A few questions about recapping this particular PSU

[espresso]

Hi.
I have this 350W ATX power supply from 2005 and I decided to recap it and maybe boost it up a little.
Here is the schematic diagram for it. Many people here have the same PSU as I can see.


I have already replaced the two biggest caps (originally 330uF/200v) with 470uF/200v. Now I'm after the caps on the output rails and I have the following doubts:

1. Can I double the capacitance of C18, C22, C23, C25, C27, C28, C29, C32 without any consequences?
2. Can I use 6.3V caps on 5V rails? The original caps on 5V rails are 10V rated.

And a general question. Do C1 and C2 need to be low esr too?
I do not have much experience with switching supplies and the main reason I'm doing this is because I wanna learn something new.

[zandrax]

Re: A few questions about recapping this particular PSU

I'm not an expert and I hope more experienced people can reply better than me.

The psu name is LC-B250ATX: this is the name of a Deer, Allied or L&C psu. Those aren't good: bad or undersized caps are the minor flaw in them.

1. Can I double the capacitance of C18, C22, C23, C25, C27, C28, C29, C32 without any consequences?

My reply is: I think there are consequences: rule of thumbs is up to a 20% variation is acceptable, over that requires a complete check of all components.
Moreover there are limitations that may prevent you from recapping or may not be worth the hassle. First caps must fit (same diameter and pin interspace) and bigger capacity caps usually have bigger dimension as well; second some of those caps are on minor lines which aren't stressed at all and won't benefit from a capacity upgrade (C18 is on 5V STB, C29 and C32 are on negative lines which are required only by some ISA cards: unless you nan eed ISA card, you don't have to bother about negative lines). Third: supposed you find caps with appropriate dimension, maximum current is limited by FETs and transformers. If you want to draw more current from your psu, I think you'll be disappointed.
In short, a bigger capacity could help by increasing tensions' stability (longer holdup time and lower ESR) but you may have to change other components as well.

2. Can I use 6.3V caps on 5V rails? The original caps on 5V rails are 10V rated.

Yes, you can.

Do C1 and C2 need to be low esr too?

Not sure: IIRC high endurance is worth more than low esr, but I can be wrong.

Zandrax

[espresso]

Re: A few questions about recapping this particular PSU

Thank you for the reply.
My idea was to make the supply more stable by increasing the capacitance and to put less stress the motherboard caps.
I will manage to fit most of the capacitors. I already have the replacements in my stock.
As I understand there shouldn't be problems with increasing the caps which are acting as filters. The only thing that I would benefit is decreasing the ripple, right?

[dood]

Re: A few questions about recapping this particular PSU

espresso- as Zandrax pointed out, caps are not your biggest concern with this supply. Even with good caps, these supplies are likely to go and take your PC components with it!

[i4004]

Re: A few questions about recapping this particular PSU

this is JNC psu: should be thrown it into garbage immediately!

tinker with it only if you have no way of getting a good psu.

but i have trouble imagining region of the wrold where you CAN get low esr caps(which you seem to be able) and NOT be able to get good psu.
get a good psu, and recap it when needed(which is probably never).
if you can't get corsair 450, get fortron.
tell all those other people they have a ticking bomb in their machines.
charge them for changing theiri psus.
much better idoea than to repair such awfull quality psu.
(although this design is not the worst jnc was doing, you should stil l steer away from it if at all possible...)

[espresso]

Re: A few questions about recapping this particular PSU

Please do not comment the quality of this PSU as I am completely aware what I'm dealing with. That's why I am emphasising the last sentance that I wrote in my first post:

the main reason I'm doing this is because I wanna learn something new

I'm asking for an advice on recapping. So basically, I would like to know if I am safe with larger capacity on the output rails? I already have the capacitors, the question is whether I should mount them.

[zandrax]

Re: A few questions about recapping this particular PSU

So basically, I would like to know if I am safe with larger capacity on the output rails? I already have the capacitors, the question is whether I should mount them.

Read the thread about the 350 W psu Trodas recapped two years ago: Gonzo0815 says (comment #24) that increasing the output capacity by a large amount could change the bandwith of the regulator loop, leading to bad tension regulation (waving voltage) or unexpected oscillations in the circuit. 2nd point is low rated rectifiers may be overloaded by high capacity: most reputable brands use rectifiers rated for at least 4 time the real output (say 120 A for a 30 A 12 V line), cheap brands may skimp on those and use the cheapest one they find (say 20-30 A ones).
On another thread Davmax points out that increasing capacitance slows a bit cap's response time.
I respect Gonzo and Davmax but my knowledge in electonics is too limited to confirm or deny their claims: I hope they'll spot this thread eventually and properly reply you.

Zandrax

[i4004]

Re: A few questions about recapping this particular PSU

mount them and inspect voltages prior to connecting it to something valuable.
what's the worst that can happen? crappy jnc psu crapping out?
no damage there.

chances are it'll work.

as for learning...doubling wouldn't be enough for me..i would try tripling it, quadrupling it etc.
just like i just tried with a dc source and adding (one by one) 1000uF caps while monitoring the output voltage...

[linuxguru]

Re: A few questions about recapping this particular PSU

I have in the past doubled the rail capacitances on Deer/L&C to 2x2200uF on +5v and +3.3v, and to 1x2200 on +12v without much detriment. However, you may want to keep the caps on the +5Vsb rail closer to the designed value (say within +50%), since there is a small possibility that the auxiliary flyback converter may not start up reliably if the secondary caps on it become too large - 2 x 1000uF works fine in practice.

C1 and C2 need not be low-ESR - in fact, lower the ESR, the higher the inrush current on switching on at any instant other than zero-crossing. Ordinary general-purpose caps will suffice, as the primary caps almost never fail on even the crappiest PSUs.

[espresso]

Re: A few questions about recapping this particular PSU

I finished the job last night and here is the summary.
I used Fujicon RK and TM serties as replacements.
All the electrolytics have been replaced with the appropriate values except for these:

C18: [B]2200µf/6.3V 105C Low ESR[/B] instead of [B]1000µf/16V 105C Low ESR ("Saturn logo")[/B]
C32: [B]1000µf/16V 105C Low ESR[/B] instead of [B]470µf/16V 105C Low ESR ("Saturn logo")[/B]
C23: [B]2200µf/6.3V 105C Low ESR[/B] instead of [B]1000µf/10V 105C Low ESR (G-Luxon)[/B]
C28: [B]2200µf/6.3V 105C Low ESR[/B] instead of [B]1000µf/10V 105C Low ESR (G-Luxon)[/B]
C29: [B]1000µf/6.3V 105C Low ESR[/B] instead of [B]470µf/16V 105C Low ESR ("Saturn logo")[/B]

The original capacitors showed no visible signs of deterioration (bulging, leaking). However I am planing to get me an ESR meter for 50€ like this one

[linuxguru]

Re: A few questions about recapping this particular PSU

The critical one is C27, on the +12v rail. That's the one that's stressed the most on P4 systems or any other mobo where the VRM is fed from the +12v rail. You can safely go up up to 2200uF/16v, maybe higher on that one. Try to go for something with ~3A ripple current specs for that cap.

[espresso]

Re: A few questions about recapping this particular PSU

Unfortunatelly 2200uf/16v is too big to fit inside (13x25mm). I could probably mount it but it wouldn't be close to pcb.

[zandrax]

Re: A few questions about recapping this particular PSU

OT: where did you find that ESR meter, espresso?

Zandrax

[espresso]

Re: A few questions about recapping this particular PSU

This pic is from a russian site where they talk about constructing ESR meter:
http://pro-radio.ru/measure/3288/
However, a guy in my country made a similar model and now he is selling it for 50€.
Here is something that looks like a simpler solution that one could make at home


I can get all the parts and I would make this one if there wasn't for double layer PCB.

[zandrax]

Re: A few questions about recapping this particular PSU

Thanks.

Zandrax

[gonzo0815]

Re: A few questions about recapping this particular PSU

@espresso
does the 50€ ESR meter have the capacitance measure capacility too or is this only a generic pic?
If yes, which is your country and would it be possible to ship one to Germany?
I am in the mood to buy an ESR kit may be, but i would love one wich does disply both, ESR and capacitance.
Until now i have only seen the offer from Atlas wich i could get for abouth 200€.
As i don`t use it in an profesional environment, i would prefer to not spend that mutch money for it.

[Wizard]

Re: A few questions about recapping this particular PSU

You don't worry about capacitance (this give a value). ESR is! Ditto to ripple current capablity when choosing good caps.

Cheers, Wizard

[espresso]

Re: A few questions about recapping this particular PSU

@espresso
does the 50€ ESR meter have the capacitance measure capacility too or is this only a generic pic?

It does have the capacitance meter (0-99999uF) with 1% accuracy. Here is the picture of the 50€ device when dissasembled:




I can inbox you the guy's email so you can try contact him. The country is Serbia.
ESR measuring capabilities are

0,001 — 0,999 ohm with 0,001 ohm resolution
0,010 - 9,99 ohm with 0,01 ohm resolution
0,1 - 99,9 ohm with 0,1 ohm resolution