Hacking an AT/ATX power supply.... / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Hacking an AT/ATX power supply....

1 2 3 next

Move Thread

LAN_403

stepher

Thu Jan 19 2012, 05:54PM

Registered Member #4357 Joined: Thu Jan 19 2012, 05:32PM
Location:
Posts: 8

Hi all,

I'm brand new to this forum. I came here because of a search I did some time back regarding modifying an AT/ATX P/S to output different (higher) voltages. My plan is to make the mods and use an ATX supply as the "front end" and then design/build a uP controlled DC regulator secondary. The thread I came across was this one:

I've never taken apart an ATX xformer (primary-to-secondary). Based on what I read in the thread, it sounded like current limits are primarily controlled by the secondary regulator circuitry and that the same wire size was used for the secondary windings (cost of material and labor considerations). I always assumed different size wire was used for each of the secondary voltage outputs because of current requirements (i.e. bigger wire for +5 and +3.3, smaller for +12, still smaller for -12 and -5).

After reading the thread, it seemed to make sense to me that a mfg'er would save money (maybe quite a bit) in production by using a single size wire that handles the max current (and power ratings) needed (i.e. +5 and +3.3 and 12 etc.) and just run the taps out as needed for the different voltages.

Anyone have any authoritative input on this? Otherwise, I may just have to rip apart a xformer from an older, unusable (and blown) ATX power supply.

Thanks in advance for any help.

Cheers...Steph

m4ge123

Thu Jan 19 2012, 06:39PM

Registered Member #4118 Joined: Mon Oct 03 2011, 04:50PM
Location: MD
Posts: 140

ATX psu transformers have about 40 turns of ~22gauge magnet wire as the primary, and 4+3+3+4 turns (3+3 for 5v winding and 7+7 for 12) of tri- or quad-filiar ~18gauge. The 3.3V rail is connected to the 5V winding through an inductor to drop the voltage a bit. All you have to do to change the voltage is remove the overvoltage and overcurrent protection (desolder the diode connected to pin 4 of the TL494) and hook up a pot between pin 1 of the TL494 and ground (or maybe the 5V reference).

stepher

Thu Jan 19 2012, 09:42PM

Registered Member #4357 Joined: Thu Jan 19 2012, 05:32PM
Location:
Posts: 8

m4ge123,

Thanks for the quick response.

So, if I understand you correctly, since all voltages use the same "diameter" wire, I could conceivably pull 25A or more (assuming the highest current output, say for the +5 or +3.3, is 25+ amps) off the secondary w/out a problem. And if I were to pull the center tap of the secondary and make the ground referenced from either the -5 or -12, I could have higher voltage (looking for +30-35VDC) with equally as high current as an output.

I recognize that power limitations are pretty much set down by how much total power either the primary or secondary side can generate without burning things up (or out :(

I understand about changing the feedback loop to change the output "cycle", and thus the voltage. How much can voltages be shifted before the chip triggers on some limit (voltage, duty cycle, etc.), i.e., how much could I change the +12VDC output with this mod?

wrg to the 18 gauge secondary wire...is that 18 gauge as a bundle (either tri or quad-filar wound) or is each "strand" 18 gauge with the final bundle being a lower gauge, maybe 10 or 8?

Again, appreciate the help.

Cheers...Steph

m4ge123 wrote ...

ATX psu transformers have about 40 turns of ~22gauge magnet wire as the primary, and 4+3+3+4 turns (3+3 for 5v winding and 7+7 for 12) of tri- or quad-filiar ~18gauge. The 3.3V rail is connected to the 5V winding through an inductor to drop the voltage a bit. All you have to do to change the voltage is remove the overvoltage and overcurrent protection (desolder the diode connected to pin 4 of the TL494) and hook up a pot between pin 1 of the TL494 and ground (or maybe the 5V reference).

Forty

Fri Jan 20 2012, 03:29AM

Registered Member #3888 Joined: Sun May 15 2011, 09:50PM
Location: Erie, PA
Posts: 649

The lower voltage, higher current windings use more strands of 18awg wire than the 12v windings do. So you won't be able to pull 25A or more out of the 12v windings like you can with the 3.3v or 5v.
Since it sounds like you want to use just one secondary winding for your output, and vary the voltage via duty cycle, you might want to consider just rewinding your own transformer.
Uzzors page might help you out to if you want to build the whole smps from scratch:

and he has a spreadsheet that will help you design your own transformer

kimbomba

Fri Jan 20 2012, 03:31AM

Registered Member #3854 Joined: Fri Apr 29 2011, 03:45AM
Location: Mexico
Posts: 95

m4ge123 wrote ...

All you have to do to change the voltage is remove the overvoltage and overcurrent protection (desolder the diode connected to pin 4 of the TL494) and hook up a pot between pin 1 of the TL494 and ground (or maybe the 5V reference).

I did what is explained in uzzors2k page with two different PSUs, in both cases i got a high pitched sound and vibration. Any idea why is this happening?

m4ge123

Fri Jan 20 2012, 04:11AM

Registered Member #4118 Joined: Mon Oct 03 2011, 04:50PM
Location: MD
Posts: 140

Kimbomba, try turning the pot while monitoring the voltage. If it squeals when you turn up the voltage, you're turning it up too high.
Steph, here's a diagram I drew of a typical secondary.

The psu will turn off at 13-14V on the 12V rail, but you can easily bypass this by breaking the trace going to pin 4 of the TL494. Then, if you turn the duty cycle all the way up, you'll get about 25 or 30V on the 12V rail. Then hook up some diodes like this

and you can get up to about 60V (+/- 30V). You'll have to replace the caps and get new diodes. It would be easiest to just desolder the whole secondary side (except the control circuitry of course) and run 3 wires from the transformer to a board with new caps and diodes. Make sure you include the toroidal multi-winding inductor to limit the peak current so your diodes don't explode. Add enough turns to keep the V/turn the same.

stepher

Sat Jan 21 2012, 03:10AM

Registered Member #4357 Joined: Thu Jan 19 2012, 05:32PM
Location:
Posts: 8

Forty,

Thanks for the input.

Your comment wrt wire size seems to conflict with m4ge123's above.

Anyway, based on diagrams I've seen, my understanding is (and let me know if I'm wrong on this) that if I "lifted" the secondary center tap ground of the xformer, and make the original -12V output the new secondary reference "ground", xformer output (ignoring mods required to the chip feedback loop) would now be: 0 (new ground reference), +7, +12, +17 and +24 (See crude xformer winding diagram below).

------ 24V (old +12V out)
(
)
------ 17V (old +5 out)
(
)
------ 12V (old CT)
(
)
------ 7V (old -5V out)
(
)
------ 0 Ref/Gnd (old -12V out)

And by adjusting the feedback loop, I could get 35VDC from the xformer ends (0-24V out). The followon question is: given you are right about the "smaller" wire size for the orig. +12VDC and -12VDC lines, if the origs provided upwards of 12-15A, would I be able to pull the 12-15A total out of the new supply lines (i.e. 5A from the 0-24V lines, 5V @5A supply from the new 0-7V connections, etc.)? Also, do you know if the same wire is used for the neg windings that is used for the positive (i.e +12 out vs. -12 out) so that would not end up being a current-limiting factor?

Again, thanks in advance for the help.

Cheers....Steph

Forty wrote ...

The lower voltage, higher current windings use more strands of 18awg wire than the 12v windings do. So you won't be able to pull 25A or more out of the 12v windings like you can with the 3.3v or 5v.
Since it sounds like you want to use just one secondary winding for your output, and vary the voltage via duty cycle, you might want to consider just rewinding your own transformer.
Uzzors page might help you out to if you want to build the whole smps from scratch:

and he has a spreadsheet that will help you design your own transformer

m4ge123

Sat Jan 21 2012, 03:33AM

Registered Member #4118 Joined: Mon Oct 03 2011, 04:50PM
Location: MD
Posts: 140

Your understanding is correct. The same wire is used for both sides of the secondary because they are both rectified and sent to the output. You can pull half the 12V rated current through the whole secondary. The 12V windings have the same gauge, just less strands in parallel.

Steve Conner

Sat Jan 21 2012, 09:22AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

This talk of "lifting the centre tap and making the -12V the new ground":

The diodes used to rectify the -12 are puny, like 1 amp sized. You'll need to replace them. When you've done that, you'll have a bridge rectifier fed by the full secondary winding, which gives the highest output voltage possible, 24V or more.

You may need to add some more turns to the output filter inductor for that to work smoothly, but I've seen it done.

For more info see uzzors website or the ATX modification guide at qrp4u.de. This shows how to make one into a 13.8V supply for ham radio gear, but the principle of operation is explained and a general schematic given that covers most makes and models.

stepher

Sun Jan 22 2012, 03:02AM

Registered Member #4357 Joined: Thu Jan 19 2012, 05:32PM
Location:
Posts: 8

OK, so far so good. Thanks.

The +/-5 windings would be the thickest (lizt-style, 18 gauge "strand", no. of strands depending on current requirements) because of the highest current requirements (upwards of 25-35A, depending on the supply). Since the +/-12 windings carry less current, then it reasons they can be a fewer number of 18G strands. Correct so far?

So, help me understand why I can draw only 1/2 the 12V rated current thru the secondary. Seems to me if the +/-5V lines can typically handle well over 20A, and the +/-12 can typically handle 10A+ current, why couldn't I draw the current the 12V lines can provide (which seems to be the upper limit to me, assuming I don't go above the original watt rating of the supply (200, 250, 300, etc.)? Is it a total power thing? Or is it something to do with the "negative" windings?

As an example....let's say I want to set up a 24V supply (using the xformer "end windings") of, say, a 350W supply, if I assume 80% efficiency (optimistic est.) then I have 280W available. Divided by 24, that means I should be able to draw ~11A from the total secondary. Granted, diodes and caps would need to be changed. So, if the orig. +12V secondary was spec'd at 15-18A (what I saw was typical of a 350W), then I should be able to draw the full 11A. Obviously, if the +12V was spec'd less, that might possibly be the limit due to wire size.

Did I simplify this too much and miss something?

Thanks and cheers...Steph

m4ge123 wrote ...

Your understanding is correct. The same wire is used for both sides of the secondary because they are both rectified and sent to the output. You can pull half the 12V rated current through the whole secondary. The 12V windings have the same gauge, just less strands in parallel.

1 2 3 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob