Variable pc power supply mod overvolt shutdown

[dr pepper]

Take a look at this schematic of a pc power supply I want to convert into a variable supply for testing heater elements.

https://www.eejournal.org/wp-content/uploads/2010/10/SMPS-IC-2003.gif

I have pulled r60 and r62 and replaced them with a 200k pot, it works well the 12v output now goes from 2.5v to 14v and then the overvolt shtdown comes in as expected.

So I've broken the tracks on the board to the 2003 driver chip, on pins 3,4 and 6 thinking that would remove feedback and stop overvolt shutdown and I'd be able to go to 16v or over, but it doesnt the chip still shuts down, if anything at a bit lower voltage, why?

The only thing I thought of next was to ground those pins I disconnected to see if its just picking up noise on those pins causing the shutdown.

Yes I am aware of the high voltage dangers around switchers.

 

[alec_t]

but it doesnt the chip still shuts down, if anything at a bit lower voltage, why?

For a fixed load resistance the load current will obviously increase with output voltage. The chip (I haven't checked its spec) may well go into internal shut-down if its current/dissipation is excessive.

 

[dr pepper]

The only load connected at the time was its case fan, maybe I should repeat the test with a load, I think I've a projector bulb somewhere.
It doesnt look as though theres any specific circuitry for overcurrent, I'd want to keep overcurrent protection, just do away with overvolts.

Edit: put the board back as it was, didnt work at first I had a bad joint, all back together all seems fine, however if I disconnect the feedback resistors r60 and r62 and put a 200k pot accross the 12v side r60 I can vary the output from 2.5 (vref) to 14v, however as soon as I load the o/p even with 1 amp the supply shuts down. really difficult to suss out, I think pin 5 has something to do with overcurrent, there dosent appear to be anything on the o/p current sense wise so maybe the psu measures the drive current to the switchers, or maybe there is no overcurrent circuitry it just relies on the o/p tranny resistance or something freaky.

 

[dr pepper]

Shelving the 2003 based pc power supply I routed through my scrap box and found an antique 230w c power supply, pre P4 connector.
It has a Tl494 (making mods so much easier) and a weird device marked LS 168A that I cant find any data sheets for, I think its a psu supervisor, it has tracks to the 3.3, 5 and 12v outputs, and tracks to the Tl494, I tried breaking the track that goes from this to the '494 to see if I could turn the handy trim pot on the board so the o/p would go past 12v, didnt work.
So fix it or break it I just unsoldered the LS168, and tried again, the supply powers up without the green power up wire grounded as hoped, and I can wind the o/p up to 15v where the tranny goes nuts and the supply outputs 30v, probably because the '494's feedback pin is grounded by the pot, the manufacturers probably thought it'd never get to ground before the supervisor shuts it down.
So this idea looks as though it might work, but if the overcurrent protection is also done by the '168 then it isnt exactly going to be reliable, also this psu doesnt give more than 8 amps on 12v, I really need 15.
Still searching for data on the 2003 ic in my other psu, or even the 1001 in a nicely made dell psu I just found.

 

[alec_t]

I'm guessing that pins 3,4,6 of the 2003 need to remain connected so that the IC knows the rail voltages are good. If it thinks a rail voltage is absent or wrong it will probably shut down (or at least cripple) the PSU. Maybe that's the effect you're getting.

 

[dr pepper]

Yes I think its because I've connected voltage feedback to the 12v side only, as soon as I put a load on the psu the controller responds with a higher duty cycle to maintain 12 volts, but at the same time the 5v rail will naturally go up, probably to the point of shutdown.

So I spose its all down to fooling the voltage monitor side of the chip that all is ok.

 

[alec_t]

Have you got room for a few zeners to hold the chip inputs at pins 3,4,6 at their expected volts?

 

[dr pepper]

Yes I'd thought of that, I need to test the supply again to see if it'll shut down due to undervolts, I dont think it did, but I've messed with it a lot and cant remember.
If it doesnt lock out through undervolts then yes thats a good idea, a 12v zener and a resistor divider to generate 12v, 5v and 3.3v from the 12v line.
I'll see if it locks out on undervolts and if not I'll try that.
There is a totally seperate 5v off line isolated supply just for the controller chip, I could chuck a joule thief together to generate 12v fairly easily from that if the chip locks out on undervolts.

 

[alec_t]

I've messed with it a lot and cant remember.

That has a familiar ring to it. On more than one occasion I've intended to be methodical: vary one parameter, check result, vary another parameter, check result, etc, but I've ended up varying several things at once and then not knowing which one caused which effect .

 

[dr pepper]

Heres a nift datasheet for those messing with later pc power supplies, its for the sem2005, dunno if its similar to the 2003 or not,

https://www.google.co.uk/url?sa=t&r...moDIDA&usg=AFQjCNE4jt5nvhiYAHD0UXqMbNKN4_iWZg