get rid of the high frequency content as early as possible, before it gets out in the wild. The later film caps ensure a low supply impedance across the bandwidth of the circuit (not just the audio band).

-Chris

[DSP_Geek]

Originally posted by anatech
Hi Francois,
You can put snubbers across the rectifiers and transformer secondary while you are at it. I would tend to use a film cap to decouple later (the last "C") and local at each stage.

I'm cool with the snubbers, but there might still be crud from the edge of the rectified waveform; don't the harmonics go up to several hundred kHz? I'm also a bit leery of paralleling films with big electros because of the resonance issue, and since I have a convenient pick-off point between the two resistors it's not a Big Deal to put the films there instead of directly on the main filter caps.

As for local bypassing, I presume the UcDs have that in spades; I just want to make sure there's no crud getting there in the first place.

[Eva]

This topic is very prone to mystification. The best recomendation that I can make is to build real prototypes and check waveforms. Don't trust popular wisdom at all.

In that thread you may find some hints about diode behaviour, particularly starting at page 7:
Fast Recovery rectifier diodes

[DSP_Geek]

Originally posted by Eva
This topic is very prone to mystification. The best recomendation that I can make is to build real prototypes and check waveforms. Don't trust popular wisdom at all.

In that thread you may find some hints about diode behaviour, particularly starting at page 7:
Fast Recovery rectifier diodes

The other thread was quite illuminating. Thank you very much.

[anatech]

Hi DSP_Geek,
Eva is so right. Read and experiment to confirm.

Film caps have not created a resonance problem for me to date, but then again I generally look at the supply with a 'scope to make sure.

-Chris

[DSP_Geek]

Originally posted by anatech
Hi DSP_Geek,
Eva is so right. Read and experiment to confirm.

I happen to have a 100 MHz scope on loan from the job since I work from home. Thanks for the tips, folks, I'll try this out in my copious free time.


Cheers,
Francois.

[mzzj]

Originally posted by DSP_Geek
Part of the problem with electrolytic caps is the equivalent series inductance (ESL), which as I understand it can reach close to 1 microhenry for large models. Placing a smaller film cap in parallel isn't such a good idea either, because if the electro's series resistance is low enough then the film cap will create a resonant circuit with the electrolytic's ESL.

You need real luck to find something with ESL of 1uH. 100nH might be closer shot for large screw-on caps, special low-inductance screw-on electrolytics as low as 30-60nH

[janneman]

If you want to use a low-ESL cap to clean the supply do it as close as possible to the circuit-to-be-powered. In that case, the wiring ESL acts for you; you have build yourself an LC filter! The load, across the film cap, damps any oscillatory tendencies of the LC combination as well.
Jan Didden