Arrgh! Bestec ATX PSU's...Dangerous to any system (So it seems)

[Hodgkinson]

Doh.
I've been using a Bestec 250W ATX PSU in my ATX shutdown modification project unaware that:

Sub-standard caps in these PSU's fail, causing +5v to go to +12v and fry any system that is connected :-o ;

And an unusually high +5v STBY rail at ~8v is in fact a result of fried, gone-low-resistance glue, probably what resulted in me ending up with the PSU in the first place.
(I noticed this, assumed that PC mobo's had extra regulation, and slapped a 5V1 zener & resistor inside the PSU to regulate the rail)

Just do a Google search for these PSU's and you'll see what I mean...

Right. I suppose that im going to have to find another PSU :-)

I’m looking for one that is 250W or more, monitor pass-through if possible, standard 1x fan on rear plate (Preferably with a non-raised grill) and sufficient space inside to mount all the necessary extra gubbins that I've built (As per my photos) - Which may affect whether a PFC transformer can be used.

Any suggestions? I might be able to get a second hand Chieftec 300W PSU with the above specs - Any bad experiences with these PSU's?

TIA,
Hodgkinson.

PS. For those of you watching the ATX mod progress, im slowly working on the switch panel in between coursework assignments ;-)

[Oliver]

Hodgkinson wrote:
...slapped a 5V1 zener & resistor inside the PSU to regulate the rail...

Hi,

Are you sure you're using a zener correctly? I know you have some expertise with electronics, but from what you've written, I wonder if you may be trying to use a zener to provide regulation of an output rail. That is unlikely to work for the output power of your supply. I haven't used descrete zeners for some time, but I think they would generally only be used to provide a reference voltage for a negative feedback output level monitor. The error signal would then be used to regulate the output power transistors.

Sorry if I've misunderstood, but thought it worth mentioning.

Cheers,

Oli

[Floid]

The Fortron Source "Green" (-GLN) supplies have been treating me well and are highly affordable.  You get active PFC if you care and a very quiet fan that's been much more durable than the one in the otherwise-okay, even-cheaper Coolmax I'd been using prior.  

I'd suggest double-checking for similar failure modes before using them with irreplaceable CBM hardware.

(Newegg product link for USians.)

My one quirk with the FSPs is probably unrelated to them -- they regulate well enough as far as I'm concerned, but for whatever reason vDIMM on a Foxconn Socket AM2 board used with one came out to 1.75v and required a 50mV "boost" in the BIOS options to get it stable at spec.  Probably more a matter of the board design, and yes, 50mV matters that much to DDR2.

[Hodgkinson]

Oliver:
Thanks for the tip.
The problem is that the +5v STBY rail starts off at 8v (Which it shouldn't...) then gradually floats down towards 5v, of which the difference then becomes less than the required minimum voltage difference for any conventional linear regulator to operate. Unfortunatly there isn't a sense rail for any STBY's - Sense rails are only on the high current +5v and +3.3v rails.

As a result, I have a 6.8R resistor feeding from the supply into a heatsinked 5W zener (To ground) to provide a 5-5.1v feed with sufficient current to feed the TTL 74LS02 latch circuit for controlling the PSU; without the voltage drop problems of a regulator or buffer transistor.

If you have a look at my photo the zener is just visible before the 3x LM317's on the white PCB. The feed resistor is probably the largest one on the board, just below the cap at the edge of the terminal block.
The zener does get warm within the first minute of PSU turn-on before the STBY voltage drops to a more reasonable level.

Floid: "irreplaceable CBM hardware."
You can read my mind :-D I'll have a look, but I’d like to find a PSU from a local shop rather than having to ship it half the way around the world. Some local brands include Q-Tec and CaseCom.

[Hodgkinson]

I've just had a look at the link :-)

Unfortunately I’ve got a huge amount of wiring, heatsinks, PCB's and brackets to cram into the PSU as well, and any PSU's with a large fan on the inside-case edge will simply take too much room up. Unfortunately, as much I’d of liked to go for a large, low speed fan approach, it means I can't fit all the gear in - Its got to be one of the old style ones. Thanks for the manufacturer recommendation though :-)

One idea was to take the PSU fan (On my original PSU), turn it around, and bolt a commercial fan-filter cover to the PSU case. However, I figured that this *might* be risky as originally straight non-turbulent air was drawn through the PSU from what would be the inside of the PC case, past the power heatsinks, then through the fan and out of the PSU. I figured that turning the fan around would result in turbulent air in the PSU, and thus might adversely affect cooling. Any ideas or thoughts on this matter?

[Braddo]

Well this is the last thing I needed to hear. I have just about finished building a supergun using a 80w AT Bestec PSU.

I certainly hope this problem doesnt affect my model.

[Hodgkinson]

From what I've read most people seem to refer to the ATX 250W versions. Although I've not read anything major referring to other wattage ratings, unfortunately I can't say that they're not affected as well (Most manufacturers use the same parts, and often the same PCB's in multiple different products...Remember that GFX card that just needed a link bridging to upgrade it to the next model?)

EDIT: You *might* get away with an AT PSU. Assuming that it’s older than the average ATX PSU, then you might have got one from before the bad caps started to be used.
FYI: By "Bad caps" I mean electrolytics with a incorrectly-copied electrolyte that - Apparently - Releases hydrogen which in turn results in the all-too-well-known bulging cap syndrome.

Anyway, I hope that someone reading this thread might have found it of some use.

I’m still open to suggestions of "Good reliable" PSU's to continue the ATX mod project with ;-)

TIA,
Hodgkinson.

[Oliver]

Hodgkinson wrote:
Oliver:
Thanks for the tip...

Hi Hodg,

Sorry, I misunderstood what you were doing. I didn't want to come off as condescending. Anyway, it seems you have paid due attention to your design.

Hope it works out for you.

Regards, Oli

[Framiga]

Hodgkinson

if this can "help" you, i can say that even some pricey PSUs acts very weirdly when used on our amigas.

I'm just "fighting" with the n° PSU (Seasonic 400 w) purchased brand new from RS.

These PSU are designed for a (no need to say to you) crap power demanding by GFX card on the 12v rail and above all, they have a totally unused by us, 3.3 volts rail that summed on some "hidden" sense wires, lands in a totally screwed/unbalanced power balancement.

In few word, i have 4.90 volts (on the +5 v rail) and 12.47v on the 12 volts one. (the funny thing is that i have 3.300 volts on the 3.3 rail perfectly stable even with a low load ...lol!!!) and all this in a quite loaded A4000 in a PowerTower.

I've fortunately found a way to re-route the fisking mis behaviour that i'll let you know, as soon as finished (my oscilloscope is under repairs .. hugh!)

Sorry for the demiOT :-)

[Hodgkinson]

Those voltages for +5v and +12v that you mentioned seem to be quite normal from what I've seen by DMM or in the BIOS.

I’m sure that you're already aware of this, but you've got to be careful with those sense wires. If a sense wire ever gets left disconnected, your PSU will think that there is a huge load on its associated rail, and will try to crank that rail up to restore the voltage at the load end of the wire.
Without the sense wire connected anywhere, this usually leads to blown up PSU's and hardware.

PS. Digital DMM's don't always seem to be accurate. I've got a cheap one that’s 0.2 - 0.3v out on the DC ranges for some reason (Compared to a Fluke DMM) - Enough stop something from working either by frying it or not giving it enough supply.

:-) Duff oscilloscope? We fixed one by giving it mains derived EHT :-o The smoothing cap on there would be enough to cook a Sunday roast :crazy:
EDIT: 40uF at 5kV (*Breaks out slide rule*) thats 1000J ... Ouch ...

Hi Hodg,

Sorry, I misunderstood what you were doing. I didn't want to come off as condescending. Anyway, it seems you have paid due attention to your design.

Nah. I’m glad that you did :-) I had a friend who used a zener to clamp a signal rail in a SCART system without any series resistor - He’s lucky that his equipment wasn't toasted. By putting in comments like that you're benefiting anyone else who might be reading the thread ;-)

Hodgkinson.

[Hodgkinson]

Right. I've just been rumaging in the multiple boxes of old AT/ATX PSU's that we have and I'd just thought I'd update this thread.

Macron Power 300W ATX - Has a temp. controlled fan and the necessary mains pass through, but most of the caps have gone bloaty-crustyfied. Probably the best bet so far out of the PSU's I have providing I can swap the caps.

No. 1/SOURCE ATX - Works, but charred PCB in areas and cheap looking construction

Q-Tec ATX - Works, but no temp. controlled fan.

I'll do some ringing around to see what else is available locally in the way of new PSU's next week.

Hodgkinson.