ATX PSU Modified for 25V

[Tenson]

I am modifying a Cosair TX 750 to make a 25V single rail high current supply for my desktop CNC machine spindle. I think I might have been silly to think I could increase the output voltage ans still draw the same current from it.

The spindle is a 3-phase outrunner motor from a radio controlled cars site and it has a 'ESC' whatever that means speed controller to adjust speed and convert single rail to 3 phase power.

The PSU has a CM6802 controller IC for the primary side and a PS233 monitoring for over-voltage / current etc.. on the secondary.

It seems the voltage is controlled by feedback from the secondary via a resistive divider network and a comparator in the CM6802. You can find out about all the parts in the PSU here

What I've done is remove the over-voltage protection from the PS233 by shorting the fault output pin to ground (all ok = low impedance, fault = high impedance). I've then changed one half of the resistor network so that the comparator circuit thinks the secondary voltage is lower than it acutally is and pulls up to a higher output. Oh, and I also uprated the caps for higher voltage on the secondary side so I hope nothing will go pop!

I now see 25V on the secondary un-loaded where it used to be 12V, but when I run the motor on it the voltage drops considerably. As I increase the speed of the CNC spindle, the voltage drops proportionally to about 18V which into the 0.46R load of the spindle is the 750watts the PSU can deliver.

Do you think this voltage drop is due to some sort of output protection limiting the total power or is it just the natural power output limit of the PSU? I mean can I draw more power from the PSU by increasing the output voltage or will the current drop proportionally?

[Tony]

aside from the existing 12volt rail, there is the -12 rail as well, if you can hack that rail to give more current, then you are in business....

[DigitalJunkie]

One 'trick' I've used is to duplicate the +12V rail from the transformer secondary onwards,except reverse the polarity of the diodes,and caps. That way you get a +/-12V (24V) supply. It's still only rated for the total power specified on the case/sticker for the 12V rail,but you have twice the voltage (thus half the current.)

[rikkitikkitavi]

Originally Posted by Tony

aside from the existing 12volt rail, there is the -12 rail as well, if you can hack that rail to give more current, then you are in business....

Negative. The -12V supply is designed to deliver a low current only, and is made by thin wire, or low power components. Something around 1A at maximum. Check the ATX 2.3 specs ,which this supply probably complies.

Yes I think that the output is limitied by the primary switch current monitor in CM6802. The PS223 is output overload protection only. You can not expect to pull the 12V rated current at 25V. Output current decreases proportionally (about atleast)

Also, by running it with maximum duty cycle (25V out) indefinately it might overheat the transformer core because of the higher flux density.

I have fiddled with similar and ended up modifying two ATX supplies (breaking secondary ground - earth connection) for series connection.
There is plenty of advise how to do this on the net.
You need to modify , otherwise they will short when series connected via earthed wall outlets.

[Rotter]

Hi,

why dont you just use 2 ATX supplies in series? This would give you 24V at the rated current, this way you would not need to modify the voltage regulation loop and the protection circuits would still operate.
One problem might be that the negative output is connected to Ground, so either don't connect the Ground/Earth to one PSU or isolate it internally.
Also you should make sure to use an atx 2.0 PSU where the main power goes to the 12V and not the 5V rail.
Greetings

[12E1]

Originally Posted by Tenson

The spindle is a 3-phase outrunner motor from a radio controlled cars site and it has a 'ESC' whatever that means speed controller to adjust speed and convert single rail to 3 phase power.

ESC = Electronic Speed Controller

This is normally digitally controlled using PWM (pulse width modulation) techniques to minimise power losses in the controller.

(ESC - same language used in model helicopters and the like, but probably smaller and lighter than in CNC machinery )

[rikkitikkitavi]

Originally Posted by Rotter

Hi,

why dont you just use 2 ATX supplies in series? This would give you 24V at the rated current, this way you would not need to modify the voltage regulation loop and the protection circuits would still operate.
One problem might be that the negative output is connected to Ground, so either don't connect the Ground/Earth to one PSU or isolate it internally.
Also you should make sure to use an atx 2.0 PSU where the main power goes to the 12V and not the 5V rail.
Greetings

You can break the secondary ground-earth connection as most of the time this is made by one of the screws attaching the circuit board to the chassis, by a copper area around the hole in the PCB.

It is better than breaking the earth connection which is a potential lethal action. NEVER EVER BREAK EARTH.

You will then get a floating secondary, and due to various capacitive couplings this might cause problems by potentail ridings all over the place , so perhaps it is wise to attach a resistor between earth and secondary ground.


You will also need to put diodes on the positive rail output of the lower supply , because one supply will ALWAYS power up before the other (within milliseconds) and it will drive current into the other and how this will affect the protection circuits ... (or the caps or whatever) is impossible to tell.

[12E1]

Originally Posted by rikkitikkitavi

You will also need to put diodes on the positive rail output of the lower supply , because one supply will ALWAYS power up before the other (within milliseconds) and it will drive current into the other and how this will affect the protection circuits ... (or the caps or whatever) is impossible to tell.

Actually, if connecting two supplies in series like this you should put a reverse biased diode across each supply, and they must be capable of passing the maximum expected current should either supply not be turned on.

As you said, either supply could power up before the other. Which one is above the other makes no difference - they both need protection.

[rikkitikkitavi]

yes of course , both supplies should have one diode as you describe it.

[tdaniel]

No offense meant to anyone, but really, a PC power supply is not designed to power CNC drives. Yeah, you might could hook four up to get the necessary current, or parallel 28 wallwarts, but Elton sez "It's like tryin' to find gold in a silver mine". Also, a bunch of audio fanatics may not be the best resource. Have you tried Planet CNC • Index page