Recapping my PSU (non PFC) & a little modification

[linuxguru]

Re: Recapping my PSU (non PFC) & a little modification

1) Check if you actually have a TOP22x device. Most auxiliary circuits are simple unregulated self-oscillating flybacks that do not use a TOPswitch (using a TO-220 bipolar or MOSFET instead), with only a small primary-side zener providing any kind of regulation. If you see a small optoisolator adjacent to the auxiliary transformer, there *may* be a TOPswitch - look for a TOP221..7 mounted near the auxiliary transformer.

2) If it works with a 1500uF, don't disturb it. The only issues I can think off are transient regulation response, stability and self-starting, with the larger cap. Since it is able to start-up without problems and maintain a steady 4.95 V, you should be OK with the 1500 uF. Go with a 1000 uF, if possible, on future recaps.

3) Datasheets, ANs, etc. for all TOPswitch family products can be found here:

http://www.powerint.com/topswitchproduct.htm

4) On self-oscillating unregulated flybacks (without a TOPswitch), that little capacitor (22 uF to 100 uF) is critical - it prevents overvoltage on +5Vsb and Vcc for the SMPS controller on the secondary. It sits on a primary-side flyback winding that also provides some of the base drive for the auxilary switching transistor. It tends to dry out and fail with time, usually with zero/low capacitance.

[gonzo0815]

Re: Recapping my PSU (non PFC) & a little modification

Well thx linuxguru. In the ss400FS there are optokopler near the AUX supply transformer. The desing is nearly identical like some similar devices from st.com, only difference is, that st`s circuits all are 4 pin. So i am sure, that is an topswitch.

[yanz]

Re: Recapping my PSU (non PFC) & a little modification

Here's for comparison:

You have a very good eyes Thanks, for the info!

3) Datasheets, ANs, etc. for all TOPswitch family products can be found here:

http://www.powerint.com/topswitchproduct.htm

Thanks!

[yanz]

Re: Recapping my PSU (non PFC) & a little modification

This PSU, on secondary side, has components mounted on the heatsink:

• MOSPEC S16C40C V6D
• MOSPEC S20C40C ======> replaced with MBR3045CT
• BYQ28E 200 PMv 8423 A7 =====> i don't know what this is for.

Primary:

• D13007K 400II
• TT2146 M4F4 =====> x2 ; i gues they are the push pull switching transistor.

[yanz]

Re: Recapping my PSU (non PFC) & a little modification

Reina was right. The original schottky for 5V is only rated for 20A (MOSPEC 20C40C). How dare they rated its +5V at 35A? This PSu was never been designed to push that much amperage.

Loking at the datasheet there: http://www.deer-group.com/switch-12cm.htm , even the 400W is only 30A at 5V. i don't care if this PSU is originally a 300W, 350W, or 400W from the OEM, but with under rated secondary rectifier scottky diode, the label/sticker on this PSu was a total bullshit.

I think the 12V scottchky diode (MOSPEC S16C20C) should be upgraded too. I replaced it with a MBR2045CT, so the voltage drop would be lower, cmiiw. In anyway, this PSu is stil ATX v1.3, and i use them for socketA and p3. The +5V rail is the most important for me.

[yanz]

Re: Recapping my PSU (non PFC) & a little modification

Log:

upgraded the +5V schottky diodes (MOSPEC 20C40C) to MBR3045.

replaced the bridge rectifier (4 x CT 2A05) with RS405 (4A 600V rated).

upgraded the o/p filter caps:

+5v = (Panasonic fc 2700uf 6.3v) + (UCC LXZ 4700uF 10v)
+3.3v = (UCC LXJ 4700uF 6.3v) + (UCC LXZ 3300uF 10v)
+12v = (Nichicon HE 3300uF 16v)

Testing PSU at idle and full load, result is ok.

all rails are good, except for the 3.3v = 3.5v (steady). WTH?


could it be that the 3.3 vsense wire broken?


also, i noticed the discolored pcb under +5v resistor (10ohm 5w). replaced it with 10ohm 15w ceramic one, and glued it to the secondary heatsink.

before the resistor replacement, two caps near it (they were UCC KY 1000uF 25v) was boling hot.


without this tweaking, i wonder how the normal unmoded/untweaked psu will survive it's first 1 year?

[linuxguru]

Re: Recapping my PSU (non PFC) & a little modification

> all rails are good, except for the 3.3v = 3.5v (steady). WTH?

Hmm - the forward drop on one of the replacement rectifiers is probably lower than the original. That's normally a good thing, except in those cases where they've designed specifically for the higher forward drop, and the feedback loop is unable to regulate it down to spec.

This is a Deer/L&C specialty - they've dispensed with the linear regulator for +3.3v, and use a magamp or saturating inductor instead. I've found that replacing both the rectifiers (for +5v and +3.3v) with identical types usually keeps the voltages in spec, but YMMV.

[PeteS in CA]

Re: Recapping my PSU (non PFC) & a little modification

This is a Deer/L&C specialty - they've dispensed with the linear regulator for +3.3v, and use a magamp or saturating inductor instead.

I've seen magamp regulators on the +3.3V O/Ps of Delta and Power One (a good vendor, but AFAIK doesn't make PC P/Ss), so this isn't a hallmark of cheap. Magamps are more efficient than linear regulators - electrically and PCB space - I believe.

[Rainbow]

Re: Recapping my PSU (non PFC) & a little modification

I've never seen linear 3.3V regulator in a PSU.

[gonzo0815]

Re: Recapping my PSU (non PFC) & a little modification

Recently most PSu`s are made with Topswitches, due to the fact, that todays motherboards do realy require the 2A, wich are stated in the ATX spec. But if you want to see a linear regulator, open a Codegen PSu from abouth 1999 to 2003 and you will find one.

[Rainbow]

Re: Recapping my PSU (non PFC) & a little modification

Topswitch might be used for +5VSB but not for 3.3V.

[linuxguru]

Re: Recapping my PSU (non PFC) & a little modification

> Magamps are more efficient than linear regulators - electrically and PCB space

Oh, I agree - use of magamps isn't a sign of cutting corners. I meant that Deer tends to use under-rated rectifiers with higher forward drops (due to peak currents higher than spec), and they've designed the circuit (transformer turns ratio, secondary inductors, etc.) to take the higher forward drop into account. When you replace the rectifiers with higher-current substitutes, the lower forward-drop sometimes messes up the voltage regulation on the low-voltage rails (e.g. +3.3v) by a few hundred mV.

Older generation ATX PSUs with linear regulators on +3.3v could have probably coped with this, but apparently the magamp circuit on a Deer/L&C cannot.

> I've never seen linear 3.3V regulator in a PSU.

If you look closely at the secondary-side heatsink (with the secondary rectifiers), you'll sometimes find a linear 3-terminal regulator.

They were standard in the P1/PII era, when the current rating on +3.3v did not exceed about 10A. A simple 3-terminal linear regulator (usually a high-current LT part) to drop +5v to +3.3v did the trick. Sometimes, a separate rail with a loosely-regulated +4v was dropped to a tightly-regulated +3.3v with a linear regulator.