ATX Conversion Project - Load Resistor Hot!

[aaronsnoswell]

Hi all!

So I'm doing the old ATX power supply conversion project - making a lab bench power supply from an old ATX supply. My project has gone pretty well so far, last night I turned it on and got the correct voltages everywhere My problem is this;

At the moment I have a 10Ohm 10W Power Resistor going from 5V to GND to 'stabilise the voltages' (several guides I read suggested this). This resistor is heating up (my temperature probe reads up to about 50 degrees Celsius and increasing fast), before I turn it off. The fan doesn't seem to be able to cool the resistor down (the resistor is zip-tied to one of the heat sinks, so the fan turns on when the heat sink gets to about 40 degC).

Should I change the resistor to go from 3.3V to GND? Or should I just remove it all together?

Photo attached for reference;

[c4757p]

This resistor is heating up (my temperature probe reads up to about 50 degrees Celsius and increasing fast)

50°C is nothing. Let it go and see how hot it gets. It's just a power resistor, 80°C is nothing. 100°C is barely something. I would suggest that a 10W resistor can handle 2.5W without much trouble.

[digsys]

As noted, it's fine. The thing that would concern me is the plastic cable tie ! With constant heating, it will very likely
come apart. I'd find a safer method to attach it to the heatsink.

[SeanB]

Those resistors will happily reach 150C at full power, so they need to be not enclosed and have free air movement around them Plastic ties will definitely not work, you need to have metal ties or a screw holding a metal clip to it. As well keep it away from any wiring as it will melt it.

[nctnico]

I wouldn't attach it to a heatsink but let it stand in free air. Now its heating the rectifiers which shortens their life. Did I mention that ATX supplies make poor bench supplies? The current they can deliver is too large. A short means lots of smoke and molten wiring.

[dr.diesel]

Don't turn it on, take it apart.
Put this in take that out, switch the parts all about.
Push the button smoke rolls out.

Isn't that how it's done?

But seriously, 50C is a-ok.

[Paul Price]

Doesn't anyone ever use Ohm's law? A 10-ohm resistor across 5V will only dissipate 25/10 or .2.5-watt.

In other words it should not get too hot.

My best guess is the 100-ohm resistor is not a 100-ohm resistor.

[c4757p]

That is correct, he said it was a 10 ohm resistor.

Doesn't anyone ever read?

[Paul Price]

Ok, my astigmatism is at fault, so it is 2.5 Watts, but the 10-W resistor should not get very hot with that amount of power.

The resistor be tied to the heatsink with plastic tie-wraps  for two reasons.

Firstly the resistor body thermally hides the heat sink from fan air and whatever semiconductor devices attached to it are what is getting hotter and would do better without this barrier to fan air cooling.

Secondly the ties will melt at a rather low temperature. The 10 watt resistor body in this case is best cooled  hanging by its own wire leads in free air.

[aaronsnoswell]

Thanks all for the feedback - I didn't even think about the zip ties getting too hot!

I've moved the resistor to the 3.3V line for now - even if 150degC is nothing, I'd still rather not have that much heat building up inside.

Appreciate all the comments.

[Matrix828]

Thanks all for the feedback - I didn't even think about the zip ties getting too hot!

I've moved the resistor to the 3.3V line for now - even if 150degC is nothing, I'd still rather not have that much heat building up inside.

Appreciate all the comments.

Dont forget that ATX power supplies (depending on the version) need a minimum load on the 5V or 12V lines to even turn on, so im not too sure moving it to 3.3v would still let the power supply work as a bench supply.

[aaronsnoswell]

Thanks - I think might mine might be a different spec - it seems to run fine without any loading (with just the 3.3V load I mean).

[bitluni]

Since I have also one of these on my bench I have another question:

Does anyone know how the main board turns off the PSU? I wasn't able to find any information about this.

By the way you need to connect the resistor to 5V

[dr.diesel]

To turn on an ATX power supply you ground the Green wire, let it float and it will turn off.

[metalphreak]

A lot of modern power supplies don't actually need a load resistor on the 5V/3.3V rails any more. Older computer designs were quite heavy on the 5V/3.3V rails, but most are using 12V now with DC-DC near the CPU on the mainboard instead. Try it without the resistor and check the voltages on the other rails (use a scope if you have one while its under some load on the rail you're testing).


http://www.seeedstudio.com/depot/atx-breakout-board-bench-power-supply-p-1222.html?cPath=1_4

This, for example, doesn't come with the load resistor soldered in, because it's not necessary most of the time.

[bitluni]

To turn on an ATX power supply you ground the Green wire, let it float and it will turn off.

ok.. this seems to work just random on this cheap old one I am using...  Since I just have used a push button to turn it on.. it stays on most of the time even the wire is floating.. Sometimes it turns off, as it should regarding your explanation. But it's one of the first ATX PSU on the market so the "green" wire isn't even green. But it also has -5V, which the new ones doesn't have. Since it also has a main switch built in, I think I will just solder the green wire to the GND and just use the mains switch. 

[Nerull]

Ok, my astigmatism is at fault, so it is 2.5 Watts, but the 10-W resistor should not get very hot with that amount of power.

The resistor be tied to the heatsink with plastic tie-wraps  for two reasons.

Firstly the resistor body thermally hides the heat sink from fan air and whatever semiconductor devices attached to it are what is getting hotter and would do better without this barrier to fan air cooling.

Secondly the ties will melt at a rather low temperature. The 10 watt resistor body in this case is best cooled  hanging by its own wire leads in free air.

Having done a similar build, 2.5W is enough to get one of those 10W resistors pretty hot.

I stuck the resistor to the side of the case with 3m thermal adhesive and it seems to work pretty well to spread some of the heat.