Input capacitors SMPS

[Tioleco]

What series are recommended for the input filter of SMPS , 470uF , 680uF and 820uF / 200V?
I´m looking for Chemicon KMH , KMQ , LXM and LXS ; Rubycon VXG , VXR ; and Nichicon GU series.

[mscserbia]

Re: Input capacitors SMPS

Rubycon VXG,they are good.

[Pentium4]

Re: Input capacitors SMPS

For those low end PSU's with the voltage doubler, I'd say it's safe to use just about any 200V snap in capacitor. They are hardly stressed, and ESR is not important. Even the worst brands will last a very very long time there

[selldoor]

Re: Input capacitors SMPS

For those low end PSU's with the voltage doubler, I'd say it's safe to use just about any 200V snap in capacitor. They are hardly stressed, and ESR is not important. Even the worst brands will last a very very long time there

I do not know enough about PSUs to question your reply but just to err on the side of caution his mains voltage is 220-240VAC does that make a difference?
or are we not discussing mains input filter caps?

[Per Hansson]

Re: Input capacitors SMPS

selldoor: The voltage selector switch puts the caps in series when set to 230v, so the 200v caps together will be able to handle 400v
Which is well within the peak voltage expected: 240VAC x √2 ≈ 340v

[mariushm]

Re: Input capacitors SMPS

A power supply rectifies the AC voltage into DC, so you have AC Voltage x 1.414. For 240v AC, you have DC voltage that peaks at around 340v.

If a power supply has two large 200v capacitors, they're connected in series or parallel depending on the switch position in the back.
For 110v countries, the capacitors are in parallel, giving you 2 x capacitance and same voltage rating.
For 220-240v countries, the capacitors are in series, giving you capacitance/2 and 2x voltage rating.

1/Ctotal = 1/C1 + 1/C2 => C1 x C2 / Ctotal = (C2 + C1 ) => Ctotal = C1 x C2 / (C1+C2)

If C1=C2 => Ctotal = C1xC1 / 2 x C1 = C1 / 2


The more capacitance on the input, the power supply can handle larger drops in power in case your house electricity is not perfect. Also, larger capacitance can help a bit in the total watts of regulated power a power supply can provide to the computer.

For 300w, about 270-330uF may be enough. For 400w, maybe 470u-680uF. For more, I would recommend 820uF. More experienced people feel free to contradict me.

These large capacitors don't need to be low esr or be any special, and in most cases you could even use GOOD quality 85c rated capacitors (Panasonic, Rubycon etc) - it depends on the design of the power supply, if there's some distance from the heatsink so they won't heat too much.

More modern power supplies have an Active PFC circuit which boosts whatever voltage it receives (let's say anything between 90v and 250v) to about 400-420v so in these new power supplies you often see just one capacitor with a voltage rating of about 420-450v. You should NOT use capacitors with lower voltage rating for those power supplies.
In these types of power supplies, you should get 105c rated capacitors.

[selldoor]

Re: Input capacitors SMPS

Thanks chaps - I know a bit more now. Remembering it might be a problem .

Also as a side I have just read that Brasil has both 220vac and 110vac


- can vary from city to city - nightmare

[Wester547]

Re: Input capacitors SMPS

For 300w, about 270-330uF may be enough. For 400w, maybe 470u-680uF. For more, I would recommend 820uF. More experienced people feel free to contradict me.

Not to say that I know more than some of the more experienced members here do, but I think for the voltage doubler at least (not the primary capacitors of an Active PFC circuit), 270uF-330uF is a bit low for 300W and otherwise. If the two capacitors in the voltage doubler can't smooth enough current for the main switcher(s) to chop and feed to the main transformer then the main switcher(s) have to work harder and the main transformer has to work harder as well (needs a greater turns ratio). It also allows more 100/120Hz ripple through which isn't a good thing (and is separate from the ripple voltage and current that the output filtering takes care of, though the control loop compensates for it to some extent, and the ferrite coils on the output can also suppress some of the 100/120Hz spikes, but only to a small degree; that sort of ripple goes straight through the output capacitors and magamp circuit otherwise unless the capacitance of the output capacitors is some unfeasibly high value). I think 470uF/200V x2 is a good minimum for 300W power supplies and 680uF/200V x2 for 400W power supplies respectively (but, ideally, IMO, 560uF-680uF is a good range for 300W units and 820uF+ for 400W units). It is also correct that more capacitance improves hold up time, though those capacitors will be charging for longer periods of time, the more the capacitance.

Also, because the capacitors in the voltage doubler are not ESR critical, they are relatively unstressed and capacitance (assuming the voltage rating is correct) is really only the important facet when it comes to them. However, because the truly bad brands especially lose their capacitance a lot faster than the Japanese ones, I would recommend sticking to Japanese brands if possible (even 105*C ones if possible since they don't dry out as fast as 85*C ones despite the lower ripple ratings). Also, some of the truly horrendous brands overrate their high voltage capacitors by quite a bit capacitance wise. Also, to my knowledge, the capacitors charge in parallel and discharge in series.

[budm]

Re: Input capacitors SMPS

Filter cap cal.

[Per Hansson]

Re: Input capacitors SMPS

One thing that's surprising to me is that most power supplies seem to not live up to the hold up time of min 16ms that the ATX spec mandates.

Scroll down a bit to "Hold-up Time"
http://www.techpowerup.com/reviews/C.../CS650M/5.html

[PeteS in CA]

Re: Input capacitors SMPS

There are ways to fudge a hold-up time test. One would be to remove AC power just after the peak of the sine wave (just after the input caps got a slug of charging current) rather than at the zero-volt crossing (as done in the article's test). And cheapie units are probably sold with the assumption that the actual system load will be less than full rated power, and that the user can't know the difference, anyway.

[Per Hansson]

Re: Input capacitors SMPS

Yea I guess so, but as an end-user it is very easy to notice.
Because if you live in a place with bad power and you get a UPS it will be designed with the ATX spec in mind.

So if the blackout happens in the peak of a sine wave it might have time to switch the load over.
But if the blackout happens not on a peak (most likely, ref: "Murphys law" and also "sandwich law") then the supply might not have the time to switch and so the end-user will blame the UPS, when it's actually the power supplies fault...

[TELVM]

Re: Input capacitors SMPS

One thing that's surprising to me is that most power supplies seem to not live up to the hold up time of min 16ms that the ATX spec mandates ...

Penny pinching on the bulk cap, those 400+V caps are expensive.

Of the two PSUs that pass ATX hold-up in that table, the Cooler Master GM 550W has one 470uF for 550W, the Antec VP450F has one 330uF for 450W.

The Corsair CS subject of the article has a much lower ratio of one 220uF for 650W, and fails.