Hi
I needed to control several servo motors with Arduino which needs higher current that Arduino couldn't provide, so I decided to use an old ATX power supply wich can make 5v-20A.
sounds perfect! but when I connect a servo motor sometimes the power supply turns off immediately, if it doesn't turn off the current is so small which can NOT run the servo motor!!!
I guess I'm missing a point! 
You need to short out PS_ON# (Usually Green Wire pin 14) to COM or GND to enable the power supply outputs
--
Mark
Try putting a resistive load on the P.S. also.
~ 4.7 to 2.5 ohms.
.
MarkDerbyshire:
You need to short out PS_ON# (Usually Green Wire pin 14) to COM or GND to enable the power supply outputs--
Mark
Thank you, But already done.
I can't get enough power out of the ATX, That is the problem. 
LarryD:
Try putting a resistive load on the P.S. also.
~ 4.7 to 2.5 ohms..
I've put a 10ohm-10w between GND and 5v terminal, should I put it between GND and PS_ON ?
GND and +5v
10 ohms may not be enough, but you can try it.
.
I have not had any problems getting all 5 of my salvaged ATX P/S to come up just by grounding the PS_ON wire. Is it possible you have one that has it's internal fuse blown or just crapped out? Is the internal fan running?
Due_unto:
I have not had any problems getting all 5 of my salvaged ATX P/S to come up just by grounding the PS_ON wire. Is it possible you have one that has it's internal fuse blown or just crapped out? Is the internal fan running?
There is a fuse inside which is normal and functioning. Yes, the internal fan is running.
how can I get desired power from ATX power supply? sounds like it doesn't provide needed power to run a servo motor and acts stingy!
Your power supply might be dead, deffective, or have a voltage regulation crossloading design issue which requires enough 12v load to keep the 5v supply working properly within spec. If your old power supply is a cheapo, odds are that it was designed primarily for a 12v load (CPU, hard drive, fans, motherboard, and video card) and may not have been designed to provide a decent 5v power output without a 12v load. If the 5v rail is regulated down from the 12v rail, it would cause a regulation problem where crossloading only the 5v rail would make the 12v rail clamp down due to overvoltage and cut off the 5v output.
Try plugging a dozen fans into the 12v rail and try again. Or, try a different power supply.
ATX supplies regulate both the 5V and the 12V but in a fairly crude way, in that there is feedback from both the 5V and the 12V which gets summed together in a resistor network and then fed back to the switcher chip (usually a TL494) which regulates the PWM cycle.
Theres also overvoltage protection on both the 5V and the 12V so if either goes hi the whole thing shuts down.
If you load the 5V rail, but not the 12V rail, the 12V will be high and will trigger the overvoltage shutdown.
Try putting a load on the 12V rail.
mauried:
ATX supplies regulate both the 5V and the 12V but in a fairly crude way, in that there is feedback from both the 5V and the 12V which gets summed together in a resistor network and then fed back to the switcher chip (usually a TL494) which regulates the PWM cycle.
Theres also overvoltage protection on both the 5V and the 12V so if either goes hi the whole thing shuts down.
If you load the 5V rail, but not the 12V rail, the 12V will be high and will trigger the overvoltage shutdown.
Try putting a load on the 12V rail.
I found a 7w - 3.9 kohm sandbar resistor in an old disposed circuit and put it on 12v rail. Still the same!
The servo motor doesn't rotate but makes some noise, sounds like it can't get enough power but it warms up!!!
With the pre-loads on the 5V as well as on the 12V... what do they then measure ( V_out = _____V )?
Many years ago, I investigated using a computer power supply for a high current project. I found that many of these power supplies only regulate one output. Either the 12V output or the 5V output. And because of this, the more current you pull out of the regulated side, The higher the voltage on the unregulated side. If you use the unregulated side, You have to pull a significant amount of current out of the regulated side as well. This made it very inefficient, and unsuitable for use as a power supply. Instead of trying to make do, and probably spending a lot of money and time trying to salvage a ATX power supply, You can buy one suitable for your application. Such as,
if you're in the US, for $26 you can have a power supply that is both efficient and Safe. if not, you can probably find one for a reasonable price in your country.
I know people have used ATX power supplies successfully, but you need to find one that is regulated on both the 5V and 12V outputs, separately. One that does not require pulling current out of the unused side in order to get voltage out of the needed side.
promacjoe2:
Many years ago, I investigated using a computer power supply for a high current project. I found that many of these power supplies only regulate one output. Either the 12V output or the 5V output. And because of this, the more current you pull out of the regulated side, The higher the voltage on the unregulated side. If you use the unregulated side, You have to pull a significant amount of current out of the regulated side as well. This made it very inefficient, and unsuitable for use as a power supply. Instead of trying to make do, and probably spending a lot of money and time trying to salvage a ATX power supply, You can buy one suitable for your application. Such as,https://www.sparkfun.com/products/14098
if you're in the US, for $26 you can have a power supply that is both efficient and Safe. if not, you can probably find one for a reasonable price in your country.
I know people have used ATX power supplies successfully, but you need to find one that is regulated on both the 5V and 12V outputs, separately. One that does not require pulling current out of the unused side in order to get voltage out of the needed side.
Thank you, it isn't worth it, I've already spent too much money and time on this, although I liked the challenge, I can buy one of those power supplies you recommended 5v-40A with 10$ in my country which I am going to buy probably.
I'd still like to know what you've measured at these presumed 5V and 12V outputs.
If you don't know, not having a voltmeter, just say so.
Sorry that it took too late to answer, busy celebrating Nowruz.
I do have a multimeter, just one servo by itself drew at most 0.6A and I couldn't measure output voltage because the moment I connected the multimeter the power supply would shut down.
3.9k ohm is not nearly enough of a load. That is barely over 1mA on the 5V line.
Did you perhaps have your meter leads connected to the amp measuring inputs?
polymorph:
3.9k ohm is not nearly enough of a load. That is barely over 1mA on the 5V line.Did you perhaps have your meter leads connected to the amp measuring inputs?
No sir, I didn't. Doesn't matter. It'll not be working well as easy as I was expecting no matter how much I spend time on it. But it never warmed up, despite the other sandbar. 3.9k ohm was on 12v rail BTW.
3.9k and 12V is V^2/R = P
12^2/3900 = 0.037W or 37mW. No reason to expect it would get warm.
There is no reason a properly working power supply should shut down when a meter is connected. Most DMMs are 10M ohm input resistance.