# Modifying Computer Power Supplies

Status
Not open for further replies.

#### alex0432

##### New Member
Has anyone tinkered with the newer, larger computer power supplies to know,
• Can the DC common (ground) be separated from the AC case ground so that these power supplies can be connected in series ?
• Can the multiple +12vdc wires be tied together for simple parallel increased 12vdc capacity ?
• Can the +5vdc some how be used in series to boost the voltage to 12+5 ?
• What is the impedance of these 12vdc circuits ? Are they voltage regulated, and so very low within their current ratings ? Or does the voltage very widely according to the load ?

I need about 100 amps @ just over 24vdc and hope to do with with generic computer power supplies, the newer (active power factor corrected) type.

Thanks for your input and ideas on this.

##### Banned
You're not going to get 24 volts at 100 amps out of PC power supplies, you would have to run them both in series and parallel it's just not practical. Why don't you use a transformer from a 120 volt mains socket? What exactly is this for? 100amps is not exactly every day use. Generally speaking for that kind of power you'd use a higher voltage and lower current. The wire thickness you need for 100 amps is huge.

Last edited:

#### alex0432

##### New Member
o.k. ... so maybe 80 amps, that is only four 500watt units, series and parallel. Its unusual, but so what ? If I wasnt a nerd, I wouldnt be here.

The cost of power-factor-corrected dc supplies, even iron-core transformer supplies, which also have low-power factor, too, is 100 times the cost of solving this problem.

I need these to drive a large DC motor. Can have high power factor because its a remote location on small generator power.

#### alex0432

##### New Member
that should read ... "cant have low power factor because its a remote location on a small generator".

#### alex0432

##### New Member
Sceadwian, "the thickness for 100 amps is huge". Yes, it is, dude. This is a basic question about series and parallel circuits, specific to generic computer power supplies .

Please dont be presumptuous about my purpose. For high power circuits, we just work "on the other side of the decimal point". I work with hundred- and thousand-amp circuits every day.

#### kchriste

##### New Member
Forum Supporter
You biggest problem will be dealing with current sharing between paralleled 12V supplies. You don't want one supply to go into current limiting just because it's voltage setting is slightly higher than it's siblings. So current balancing resistors would be required.
Some redundant module supplies have a communications buss between modules so that they can intelligently share the load. Your run of the mill ATX supply won't have that.

#### tcmtech

##### Banned
For a motor drive application I would honestly look at a simple constant voltage welder as a good power supply. (MIG, wire feed type)
These also have a variable voltage output too so you will gain the added benefit of variable speed!

#### mneary

##### New Member
Is this Yet Another Electric Boat thread?

#### alex0432

That arc welder idea is interesting... do you know what RMS voltage those transformers are ? Using iron core transformers is an option. Ive estimated the cost doing it this way is about $200 and maybe 100lbs weight, but it is low power factor and so seriously limits the available power I can get from the generator ... probably less than half I could get from APFC computer supplies. The rectified transformer only supplies power on the peaks of the wave form.... maybe 40% of the total time.... very non-linear. Ebay regulated 24vdc supplies would be much lighter and cheaper... but they are low power factor , too. If the computer supplies are low-impedance type, voltage regulated, they would need a small series resistor, but where did "one ohm" come from ? I calculate a small fraction of an ohm is required.... but there is no way to figure it without answer to the original question: Are these power supplies regulated or does the voltage vary with the load ? When disconnecting the case from the DC ground, why would the case voltage go wierd ? The case is solidly grounded on the a/c side. #### solis365 ##### New Member the 1 ohm resistors are to balance the current between parallel-connected supplies. i.e. if one supply is at 11.9V and another is at 12.1V and they are connected in parallel. the resistance in the wire is not enough to drop those .2V without lots and lots of current. it will be like the supplies are "shorted" it just won't be a short across 12V. because of the "short", the supplies will try to output a ton of current to make up for it (keep the voltage constant at 12). they can't handle that much current, so they go into current-limiting mode. the 1ohm resistors between parallel supplies will give some resistance to drop those little voltage disparities without requiring a huge amount of current. and about the chassis ground: I believe that, if you disconnect it then, while the chassis is grounded to the AC mains, it will be floating with respect to the DC., and for supplies in series you don't want to touch the floating chassis together, since each chassis is floating independently and you are referencing the DC from each supply to the same reference. #### alex0432 ##### New Member perhaps blocking diodes is a better answer for the paralleling issue. then i would have to add in another unit at 3.3v or 5.0v to get the voltage back up. Solis365, it sounds like you are saying the supplies are voltage regulated, is that right ? #### solis365 ##### New Member perhaps blocking diodes is a better answer for the paralleling issue. then i would have to add in another unit at 3.3v or 5.0v to get the voltage back up. Solis365, it sounds like you are saying the supplies are voltage regulated, is that right ? yes I believe they are regulated; they are required to be precise for the operation of the computer. i.e. its still 5V whether i hook up a 1k or a 100k resistor, the current changes. how would blocking diodes solve the problem of mismatched voltages in your parallel configuration? if the voltage differential between the supplies is less than the diode voltage, no current flows. if its more, even MORE voltage than usual flows. you need power resistors, if this scheme even works at all. #### alex0432 ##### New Member With blocking diodes on each unit, no power can flow backward into the supply, but only forward into the load. The resistor then goes after the diode and feeds into the lower voltage load. Lets say the power supply is adjusted up to 12.6v, after the diode its down to 12.0v and the giant, low-impedance load is rock solid at 11.0v. So, if the power supply is rated for 20 amps, the required resistor is e/i = .05 ohms which is easily made from a dozen feed of #14 copper wire. If the next power supply was at voltage less than 12.0v, then because of its blocking diode, no current would flow. But if it was adjusted up to 12.6v or more, it also would provide the i=delta e/r current into the load. #### AlainB ##### Member Hi, Here is a reply that I posted last week about computer power supplies. http://www.electro-tech-online.com/threads/exploding-l298-stepper-controlers.91847/#post729232 I don't know what would be the effect of paralleling them. About your original question, all the 12 volts wires (yellow) are tied together inside the cpu. Same with the red and the black ones. Putting them in parallel will not change anything, unless there is something new (to me) about newer, larger power supplies. Alain Last edited: #### tcmtech ##### Banned Most Helpful Member What is the motor powering? And what type of motor is it? A standard Brush type motor is not fussy about the power it gets. A fair amount of DC ripple will usually only result in a noticeable hum when loaded down. The actual efficiency loss from that hum is not enough to need to worry about typically. And as far as welders or any other transformer based power source, the power factor correction is simple. Just use power factor correction capacitors! With an amp meter and a little trial and error you could get a near perfect power factor! Most good welder power supplies have power factor correction devices built right in. Plus as I stated before, a MIG welder has a variable output. A 150 amp rated one would have a typical range of 15 - 30+ volts. MIG welders use a constant voltage type power system. They already have a large filter capacitor and a inductor built into the DC side. Plus by the intended application of a welder, the power supply design is already built to take high current overloads without any damage. Its a common farmer and construction equipment owner trick around here to use the wire feed welders as a big battery charger/jump start unit if you really need to get something started in a hurry! I have often used constant voltage welder/generators as charging and jump starter units. You just set the welders voltage at around 15 volts for a 12 volt system and turn it up to 30 volts for a 24 volt system. I do like the big switching power supplies too! But when you start getting up into that 24 volt 100 amp range dont be surprised to see a$2500 or higher price tag on it! One flaw I have seen with the big switching type power supplies is that they can have some problems with voltage reagualtion and general stability when powering brush type DC motors. They need added filtering to keep the electrical noise from the brushes down.

I pulled 24 volt 60 amp inverter type DC power supply unit off some large industrial equipment a couple of years ago. It was used and a little rough looking but I still went for $800 for it on eBay! ( was expecting$200, maybe!)

So even when buying used dont be surprized by the price the good ones can still bring! I was.

#### alex0432

##### New Member
Thats good input Alain. Ability to disconnect and float the dc ground is the key for series operation.

I hear there is some standard that requires limitations on the available power from each 12v circuit and so there are multiple 12v outputs, each power limited. Do these come from the same voltage bus and so can be tied directly together ?

#### alex0432

##### New Member
I had no idea what was in those welders, but that is very handy. Too bad they weigh a ton.

The power factor problem is on the 120v line side because of the wimpy generator providing the 120v power. Thats why Im interested in the APFC type computer supplies. THese will feed into a large battery bank which is running the motors, so no problem whatsoever with waveform on the dc side. I hope it will be cheaper, lighter, and provide considerably more power.

Thanks for your unusual perspective, tcmtech.

Status
Not open for further replies.