Corsair CMPSU-550vx / vx55w hack

[domas]

Hi there, long time since last posted..

I've got my hands on this ATX psu recently that has been modded by someone I know - output connectors cut, pson switch added, stby and pwr_good leds installed on the wall. A good strongly built p/s for projects. He did not need it anymore, I took it over. As expected, I am making this to adjustable PSU.

It has main transformer and stby transformer, the fets are controled by optocoupler which has CM6800G on the hot side and a little daugherboard on the cold side. It has a few transistors for fan control and PS229 for protections and WT7518D for more protections ;-]
Couldn't care less for hot side as I am not doing anything there so cold side protection chip datasheets follow here:
https://cdn.badcaps-static.com/pdfs/...e09b7dce1a.pdf
https://cdn.badcaps-static.com/pdfs/...d6d7f00ff7.pdf

The good about the power supply: strongly built, good caps, ok soldering, humongous rectifiers, shitload of input filtering, you name it. Actually i am sorry to see this thing not being used in a pc anymore. Has sense wires for 3,3, +5 and +12 rails. Has 4 12v rails, which is coming from one coil, so single rail unit, but has 4 current sensing resistors for output protections. not sure why not a single one.
3,3v and 5V rails are rectified by STPS30L45CW rectifiers, while 12V rail has 2x STPS3060CW in parallel.
https://cdn.badcaps-static.com/pdfs/...48a715a28f.pdf
https://cdn.badcaps-static.com/pdfs/...e12740bf89.pdf

The good about the hack: tl494 controls optocoupler, seems very easy job to hack one resistor, change to adjustable, job done.
The bad: the daughterboard. ps229 as seen in datasheets wants exact voltage inputs (3,3, 5, 12) to voltage sense pins, not 3-ish volts where you have to reduce the output by voltage divider from respective rail but actually fake it from 5vsb rail while doing the hack.
Current sense is one wire only, the other input is used from voltage sense input to compare tiny voltage drop. That being said, no chances to use the chip for voltage protection, thus no chances to use it for OC protection

Daughterboard has really 18 pins, most of them U/O VP and OCP
Two pins for each rail except for -12
PG output, on/off optocoupler control, pson input, 5vsb in , -12 in, some strange 15.43v thing (might be stby transformer output before rectifier), then 2x for 3,3v, 2x for 5v, GND, and 7X for 12V . Four of them are rail outputs, 1 pre current sense resistor input for PS229 chip and 2 for WT chip, as PS229 can only control two rails.. So yes, the board can be removed and you loose PG indicator and protections, just grounding pin 2 starts the supply.
Oh and daughterboard has 3 connectors for fan control - 12V input (separate), NTC sensor and fan output. IT works with the board out of main board if you plug the three connectors. So if I decide to ditch the little board, i might try to separate those components. Might as well not.

All rails, apart from 12V are leaving the p/s for good.

Questions:
So I really have no chances of keeping OC protections? isn't there anything on the hot side that might save my p/s if I short it? Any advices regarding that? Automotive fuse? I might use WT chip for two rails as it is not fixed voltage unit and used only for current sensing.

Do I remove 12Vsense wire with control crap at all or keep it after I make it adjustable?

Am I expected to unwind -12V wire from the big toroid? Any other thoughts?

What the hell is the white huge resistor on the little board for? Minimal load? there? really? 15ohms It is close to fan control circuitry

This is CWT unit. Nobody has diagram, do you? I do not love smd components. Hard to follow things without diagram

[momaka]

Re: Corsair CMPSU-550vx / vx55w hack

Probably not the answers you are looking for... but I think it would be easier to take a cheap-o half-bridge PSU and modify that for your needs. This PSU would be better suited for PC use.

[domas]

Re: Corsair CMPSU-550vx / vx55w hack

Hi there.

This psu had its connectors cut off, 2 leds installed and a pson switch - so it wasn't worth the hassle.

Actually I have already stripped the psu. Unfortunately protections had to go as described before. Actually, the little daughterboard had only 2 outputs - PGO to main board and to pgo wire, and the actual PSON to optocoupler. So basically after removing the board PSON wire was changed to one of the daughterboard outputs to mainbard ;-] easy. I have stripped all not usable components of it -12, +3,3, +5v was removed. 12v still working (remember protection daughterboard is removed so those are not monitored anymore thus psu keeps runnning with 12V only) Haven't hacked the 431 yet (and yes i did a mistake typing tl494 in the first post, had 431 in mind.) so it is fixed 12V supply. I may try to keep the sense wire working though depending of how it is implemented. All 3 rails had sense wires. Incredible.

The OC protection exists on the hot side. In case i short anything, PSU enters limited current mode. So not a huge risk, still, i feel a need for ov protection in case feedback loop is damaged, don't want to damage the caps. Probably will implement it with another tl431 voltage reference.

Fan control will be kept the same, i hope I can manage to move components to another board, if not, i will keep the daughterboard for the purpose. I will need a simple 12v supply for fan though. I have a separate tap on transformer that is no longer in use.

The interesting part: the main trafo has 2 outputs, 12V and 3+5V and GND of course.
12V is pretty standard apart from sense wire.

3,3V and 5v are regulated form trafo this way: output splits into two separate coils, controlled by some active component, looked like transistor, and after rectification they had 2 coils each again, one for vsense compensations (or a separate filter not sure), and another for filtering, 2x2200uf caps for each rail. That is impressive IMHO. Never seen this before.

12V filter coil had only -12v rail wire on, so it was easy to remove, didn't have to unwind 12V rail.

So answers to my questions:
yes there is overcurrent/overload protection on the hot side and is very efficient, i would say it is even too careful.
Just had to unwind -12v rail wire from 12v rail coil, no problems.

Pictures in next post

[domas]

Re: Corsair CMPSU-550vx / vx55w hack

In the first 2 pictures you can see +3,3 and +5v rectification components arranged to show you how it was implemented. might have swapped some around, can't remember, they are very similar

The two active components next to each rectification assembly are sense/feedback/whatever controls. Probably +5v and +3,3 V regulation as well, as the output of the trafo had to be split accordingly.

The small pile of components is -12v rail and some other components (like a diode between +3,3 and +5v rail )

[momaka]

Re: Corsair CMPSU-550vx / vx55w hack

Actually I have already stripped the psu. Unfortunately protections had to go as described before.

Ah okay, I got you now. I thought you wanted to modify the secondary protection circuitry without removing the original ICs - in that case, that would likely have been very hard.

If I understand it correctly, the route you took was to just use a 431 shunt to regulate the 12V rail?

As for the fan, does the standby power supply have a secondary auxiliary rail? If the fan isn't too big, you might be able to use that. Or maybe have the fan running on 5V if it's a fairly powerful fan.

In the first 2 pictures you can see +3,3 and +5v rectification components arranged to show you how it was implemented. might have swapped some around, can't remember, they are very similar

The two active components next to each rectification assembly are sense/feedback/whatever controls. Probably +5v and +3,3 V regulation as well, as the output of the trafo had to be split accordingly.

That explains a lot. So the 3.3V and 5V rails use what is called a "mag-amp" regulator. Almost all power supplies have used this for decades now for the 3.3V rail only.
So that is a pretty decent design there on CWT's part.

[domas]

Re: Corsair CMPSU-550vx / vx55w hack

Ah okay, I got you now. I thought you wanted to modify the secondary protection circuitry without removing the original ICs - in that case, that would likely have been very hard.

If I understand it correctly, the route you took was to just use a 431 shunt to regulate the 12V rail?

As for the fan, does the standby power supply have a secondary auxiliary rail? If the fan isn't too big, you might be able to use that. Or maybe have the fan running on 5V if it's a fairly powerful fan.

Protections:
Yes, the protections had to go. As I have told, this chip really needs 12V, and it would shout down the supply if the voltage supply was higher and/or lower. It is easier to use lm339 or similar (maybe 431 voltage reference or just a zener) to adjust to max Voltage of supply to disable PSON in case voltage is exceeded. OC protections are not really needed anymore. UV protection had to go, as the supply will be adjustable from very low voltage.

Re: regulation. There already is an optocoupler with 431-alike component. I will have to re do the voltage divider by changing few resistors and adding a trim pot.

Re: fan
I just want to keep the original fan control really. I will find a small 12V or adjustable dcdc chip (i have used it in 2 adjustable supplies like here: https://www.badcaps.net/forum/showpo...2&postcount=13)

it needs only 4 ext components so it is easy. And I have a separate tap with lots of voltage. In case it is not enough, I will use 5VSB with boost converter.

After that, the original fan speed control circuitry will take care of the rest.
Probably 5V would be enough but.. I want to do it properly - not sure why though. Probably only because I can?

And no, this time stby transformer does not have an aux rail. Only two pin outputs followed by transistor or voltage ref for 5V

That explains a lot. So the 3.3V and 5V rails use what is called a "mag-amp" regulator. Almost all power supplies have used this for decades now for the 3.3V rail only.
So that is a pretty decent design there on CWT's part.

Mag-amp.. ok thank you did not know that term. But I know what you are referring to.

[Behemot]

Re: Corsair CMPSU-550vx / vx55w hack

Magnetic amplifier it is.

As for regulator, there are all-in-one regulators which do not need ANY external component, also capable of some 1 A or so, for a long time. These days there is many buck/boost converters-in-a-chip appearing.

[domas]

Re: Corsair CMPSU-550vx / vx55w hack

Magnetic amplifier it is.

As for regulator, there are all-in-one regulators which do not need ANY external component, also capable of some 1 A or so, for a long time. These days there is many buck/boost converters-in-a-chip appearing.

Yes, I agree, but I will just use whatever I can find in my trash pile

Actually, CWT has used mag amps for all 3 main rails.. including +12v. but that one might stay with sense wire, not sure how adjustment will be implemented.