Repurposing an ATX PSU based on DR0183

[Th3_uN1Qu3]

Here's what we have on the dissection table today...

JNC LC-A300ATX
Platform DR ASIC ATX Ver:1.0B

The custom DR0183 controller works at 60kHz based on info i've found on the net, see attached archive for more photos and partial schematics. His version was rev "A" though while mine is rev B, mine is a bit better built and more honestly rated. It still had crap secondary caps unfortunately (JEE, all bad). Main transformer (EI-33 ASB) had 19+20 turns primary. Together with the 1uF coupling capacitor, this resolves to just under 200mT flux which is appropriate for this core at this frequency and power. 3+3 turns for 5v, another 4+4 turns in series with 5v winding for 12v. So we have 3 turns for 5v and 7 turns for 12v.

30A schottkys for 3.3v and 5v, 10A fast recovery for 12v.
Label was honestly rated if you want to believe this...
3.3v @ 20A, 5v @ 30A, 12v @ 10A, -12v @ 0.8A, -5v @ 0.5A, +5vSB @ 2A
Max combined load on 3.3v and 5v: 160W, 300W total.

It also has a temperature controlled fan. Hey, they didn't cut so many corners on this one.

I proceeded to remove all the wiring, the bad caps, the secondary rectifiers, the secondary heatsink, the inductors, the little diodes for the negative rails, the main transformer, one of the TL431s... And finally the controller as well. Sharp eyes will see that i have also replaced a bad cap in the standby section. And believe it or not, i could strip even more stuff out and still have the standby section work, but i can't be bothered.

I left the standby supply intact as we're gonna need it. Btw, the standby section on this unit provides two voltages, 5v and 16v. The presence of the 16v rail is welcome, as that's what we'll use to feed our TL494. If you feel like adding some of the missing input filtering, now is the time to do so.

Next step, coming soon...

[Th3_uN1Qu3]

Re: Repurposing an ATX PSU based on DR0183

Take this typical schematic of an old 300W ATX PSU. Look over it again and again because this is a typical design and you're gonna see it a lot. The controller IC is a SG6105 in this one, but it doesn't matter, because the SG6105 is just a TL494 with some extra stuff built-in, like a couple voltage references and a bunch of comparators (for overvoltage/undervoltage protection). What i'm trying to show you is something else.

If you look at the comparison between the two driver sections you will notice that they are more or less the same. What that means is you can remove the proprietary controller and drop a TL494 in there and get whatever output you desire. 2003, 2005, AT2005B, DR0183, Linkworld, Weltrend... It doesn't really matter because ALL halfbridge PSUs are the same in the driver and power sections, and the proprietary controllers can be replaced by a TL494.

Add a typical comparator like the LM339 and a bunch of resistors, et voila, you also get protections. Actually the TL494/LM339 design was found in many, many PSUs, with another common version being just the 494 and the protections implemented discretely with a bunch of NPN transistors (usually 2SC945) and 1N4148 diodes.

[everell]

Re: Repurposing an ATX PSU based on DR0183

Are you saying that all of these pwm chips are pin for pin compatible?

[Th3_uN1Qu3]

Re: Repurposing an ATX PSU based on DR0183

Unfortunately not - they even have different pin counts. What i'm saying is there's nothing keeping you from putting a 494 on a daughterboard and wiring it into the circuit.

[Th3_uN1Qu3]

Re: Repurposing an ATX PSU based on DR0183

Got a little more work done. Installed the new secondary heatsink and diodes, and i've also gotten around to moving the fan circuit to the 16v rail of the standby supply (since the 12v rail it was wired to will become a 60v rail. ) That of course means the circuit will have to be reworked a little to accomodate the higher supply voltage.

I have also found that the schematic i found on the internet and posted earlier doesn't match my version. Mine is slightly different, see attachment. Btw, the transistor is a 2SD882 not a BD139 (that's just what i had of roughly the same type in LTSpice). If anyone wants to simulate it in LTSpice, the simulation file is attached.

The NTC wasn't even fitted to a heatsink, it was sitting next to the 3.3v inductor. But anyway. Did everybody figure out how to decrease the initial speed of the fan? Yup, that's right - add a resistor in series with the NTC. The resistor will have to be selected in such a way that the NTC can still ramp the fan up to maximum speed when it gets hot.

Edit: Just did a little testing: With theFan starts at 10.4v with the NTC at 26C, then the NTC self-heats to roughly 31C, but by this time the fan voltage has already become 13 volts. It reaches 15 volts with the NTC at 40C, and it changes very little above that. I'm too lazy to do resistance measurements on the NTC, but anyway, it is too sensitive. I'll want the fan to start at around 7 volts, and only get to highest voltage at 70C or so, measured on the hottest heatsink (obviously, that's where i'm going to place the NTC as well).

Edit 2: My bad, there is actually a 6k resistor in series with the NTC, just like in the original (it might as well be the same 6k2 but i'm too lazy to measure it out of circuit), and the zener is 5v. It's the zener which actually gives the fan that minimum voltage - so for a 12v supply it appears the fan would have actually started at 7v. Given the layout, i'm not sure how much time it would have spent being that slow though.

Edit 3: Removed the zener (no more minimum fan voltage) and replaced the 6k resistor with a 10k. This brought things just where i wanted them. With the NTC cold, the fan starts under 4v, but already gets to 8v by the time it reaches 30C, reaches 12v at ~60C, and then kicks into overdrive with 13.8v at 70C (this is a 12v fan, remember). Perfect.

And if you're wondering how the fan starts with under 4v, there's that little 10n capacitor parallel with the NTC. What that does is create a time constant, during which the NTC is effectively shorted and the fan runs at full blast. It also filters out some of the nasties the NTC may pick up, so that the variation in fan speed is relatively smooth. Now, 10n and 4kOhm is just 40 microseconds, so that is a very small spike, but it seems to get the fan going just fine. However, what i will do is increase that time constant by a few orders of magnitude - i will put a 100uF capacitor in place of the 10nF, making the time constant 400 milliseconds or 0.4 sec, to make sure the fan always starts properly, and to be able to hear any signs of stress from the fan as i start the power supply, so i don't get the surprise of the fan failing when i'm pulling a few hundred watts.

This is of course valid only when the NTC is cold, as temperature rises and its resistance decreases, the time constant decreases accordingly.

[Th3_uN1Qu3]

Re: Repurposing an ATX PSU based on DR0183

Anyone asked for pics? You can see the new heatsink and diodes, the insanely low voltage that the fan runs at when idle (btw, i went for a 470uF cap eventually, upping the time constant to 1.8 seconds, and i left the wires long on the NTC so that i can fit it on either primary or secondary heatsink, whichever will happen to run hotter), and a partly assembled TL494 daughterboard.

It occured to me just now that running the fan from the standby supply gives a huge advantage - the fan will keep cooling the power supply even after it is turned off, and at 3v it is quieter than a whisper, so i don't see why i shouldn't keep it running at all times.

[Th3_uN1Qu3]

Re: Repurposing an ATX PSU based on DR0183

Alright... Too bad i didn't catch some pics.

I put everything together. I used my old transformer and 1.1mH coil that i made for a 25kHz PSU which blew up. No modifications were necessary, since as you go up in frequency you can lower the number of turns, if you have too many the transformer actually works more relaxed, the only risk being copper loss too high. The wire was thick enough, so it was okay.

I still need to work on compensation because it hissed angrily at me all the time, but for a few minutes i got 560W out of it. Then, something stopped. I turned around and thought the overtemp protection kicked in. Nope, supply was still running, but the amp meters showed zero.

Guess what happened?
.
.
.
.
.
.
.
.

I blew up the load resistor.

Remember my toaster? It was rated for 800W @ 230v. It has 3 banks of wirewound resistors in series, 20-some ohms each. I was only using one bank so i can have the resistance low enough for this supply voltage (+/-60v maximum). It turns out it couldn't take the heat.

I did get some numbers, efficiency was around 87% at this point. At 300W i was close to 94%. As the load gets higher, efficiency is inversely proportional to the temperature of the primary transistors. As in, the hotter they get, the worse the efficiency. If not cooled adequately, this continues until the transistors blow up. They are genuine Fairchild MJE13009s in TO-247 btw. In this supply the primary heatsink is isolated from the mains, so i stuck both overtemp and fan control thermistors in it. Works a treat.

[Th3_uN1Qu3]

Re: Repurposing an ATX PSU based on DR0183

The resistor wasn't damaged, it was the wire which melted off the connector. Here's the latest progress, pic taken with my trusty ole' N93i, coz mom left for the weekend and took the camera. You can still make out what the meters say though. You do the math. My other meters are out of battery so no line voltage readings, sorry. Just assume it's 230v. It's likely a tad lower, but make it 230v so you can't say i'm overrating the efficiency.

The power supply has been running like this for an hour already. Enough to make it pretty darn hot in here. And the thermally controlled fan isn't even at full blast yet.

You said my test bench is messy? You can say that again.

[ss627]

Re: Repurposing an ATX PSU based on DR0183

Can you replace secondary with these?

[Th3_uN1Qu3]

Re: Repurposing an ATX PSU based on DR0183

Only if what you're planning is less than 30 volts output, reverse voltage rating is low. They'd make one hell of a 12v supply tho. Here's what i have in mine: http://www.datasheetcatalog.com/data...YW29-200.shtml