Need to switch an ATX power supply every 20 min...

[prosper50]

Hello everyone!

I'm brand new to this forum, and I am hoping to acquire some assistance.

I need to design a very simple circuit that will switch an ATX power supply on and off. It needs to cycle every 20 minutes (running for 20 min, off for 20 min, back on for 20 min, back off for 20 min, etc.)

The power supply will be plugged into a computer, as one would expect, and the computer will be running a burn-in test. The goal is to stress the pc, let it cool, stress the pc, let it cool... etc.

The cycle should be indefinite - we will likely run this in 24 hour intervals.

Like I said, I am looking for a simple and small solution - maybe something where I can remove the power switch of the ATX power supply, and replace it with this simple circuit. Or, I have seen postings where people send trigger signals to the power supply. However, since the supply will be plugged into a PC, this could be a problem.

We will need a very cheap and simple circuit - nothing that is microprocessor controlled. Perhaps a 555 and a relay, or something to that effect??

Can ANYONE think of a good solution to this?? I don't have a ton of electronics background, so if I can save a few days of scratching my head, I would definitely appreciate the assistance. Thank you all!!

-Dave

 

[Harald Kapp]

Welcome, Dave, to our forum.

ATX power supplies have a power-on signal/pin. This is normally used to turn the PC on or off using the front button on the PC. It can e "abused" to control the power supply externally. This instructable shows how to use a manual switch (or read this discussion). You can replace the switch by a transistor (NPN) or a relay which in turn would be controlled by a timer (a 555 is not very suitable for 20 min. intervalls, althought it will do as long as you don't expect precision timing).

This ressource shows you how a transistor can be used to turn a load on or off, this same circuit can be used for the power on.

This ressource gives you background information on timers, specially for long times.

Harald

 

[Gryd3]

Make sure you set the BIOS to Resume last state on power-loss so it will turn back on. (But I didn't need to tell you that )

This is the first thing that popped into mind... Keeping it simple.
http://www.homedepot.com/p/Good-Choice-15-Amp-24-Hour-Plug-In-Dial-Lighting-Timer-602/203729962

Edit: Bad example sorry... has to ground...

 

[prosper50]

Thank you both for your responses - the knowledge is appreciated!! Harald, your references were fantastic - I think I may be able to put something together from that. I'm not skilled enough to be able to piece together a working circuit from an NPN or the 555 timer, but I'm sure with the info provided I can figure something out!

-Dave

 

[prosper50]

Harald, upon further investigation, I have a followup question...

The Instructable you provided describes how to turn on an ATX supply using a switch. I have used the jumper wire technique many times, myself. The problem is, that header will be connected to the motherboard. So in lieu of shoving a wire into the green and black header, how else might I trigger the same signal?? Is it suggested to tap the green and black wires above the plastic header, and apply the signal/voltage that way?? I would ASSUME that is not recommended, but I'm curious how else I would trigger this if it is plugged into the motherboard already.

THANKS AGAIN!

 

[Gryd3]

Harald, upon further investigation, I have a followup question...

The Instructable you provided describes how to turn on an ATX supply using a switch. I have used the jumper wire technique many times, myself. The problem is, that header will be connected to the motherboard. So in lieu of shoving a wire into the green and black header, how else might I trigger the same signal?? Is it suggested to tap the green and black wires above the plastic header, and apply the signal/voltage that way?? I would ASSUME that is not recommended, but I'm curious how else I would trigger this if it is plugged into the motherboard already.

THANKS AGAIN!

The best approach to that would probably to cut the enable wire. Take the end from the PSU and use it as an 'input' to a transistor, or relay. Connect the 'output' to the pigtail left on the main-board connector.

 

[prosper50]

Aaaahhh - OK, so I don't need BOTH the green and the black wires if it is attached to the motherboard? I can just use the green wire and place an inline transistor (or relay?) above the motherboard connector...?

I hope I understood that correctly.

 

[Gryd3]

Aaaahhh - OK, so I don't need BOTH the green and the black wires if it is attached to the motherboard? I can just use the green wire and place an inline transistor (or relay?) above the motherboard connector...?

I hope I understood that correctly.

The green wire is not always green but yeah, it typically is.
When the wire is disconnected, the PSU remains off... You can interrupt this connection to turn the PSU off with a relay or transistor.
This wire requires that it gets pulled down to ground to turn on the PSU.

 

[prosper50]

Got it! OK. Thank you for the clarification.
We have dozens of the same PSU around here, and the signal wire is green.... but I will have to keep an eye on the coloring.

I think I will cut some wires and experiment before creating a circuit - that way I have a more clear understanding of what needs to happen to get it to turn on and off.

 

[BobK]

Presumably, there is a switch connected to the motherboard via a header that is used to start it up. This is where you should connect. No need to cut any wires at all. The transistor can even be wired in parallel to the switch and either one will start it up.

Bob

 

[prosper50]

Aaahh - yes, the motherboard standard power pins. I hadn't really thought of that. Since we swap motherboards so often (for burn-in testing), we usually just use the PSU switch to start and stop. But yes, there is the header that I can use as well.

 

[Harald Kapp]

Cut the wire and connect it to your timing circuit's output (relay or transistor). Here is a basic schematic. You may have to change the values for resistors R2 and R3 as well as capacitor C2 to match the timing to your requirements.
Do not use R4, it is there for simulation purposes only.
Use any off-the-shelf NPN transistor, e.g. 2N2222 for Q1.
Instead of the 12V you can use the 5V ATX standby power (purple wire). You'll have to tap into this wire somewhere along its length since you may nedd this voltage on the mainboard, too, so you can't just cut it.

 

[prosper50]

Another great reference, Herald. Thank you again for this information!!

 

[BruceS]

I know this forum is all about building up 'projects' and not buying them off the shelf but what about this item?
One program will allow indefinite cycling for set times (on & off) once initial pulse/trigger.
http://www.ebay.com.au/itm/200973250865?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1497.l2649
Just cut your active AC wire & feed it through the relay. NO of course.

 

[Harald Kapp]

This module will most liekly allow for much more precise timing as the 555 circuit - at least when it comes to such long times.
The drawback of the module is that you will require an additional 12V power supply whereas the 555 circuit can be powered by the PSU's own 5V standby power.

Just cut your active AC wire

That's technically possible but ill advise for the unexperienced. But the module can be used to switch the ATX power-on wire to ground, too. That's on the low-voltage side of the PSU and therefore muuuuuch less dangerous.

 

[prosper50]

That is a fantastic find, BruceS!!!
I actually checked ebay before posting this, but because I didn't know exactly what I was looking for, I came up empty.
I will have to look into this.

I am still a bit confused as to WHERE this would fall in the setup. Are we still talking about the "Green" signal wire here?? Would that green wire plug into this module as in input? I am thinking that is NOT what is described.... I would imagine that the output from this module would jack into the existing signal wire, so that it's output would triger the signal wire?? Is my logic accurate??

It looks (to me) like this is a self-contained unit, and once it is determined WHERE it goes, it can simply be plugged inline. Correct? It has the relay built in, so I wouldn't need any additional components except for an AC adapter, which is no big deal. As long as this does what I need it to do, it can take up 10 plugs if need be. LOL.

Thank you again for the advice and clarification!

-Dave

 

[Harald Kapp]

You connect the "green wire" to one contact of the relay, the other contact of the relay to ground of your ATX power supply. The relay merely replaces the switch that was used in the instructable I first linked.

 

[prosper50]

Got it! That's fantastic information. Thank you again - I am going to explore this module a little further now!

 

[Harald Kapp]

Note that this setup will have the power supply on mains voltage all the time. So take all necessary safety precautions.
Also note that this test will stress the power supply in a different manner than disconnecting it from mains, since the PSU is always on mains and is active to produce the standby power.

 

[prosper50]

Yes, that is definitely a concern worth noting, and something I will have to look into. Because not too many hands will be in this particular pot, I don't think it should pose a hazard, but it does add a layer of uncertainty to the process.