Recently Opened up the Computer SMPS cover. The SMPS is rated 400Watts and gives o/p of 12V@17 amp, 5V@32amp, 3.3V@30amps. However need advice on the capacitance values on each rails. The 12v has 2200micro, 5v & 3.3v both have 1000micro. These are the present capacitors on the SMPS PCB. Do you people think these are adequate cap values considering the high load currents. I know there are some calculations for finding the o/p capacitance values- but i think the present values are too less. Seen such small caps values on very low power SMPS like adapters etc. Are these values OK, if not considering adding capacitor banks externally on the molex connectors--approx 40,000 micro for each rail.
Generaly computer SMPS dont have to supply large current swings, re bass transients etc, ie the current demand is steady, also there is a large amount of distributed capacitance on the PCB's, where it can supply the local demand for switching transients (ie 32 address lines togling), usualy at least 0.1uf per chip, with larger ones having 2 or 3 different values parralleled.
Our SMPS probably have about a quarter of the OP capacitance than they had 4 years ago.
So IMHO for audio I would increase the resorvoir caps, cos if you tax SMPS the controlling chips go into protection mode which aint gonna due the rest of the stuff down the line any good.
Our SMPS probably have about a quarter of the OP capacitance than they had 4 years ago.
So IMHO for audio I would increase the resorvoir caps, cos if you tax SMPS the controlling chips go into protection mode which aint gonna due the rest of the stuff down the line any good.
Yep have seen the locally distributed caps on the motherboard, even then when opening heavyduty applications like games and stuff, the SMPS temp rises by a good amount, indicating rise in currents, will additional caps help in that case.
Asking because for my old Pentium II comp, after 5 yrs of use the HDD would automatically go in standby mode for a few seconds, and would periodically produce clicking sound. Finding no fault in s/w, opened the SMPS and changed the caps from 1000 to 4700 for all rails. The HDD problem was solved then. The present question is for my new PC.
Also if i add large cap banks- the caps will initially be in discharged state, when powering up the SMPS, a large surge current will flow. Will the SMPS controller sense this surge as o/p short and refuse to power up?
Asking because for my old Pentium II comp, after 5 yrs of use the HDD would automatically go in standby mode for a few seconds, and would periodically produce clicking sound. Finding no fault in s/w, opened the SMPS and changed the caps from 1000 to 4700 for all rails. The HDD problem was solved then. The present question is for my new PC.
Also if i add large cap banks- the caps will initially be in discharged state, when powering up the SMPS, a large surge current will flow. Will the SMPS controller sense this surge as o/p short and refuse to power up?
Usual no-name computer supplies are cut right down to the bone in terms of component ratings. It would not surprise me at all if your output caps had dried up after 5 years duty, especially as they were probably off-brand units of dubious specs and reliability. What can you expect when you can get a case and supply for $30 US RETAIL?
Agree with Wrenchone, I work in the design dept. of an electronic OEm and unfortenately its true that products are designed down to a price, and as PSU's are one of the most expensive components these suffer more than most. Cheep capacitors and reservoir caps cut to the bare minimum.
The best thing you can do is buy a decent supply rated for at least 1.5x 2x your expected amximum demand. We have had more computers blow up do to PSU going than anything.
The best thing you can do is buy a decent supply rated for at least 1.5x 2x your expected amximum demand. We have had more computers blow up do to PSU going than anything.
Mine is a medium end comp- got the SMPS in the end-- of medium quality as was short of money.
Anyways cover says 12V@17A. However for the 12V rails the diode is MOSPEC F12C20C. The datasheet for the diode says 6A/leg and 12A total absolute max current. So from where 17amps 🙁 .
For the 5V rail diode is SBL3040PT- max 30A capacity--now the 3.3V is derived from 5V rail via mosfet- so the actual current is very less. The PS has a CE marking on the cover.
Will add a GPU card later- don`t know what will happen then.
Anyways cover says 12V@17A. However for the 12V rails the diode is MOSPEC F12C20C. The datasheet for the diode says 6A/leg and 12A total absolute max current. So from where 17amps 🙁 .
For the 5V rail diode is SBL3040PT- max 30A capacity--now the 3.3V is derived from 5V rail via mosfet- so the actual current is very less. The PS has a CE marking on the cover.
Will add a GPU card later- don`t know what will happen then.
Imerse PSU in liquid nitrogen, purley resistive load, and you might get a 17A current pulse for approx one femto-second. Ratings are quite often dictated by marketing departments, now IMHO all marketing people come from a far far far away planet, and travel here in spaceships powered by the by-product of male bovine herbaceous digestion.
When I buy a case these days, I strip out the cheap no-name supply and repace it with a brand that I can at least recognize - Antec and SPI are a couple that come to mind. Neither of these are first-tier suppliers, but are still a lot better than the bottom feeders. The difference in weight alone is immediately apparent. The cheapo stuff feels like it would blow away with the silightest breeze. Scarier, though - I've popped the cover in a lot of these supplies and found only a cursory sketch of an EMI filter, with standard film and ceramics used instead of safety rated components.
in short, most computer PSU's today are c**p. you need to dig to find the really good ones.
I hope I could find a shop that would allow me to open up a SMPS and take a peek inside before buying them....
I hope I could find a shop that would allow me to open up a SMPS and take a peek inside before buying them....
- Status
- Not open for further replies.