AT PSU rebuild

[pdavid]

I noticed that barney is looking for a good AT psu so I decicded to rebuild a generic crapy one I have laying around, just to see if it works out. Might be good for some old Pentium retro rig.
I replaced the EMI filter components, added a 4A graetz bridge, E13007 primary switchers, Samxon 470/200 primary caps, EI35 transformer, MBR3045 for 5 volt and SBL1660 for the 12 volt rail. Added the missing pi-filter coils. Yageo SY and Arcotronics ESX caps.
Had to replace two small 4.7uf and 22uf caps, I put Nippon KME there.
...also I fited bigger heatsinks and a temperature controlled fan circuit. I'll put the pcb into a grey anodized Skyhawk psu case. It has holes for both the AC switch and the periphelial cables. This desing has an LM494 pwm chip an a few transistors for the regulation. I'll add a 220k trimmer to where just two resistors where soldered underside the pcb for voltage adjustment.
I hope my modifications worth it and this thing will live
Also cabe gauge will be awg18 for the periphelial and the AT style motherboard connectors.

[PeteS in CA]

Re: AT PSU rebuild

The big uncertainties may be whether that new transformer is sufficiently similar to the old, smaller, one, and whether those output caps are good enough but not too good (which might mess up the loop compensation). If that's a new output inductor, there may be compatibility issues there as well. But for adding robustness, I think you've accomplished that.

[Th3_uN1Qu3]

Re: AT PSU rebuild

For the inductor i wouldn't worry - i've tested a bunch of output inductors of various sizes from cheap halfbridge PSUs and all had the same values.

[PeteS in CA]

Re: AT PSU rebuild

I remember seeing your post to that effect. The P/S pd rebuilt is nearly 14 years old (the date code on the PCB is "9816", 16th week of 1998). I was just wondering whether output inductors have been more or less identical for that long.

[pdavid]

Re: AT PSU rebuild

I remember seeing your post to that effect. The P/S pd rebuilt is nearly 14 years old (the date code on the PCB is "9816", 16th week of 1998). I was just wondering whether output inductors have been more or less identical for that long.

The supply is an ancient UTT model. The output inductor has the same size as the original. The only differance is that the two 5V windings are a bit thicker. Right now I'm looking for white and blue wires for the AT motherboard connector... though the originals would be fine as well for those.

[Th3_uN1Qu3]

Re: AT PSU rebuild

I was just wondering whether output inductors have been more or less identical for that long.

Considering that the Chinese keep using the same early 90s designs for their cheap PSUs, there's a good chance they've been like that from the get go.

[PeteS in CA]

Re: AT PSU rebuild

The places where P/Ss are built has been shifting over the past 3 decades or so. In the late 70s, Boschert was building production P/Ss in Sunnyvale, CA (Silicon Valley). I'm sure they weren't the only company building P/Ss in the US. Production has since migrated to (then) lower labor cost places like Mexico, Japan and Taiwan, and then to even lower cost places like Thailand, India and China. Putting it in perspective, the target price for power supplies in 1980 was ~$1 per watt. IOW, the cost to the computer manufacturer for a 250W P/S would have been $200-$250; in 2000, Delta Products sold 250W P/Ss to Compaq or Dell for $13 or $14 or to Apple (who had some unique aspects in what looked like a standard ATX P/S) for $14 or $15. And competitors were under-selling Delta on price (but for, then, lesser quality). So when things like inductor inductance values/winding ratios, cores, and transformer ratios and pin-outs became more or less identical is uncertain to me.

[pdavid]

Re: AT PSU rebuild

It's finished! Sort of... I've turned it on, voltages were good (except for -5 and -12, these were low) and I noticed that the fan spins faster and faster. The primary switchers were cool, but the secondary heatsink got so hot I could see the heat. The thermometer read 79C! Also there was an audable buzz coming from the trafo. So I put back an EI-35 transformer and it runs cool now, without load. Maybe I'll take it to work to measure ripple and stuff.

[PeteS in CA]

Re: AT PSU rebuild

Remember that the -5V and -12 may be regulated by 79XX 3-terminal regulators, which are 5% devices. Or if it's a reeeeeeally cheap piece of @#$% those two outputs might not even be regulated.

[PeteS in CA]

Re: AT PSU rebuild

Looking again at the pic, I don't see a 7905 or a 7912.

[pdavid]

Re: AT PSU rebuild

Looking again at the pic, I don't see a 7905 or a 7912.

Hooked up to a mobo and few hdd the voltages are good. Even the negative ones. Still, this psu is a low end at design and parts. A have a Sparkle and a Seventeam unit, now those are well built.