Theory question / exercise in determining capabilities of ATX PSU

[b700029]

Re: Theory question / exercise in determining capabilities of ATX PSU

- When the PSU is used in a country with 220-240 VAC, both the big and the small diodes are used to form a full-wave bridge rectifier. Therefore, the current pulled from the wall by the PSU is limited by the rating of the smaller diodes.

And on 220-240, the amount of current drawn for a given output is on average 1/2 that for 110-120, so they save a few cents by using smaller diodes. Really clever cost savings, too bad they can't get anything else right

[momaka]

Re: Theory question / exercise in determining capabilities of ATX PSU

I don't see how two small diodes and two big diodes can be used in the bridge rectifier.

It's just like using 4 diodes of the same size.
Leadman/PowMax and low-end Sun Pro PSUs often do this.

In such a case, the big diodes are connected to the Live, and the smaller ones to Neutral.
- When the PSU is used in a country with 110-120 VAC, the smaller diodes are bi-passed by the voltage-selecting switch. Then the Neutral becomes connected to the common connection between the two primary filter caps and you have a voltage doubler circuit. Therefore, the current pulled from the wall by the PSU is limited by the rating of the bigger diodes.
- When the PSU is used in a country with 220-240 VAC, both the big and the small diodes are used to form a full-wave bridge rectifier. Therefore, the current pulled from the wall by the PSU is limited by the rating of the smaller diodes.

[lti]

Re: Theory question / exercise in determining capabilities of ATX PSU

I have a 75W power supply with more guts, and the manufacturer had to cram everything into a smaller enclosure.

Only one cap filters each output, there is no room for a pi filter, and the main filter coil is smaller than the 3.3V coil. If the filtering components had been installed and the heatsinks were bigger, the transformer would limit the output power.

I don't see how two small diodes and two big diodes can be used in the bridge rectifier.

[c_hegge]

Re: Theory question / exercise in determining capabilities of ATX PSU

ah, OK. I have an ISO -branded- PSU and it has CWT markings all over the transformers. In any case, though, they must have taken a low end 100W CWT and put a 450W label on it.

[370forlife]

Re: Theory question / exercise in determining capabilities of ATX PSU

^exactly.

[rogfanther]

Re: Theory question / exercise in determining capabilities of ATX PSU

Hi,

green in like these ?

[370forlife]

Re: Theory question / exercise in determining capabilities of ATX PSU

@370forlife, You sure that's a CWT? I can't imagine CWT would build something that bad (and there are no "CWT" markings on the transformers and no green tape). It looks like an Okia/Broadway Com Corp

BCC use CWT and yuelin as an oem. This is not a yuelin. But like I said, it is probably from CWT's ISO lineup which do not use CWT marked transformers and do not necessarily use green tape.

[c_hegge]

Re: Theory question / exercise in determining capabilities of ATX PSU

You can use some old drives and fans, but don't connect anything to this PSU unless you don't mind killing it.

My load tester is one I put together myself with about 30 large power resistors.

To measure the ripple, you need an oscilloscope. I use an EasySync/USB Instruments DS1M12 Stingray.

[rogfanther]

Re: Theory question / exercise in determining capabilities of ATX PSU

Thanks, I forgot to search for current specifications for the switching transistors. Will read the datasheet again and add that to my notes.

Shouldn´t CWT be somewhat like a reputable company, producing like maybe not-so-great PSUs, but at least better than this ? I believe most of the common PSUs sold here are rebadged L&C´s ( will post pictures when I find them in the PSU box ) .

Because I don´t want to use good caps in this, I can´t measure the ripple, but will try to do it when some new one crosses my path. Is there a "proper" way to measure it , with some defined load, or can I just use some old harddisks and fans ?

Thanks.

[c_hegge]

Re: Theory question / exercise in determining capabilities of ATX PSU

In this case, though, it wouldn't take much at all to overheat them with those tiny little heat sinks. I recently put an okia PSU on my load tester which looked similar to your unit. It managed only 79% efficiency on 230V (120W load), and failed at only 250W load. On a lower input voltage, the efficiency would be lower, so it would generate more heat and fail sooner. The ripple was WAAAYYYYY over the limit, even at 120W, at around 180mV on the 5V rail and over 200mV on the 12V.

[Th3_uN1Qu3]

Re: Theory question / exercise in determining capabilities of ATX PSU

is my analysis correct, in that with the maximum current for the secondary rectifiers, or the input diodes, limits the real maximum this psu would be able to do ?

In cheap PSUs generally the maximum current of the secondary rectifiers limits the output power - but to reach that maximum current the primary switchers must be able to take it, and most of the time the heatsink they're on is way undersized for the task, and unless the overload was on only one rail they will usually fail before the diodes do. On your particular PSU you also have a really small output filtering toroid coil (the round thing with lots of wires on it). It'll do 200W... barely.

Approaching maximum current also means high voltage drop thus a lot of wasted power (and more heat to get rid of). That's why cheap PSUs have lousy efficiency. Picture this - a 300W supply with 70% efficiency eats 429W from the wall. If the efficiency were 80% said PSU would be able to provide 350W from the same input power.

As for primary rectifiers, their current handling capability depends a lot on their temperature. It's not uncommon for them to survive almost twice the rated current continuously if they're placed in the path of airflow.

[c_hegge]

Re: Theory question / exercise in determining capabilities of ATX PSU

@370forlife, You sure that's a CWT? I can't imagine CWT would build something that bad (and there are no "CWT" markings on the transformers and no green tape). It looks like an Okia/Broadway Com Corp to me.

@rogfanther, That is correct. The input diodes could in theory handle 381W assumming a 127V input which will equate to about 285W output assumming 75% efficiency, however, those switching transistors will go kaboom at before that.

[rogfanther]

Re: Theory question / exercise in determining capabilities of ATX PSU

The primary transistors are KSH130007A.

370forlife, but is my analysis correct, in that with the maximum current for the secondary rectifiers, or the input diodes, limits the real maximum this psu would be able to do ?

Thanks.

[b700029]

Re: Theory question / exercise in determining capabilities of ATX PSU

What are the primary transistors?

[370forlife]

Re: Theory question / exercise in determining capabilities of ATX PSU

This is a CWT unit, unfortunately. Probably from their ISO division. It could probably do 100W barely in spec, maybe 150 before blowing.

While the amperage rating of the input diodes would indicate they would fail at 381W, typically they can do much more than that. But it dosen't really matter as it would never reach that level of loading, probably more along the lines of 250W input to get 150W output before 'sploding.

[rogfanther]

Hi,

First, I know this is a gutless crap. They sell well here, and are the kind of model found in popular shops.

It is a Coletek CT-2045S. I collected a bunch of burnt PSUs from friends just for studying how to fix them, and was lucky to get 2 or 3 of the same model each. I ask you for please reviewing my conclusions.

From the label :

Peak Power : 450W

3.3V - 18A
5 V - 22A
12V - 25A
5Vsb - 2A
-12V 0.3A

Maximum combined power for 3.3V and 5V : 140W

Primary diodes : 2 x 1N5406 ( rated max 3A ) and 2x 2A05 ( rated max 2A ) .

Secondary rectifiers : 3.3V - S20C45C ( max. 20A )
5V - S2045C ( max 20 A )
12 V - F12C20C ( max 12A )

The main controller is a Welltrend WT7520. Secondary caps are marked 1000uF x 16V KYS . The -12V line gets a 470uF x 16V KYS.

The voltage here is 127VAC.
So, I believe I could argue that the most power this would be able to have was 127 x 3A = 381W.

With those secondary rectifiers, while the 3.3V current is inside the possible, the 5V exceeds it ( and in one of the psus, the 5V diode was shorted ) , and the 12V line is very exagerated.

That , along with the ausence of pi filters, would make this a not-that-good 350W PSU with much ripple.

Sorry for the bad photos, if necessary I can get a better camera in the office.