Super Flower Power Supplies

[japlytic]

In a computer used for video editing, there is a Super Flower 500W power supply inside it. During heavy CPU loading, the 12V rail drops to about 11.4V. From the ventillation holes, I can see a JP (Jen Pan - www.jpcon.com) electrolytic capacitor.

Does anyone have a bad unit with photos of the internals?

[Newbie2]

Re: Super Flower Power Supplies

In a computer used for video editing, there is a Super Flower 500W power supply inside it. During heavy CPU loading, the 12V rail drops to about 11.4V. From the ventillation holes, I can see a JP (Jen Pan - www.jpcon.com) electrolytic capacitor.

Does anyone have a bad unit with photos of the internals?

JPCON caps spell trouble! I'm guessing the JPCON caps are doing part of the 12V rail voltage dropping when the Super Flower PSU is under heavy load. I do not have a picture of the internals of a Super Flower 500W PSU. Is this a computer at the place where you work or it's your computer?

[japlytic]

Re: Super Flower Power Supplies

It is at home (not mine). The power supply is still under warranty, with the seal intact. Anyone else with a picture of the internals?

[PeteS in CA]

Re: Super Flower Power Supplies

Bad cap(s) is a possibility. Also, if a VRM draws power from the +12V O/P, and if the P/S doesn't use a shared regulation scheme for the +5V and the +12V (i.e. the +12V isn't regulated), the extra current demand from the +12V may just be dragging it down. A concurrent light load on the +5V would make this worse; crap-caps would make it worse, too.

Haven't heard of Super Flower.

[Newbie2]

Re: Super Flower Power Supplies

It is at home (not mine). The power supply is still under warranty, with the seal intact. Anyone else with a picture of the internals?

If you don't mind getting into trouble, you can try cutting the middle of the warranty seal with a knife. I have tried that and it's very hard to notice that the warranty is void! I recommend Enermax power supplies, they don't have a warranty seal to rip off or cut!

[japlytic]

Re: Super Flower Power Supplies

I can see all of the caps through the vent for the large fan, without opening the case. None were bulging, and there were two 1000uF 200V electrolytics in the primary - I think this is adequate for a supply of this wattage.
I guess the 12V rectifiers may be inadequate - a good power supply of such a high wattage should have a separate power transformer for each rail (+12V, +5V, +3.3V), so a change in load on one of the rails woudl not affect the other. I have seen such a supply, which was out of a server.

[PeteS in CA]

Re: Super Flower Power Supplies

Using separate converters like that would be expensive - in $$ and space. It would require extra power transformers (custom parts), extra switch devices and more heatsink space, extra PWM ICs (which would have to be synchronized to prevent O/P noise problems) probably separate O/P inductors for each converter (PC P/Ss usually have a coupled inductor, one core with separate windings for each O/P; with proper turns ratios, this improves regulation of the various O/Ps). PC-grade P/S are too inexpensive to do this, though, properly implemented, it would solve the problem.

O/P rectifiers typically have forward drops of ~.5V for Schottkies, ~1V for ultrafasts, with variation due to load current of .2V-.4V. Unless the +12V is 11.6V-11.8V under light load, I don't think a marginally rated 12V O/P rectifier is the cause of the voltage drop. And the rectifier may be marginal, but the problem would show up as extra heat and shorter life. There are a several ways to improve the regulation of the +12V. One is thru the use, noted above, of a coupled inductor. Another is to use the +5V tansformer winding as part of the +12V winding. A third is by including the +12V in the the feedback of the main regulator. You give up a little regulation in the +5V, but gain regulation in the +12V, and this is accomplished by adding a resistor, making the sample of the O/P provided to the error amplifier a combination of the +5V and the +12V, rather than just a sample of the +5V. A technique I've seen used for the +3.3V O/P is a sense wire added to one of the ATX connector +3.3V pins, so that the regulator compensates for some of the voltage drop in the O/P harness wires.

My guess is that Super Flower either didn't include the +12V in the main regulator feedback or messed up the turns ratio of the coupled inductor, or both.

"Stupid" question time, are all P/S O/P connectors properly seated? Another "stupid" possibility is that the solder joint on the P/S PCB for the +12V wires could be bad. A third "stupid" possibility is bad connector socket crimps.

[arneson]

Re: Super Flower Power Supplies

are all P/S O/P connectors properly seated? Another "stupid" possibility is that the solder joint on the P/S PCB for the +12V wires could be bad. A third "stupid" possibility is bad connector socket crimps.

Forrest Gump always said, Stupid is as stupid does.
I've seen all three of those issues in power supplies,
but I've also seen them fly across the lania.

[PeteS in CA]

Re: Super Flower Power Supplies

Yep, those are all common Oh-S__ts. One time a company I worked for had O/P harnesses made by a division of a major connector mfr. After making 10K-12K P/Ss with those harnesses, the customer, a major computer mfr was experiencing a rash of failures in test (not at end-users' sites, thankfully). When it got down to the root cause, that division of a major connector mfr used the wrong crimping tool on the major connector mfr's own product.

[arneson]

Re: Super Flower Power Supplies

I was doing life safty systems in hospitals years ago.
It was Edwards fire alarm controle panels, totally loaded with Molex, 12 and 16 pin, hundreds of them interconnecting panels. There's nothing quite like troubleshooting a panel full of molexes.

[japlytic]

Re: Super Flower Power Supplies

A multimeter reading on each rail corresponded with the readings on a monitoring utility, which ruled out any problem realted to the connectors. I do not think there would be a bad solder joint inside the supply.

www.super-flower.com.tw - They were established in 1991.

(Many utilities do not monitor the negative rails, since they are rarely used).

[japlytic]

Re: Super Flower Power Supplies

Today, I opened up the supply (nothing was bulging!)

The SBL3040 (2 in parallel) (+5V) and SBL4040 (+3.3V) rectifiers have a foward voltage of 0.55V, while the F30D20 (+12V) has a foward voltage of 1.3V. So that's why the 12V rail is 0.6V lower than normal during high CPU usage, but this is not causing any problems .

Acutally, the capacitors were Jenpo CE-WG (1000uF 200V 85C - at this temperature rating) in the primary, with the "JP" units (often called PCE-TUR)

The capacitors (JP) were:
3300uF 16V (TUR) on the +12V rail (none before the primary inductor)
2200uF 6.3V (TUL) on the +3.3V rail (one before and after primary inductor)
4700uF 10V (TUL) on the +5V rail before primary inductor; 2200uF 10V (TUR) after primary inductor on the +5V rail.
1000uF 10V (TUR) on the +5V standby rail.
470uF 16V (TUR) on the negative rails.
There is a 10 ohm, 10W wirewound load resistor on the +5V rail (2.5W, 0.5A load)
The X2 and Y-class capacitors are approved by various safety agencies.

The SMPS chip is a KA7500, and there is a custom chip; which is a SFB03B24A2U1.

http://www.0759.com/products/ecap/ecap_e/16-17.htm (TUL Series)
http://www.0759.com/products/ecap/ecap_e/14-15.htm (TUR Series)

[PeteS in CA]

Re: Super Flower Power Supplies

The SBL3040 (2 in parallel) (+5V) and SBL4040 (+3.3V) rectifiers have a foward voltage of 0.55V, while the F30D20 (+12V) has a foward voltage of 1.3V.

We-e-e-elllll ... that may not be a problem. The +5V and +3.3V rectifiers are schottky types - 30A@40V for the +5V and 40A@40V for the +3.3V. OTOH, the +12V rectifier (possibly 30A@200V) will be ultrafast recovery types. Schottkies have a V(F) of .4-.7V; ultrafasts' V(F) are 1.0-1.3V - two different technologies (and max reverse voltage available for a schottky is 100V, IIRC).

[japlytic]

Re: Super Flower Power Supplies

The custom chip is acutally a WT7512, used for the PowerGood signal, which is in the same position in a similar model on http://terasan.info/dengen/no083/index.html
WT7512 datasheet:
A 14cm fan blows air over the heatsinks from inside the case.
Since two SBL3040 rectifiers for the +5V rail are in parallel, the voltage drop would be 0.275V.
I forgot to mention that the primary transistors are 2SC3320s, with ratings of 500V 15A having a maximum dissipation of 80W.

[kc8adu]

Re: Super Flower Power Supplies

Today, I opened up the supply (nothing was bulging!)

The SBL3040 (2 in parallel) (+5V) and SBL4040 (+3.3V) rectifiers have a foward voltage of 0.55V, while the F30D20 (+12V) has a foward voltage of 1.3V. So that's why the 12V rail is 0.6V lower than normal during high CPU usage, but this is not causing any problems .

Acutally, the capacitors were Jenpo CE-WG (1000uF 200V 85C - at this temperature rating) in the primary, with the "JP" units (often called PCE-TUR)

The capacitors (JP) were:
3300uF 16V (TUR) on the +12V rail (none before the primary inductor)
2200uF 6.3V (TUL) on the +3.3V rail (one before and after primary inductor)
4700uF 10V (TUL) on the +5V rail before primary inductor; 2200uF 10V (TUR) after primary inductor on the +5V rail.
1000uF 10V (TUR) on the +5V standby rail.
470uF 16V (TUR) on the negative rails.
There is a 10 ohm, 10W wirewound load resistor on the +5V rail (2.5W, 0.5A load)
The X2 and Y-class capacitors are approved by various safety agencies.

The SMPS chip is a KA7500, and there is a custom chip; which is a SFB03B24A2U1.

http://www.0759.com/products/ecap/ecap_e/16-17.htm (TUL Series)
http://www.0759.com/products/ecap/ecap_e/14-15.htm (TUR Series)

sounds like a "mad dog" unit i have on the scrap pile.
the molex burned up on +3.3.
the fan says super flower on it and has the big resistor on the heatsink.

[kc8adu]

Re: Super Flower Power Supplies

Since two SBL3040 rectifiers for the +5V rail are in parallel, the voltage drop would be 0.275V.
.

vf stays the same regardless of how many are in paralel.
current handling doubles though as long as one doesnt have a lower vf and hogs all the current.most semi's are pretty consistant nowadays so you are not likely to have that problem.

[larrymoencurly]

Re: Super Flower Power Supplies

The SMPS chip is a KA7500, and there is a custom chip; which is a SFB03B24A2U1.

Is it possible that the second chip is actually an LM324 quad op-amp? I've seen them in some PSUs, including one with a KA7500 chip in it.

[Chris1992]

Re: Super Flower Power Supplies

btw @ kc8: mad dog are usually decent power supplies. wonder how the molex got burnt on 3.3.

[kc8adu]

Re: Super Flower Power Supplies

btw @ kc8: mad dog are usually decent power supplies. wonder how the molex got burnt on 3.3.

all 3 +3.3 were burned.
maybe just a bum molex?
iirc its an abit ka7.replaced the molex on it and its running fine.
been running it in my testbed just to beat it up and see if it does anything stupid.
customer replaced the maddog with a seasonic.
havent seen any other reports.
the maddog has crap caps though.

[Rainbow]

Re: Super Flower Power Supplies

ATX connector usually burns as a result of bad contact. The board probably draws a lot of power from 3.3V (although KT7 does not, it has its own switching regulator that runs from 5V - so 3.3V is adjustable, don't know about KA7).