PSU modding - adding better capacitors. Good idea? Bad idea?

[Jeff7]

I've got a few cheapo PSU's here, and I notice that their filter capacitors are outright puny compared to those in high quality supplies. Would it compromise the circuit's functionality to solder in capacitors with higher capacitance? Or would it improve the integrity of the power output?

 

[ronach]

If you have the electronic knowledge and skills to do the job, go for it. If this is the case, you more than likely have built yourself a load bank to test out your modifications, rather than use your computer as a dummy load. Am I making any sence here. ronach

 

[KillaKilla]

Originally posted by: Jeff7
I've got a few cheapo PSU's here, and I notice that their filter capacitors are outright puny compared to those in high quality supplies. Would it compromise the circuit's functionality to solder in capacitors with higher capacitance? Or would it improve the integrity of the power output?

Will it improve the PSU? yes. Is it worth it to risk your entire computer over $30?

 

[Lonyo]

As ronach suggested, buy some cheapo stuff to test the PSU's first, so you don't break your "real" computer.
You can also use the stuff as a test rig if oyu have any issues (like RAM messing up, graphics card messign up etc) and run it as a router/firewall to protect oyur main PC.

 

[Mark R]

Which capcitors are you interested in?

The mains reservoir capcitors (the biggest ones) - these are high-voltage and under relatively low stress, ripple here isn't a desperate problem, their most important function is to keep the power stable in case of a transient loss of power. Unless your mains power is really awful, I probably would leave these, as these are expensive and replacing them is likely to give only little benefit.

The output filter capacitors - these are under high stress, with high ripple current. Some of these may be inside the regulation feedback loop of the PSU, so some care is required in selection of replacements in case they destabilise the feedback loop. If you improve ripple current rating and ESR as well as increasing capacitance, you shouldn't run into problems. Simply swapping in higher voltage rated capacitors will usually improve ESR at the same time (at the expense of a larger physical size). Whatever you do, don't replace the capacitors with ones with higher ESR - this will degrade regulation quality and phase margin in the feedback loop.

There are potential benefits for upgrading these capacitors - increasing capacitance can decrease ripple voltage and improve transient response. Decreasing ESR can improve voltage regulation.

[Edit - snip safety advice of dubious relevance to the present task]

Another potential gain can come from replacing the diodes in the PSU - again, I don't mean the high voltage ones. I mean the low-voltage rectifiers after the transformer. These are often packaged as a pair of diodes in a 3 pin package looking exactly like a transistor, and are attached to a sizeable heatsink. Replacing these with high-grade rectifiers (e.g. dual 50A SMPS rectifier Schottky diodes) can reduce heat and potentially improve regulation.

 

[VirtualLarry]

Originally posted by: Mark R
Also, some care is required in disposal of the old high voltage capacitors, if you choose to replace these. There is a rare phenomenon, where capacitors which have been charged to a high voltage for a long period of time, can reaccumulate a charge if left disconnected, even if they have been fully discharged. To prevent this - solder a wire to both terminals to keep the capacitor shorted out.

This phenomenon really intrigues me. You seem to know what you are talking about in the rest of the post, so I'm going to nominally believe you here, but this seems incredible - could a bank of these "specially modified" capacitors, be used to somehow pull charge out of the air? That sounds almost like some form of perpetual-motion alternative energy source, if it could be harnessed. Can you provide any references to this phenonmenon?

Alternatively, could this simply be a mistaken observation, resulting from working on a system with a failed bleeder circuit hooked up to the mains capacitor bank in a PSU?

 

[sunase]

He's probably referring to soakage, this was the most interesting page I could google on it:
http://www.national.com/rap/Application/0,1570,28,00.html

 

[Mark R]

Originally posted by: sunase
He's probably referring to soakage, this was the most interesting page I could google on it:
http://www.national.com/rap/Application/0,1570,28,00.html

Yup, that's it. Although reading the article, it looks like its of no safety relevance in modern caps.

I'd never known the name of the effect, it was just something my mentor had once told me as part of an old war story.

He used to do some military work - so presumably, when he meant 'high-voltage' he meant 30,000V not 300.

 

[Jeff7]

Originally posted by: VirtualLarry

Originally posted by: Mark R
Also, some care is required in disposal of the old high voltage capacitors, if you choose to replace these. There is a rare phenomenon, where capacitors which have been charged to a high voltage for a long period of time, can reaccumulate a charge if left disconnected, even if they have been fully discharged. To prevent this - solder a wire to both terminals to keep the capacitor shorted out.

This phenomenon really intrigues me. You seem to know what you are talking about in the rest of the post, so I'm going to nominally believe you here, but this seems incredible - could a bank of these "specially modified" capacitors, be used to somehow pull charge out of the air? That sounds almost like some form of perpetual-motion alternative energy source, if it could be harnessed. Can you provide any references to this phenonmenon?

Alternatively, could this simply be a mistaken observation, resulting from working on a system with a failed bleeder circuit hooked up to the mains capacitor bank in a PSU?

In my online travels looking for information on monitor repairs, I found mention of this with respects to the CRT, that even a "discharged" CRT can re-accumulate charge. I guess it might be similar to how a "dead" battery, when left alone for long enough, can put out enough charge to power a device for a little longer.

Whatever you do, don't replace the capacitors with ones with higher ESR - this will degrade regulation quality and phase margin in the feedback loop.

Ah, the great low-ESR capacitor problem, which I encountered when re-capping my 8RDA+.