Author Topic: Power Supply Conversion  (Read 7285 times)

0 Members and 1 Guest are viewing this topic.

Offline fsleeman

  • Regular Contributor
  • *
  • Posts: 54
Power Supply Conversion
« on: March 13, 2010, 08:35:46 am »
I am in the process of converting an unused PC power supply to a bench type supply as mentioned in many places online. Most of the time I will be using the LM317T circuit I am adding to run off the +12V line but will also be using the 3.3/5/12 lines directly. My only concern is short circuit protection for the three voltage lines. The variable output should shut down if shorted but I am not sure what will happen if, say the 5V line, is shorted, and not sure I want to find out. Does anybody have experience doing this or ideas how to make this a little safer? What dose a commercial bench power supply do when it is shorted?
 

Offline armandas

  • Frequent Contributor
  • **
  • Posts: 336
  • Country: jp
    • My projects
Re: Power Supply Conversion
« Reply #1 on: March 13, 2010, 09:02:17 am »
What dose a commercial bench power supply do when it is shorted?
Shuts off I suppose. It's called current limiting.
 

Offline desolatordan

  • Regular Contributor
  • *
  • Posts: 64
  • Country: 00
Re: Power Supply Conversion
« Reply #2 on: March 13, 2010, 10:06:19 am »
They usually use an opamp to compare the voltage across a sense resistor to a setpoint, then use the output of the opamp to mess with the feedback in order to lower the output voltage.

There is an example of an adjustable constant current/constant voltage circuit in the LM317 datasheet.
 

Offline mkissin

  • Regular Contributor
  • *
  • Posts: 93
Re: Power Supply Conversion
« Reply #3 on: March 13, 2010, 01:58:19 pm »
I urge people to be very careful when using an old computer power supply as a lab bench supply. It's perfect as long as everything is fine, but when something goes wrong, it tends to go horribly wrong.

Take the 300W supply that I have lying around in my workshop. It's rated to supply 20A on the 3.3V, 30A on the 5V and 15A on the 12V rail.

Now say you hook something up incorrectly and your +5V rail gets shorted to ground through one of your components. That PC power supply is just going to pump 30A through the fault, happy as Larry, until something on your board dies catastrophically (most components tend to die as short circuits, not open, so whatever dies will probably actually be blown apart) or one of your PCB tracks vaporises.

Im not saying don't do it, because it's obviously the cheapest and easiest way to get a bench supply, but be very careful when powering a new circuit with it!

You should probably add fuses inline with each of the high current voltage rails. They won't save semiconductors from death, but they will generally prevent more extreme failures.

Also note that often computer power supplies require a specific output rail to be loaded to a certain point before any of the other rails will come into regulation. It's probably either the +12V or +5V rail, and you can just add a fixed resistor inside the supply to draw a constant few hundred milliamps. That tends to be enough.
 

Offline fsleeman

  • Regular Contributor
  • *
  • Posts: 54
Re: Power Supply Conversion
« Reply #4 on: March 13, 2010, 03:10:44 pm »
Now say you hook something up incorrectly and your +5V rail gets shorted to ground through one of your components. That PC power supply is just going to pump 30A through the fault, happy as Larry, until something on your board dies catastrophically (most components tend to die as short circuits, not open, so whatever dies will probably actually be blown apart) or one of your PCB tracks vaporises.

Yep, that is what I am concerned about and fuses seems like the simplest solution. I have never added fuses to anything I have designed before, is there anything special I should know about?

Also note that often computer power supplies require a specific output rail to be loaded to a certain point before any of the other rails will come into regulation. It's probably either the +12V or +5V rail, and you can just add a fixed resistor inside the supply to draw a constant few hundred milliamps. That tends to be enough.

From what I have read, most people suggest a 10W 10 Ohm resistor on the 5V line.
 

Offline DJPhil

  • Frequent Contributor
  • **
  • Posts: 511
Re: Power Supply Conversion
« Reply #5 on: March 13, 2010, 11:10:04 pm »
I rambled on quite a bit here. Please forgive the detailed coverage of the basics if you already know this stuff. I include them for the use of any others who might stumble on this. If I make any mistakes along the way, someone bust my chops on it. This is safety stuff after all.

Yep, that is what I am concerned about and fuses seems like the simplest solution. I have never added fuses to anything I have designed before, is there anything special I should know about?

It is important to choose a fuse that will blow before any connection between the power supply board and your load becomes a fuse itself. Have a look at the power supply mainboard connections for the output wires. In most that I have seen there are several wires for each high current supply voltage (+5V, +12V, and likely +3.3V) and an enormous amount of ground wires. The reason for all this redundancy is that each wire is expected to share a portion of the maximum rated load. If any single wire is forced to conduct a large portion of the load alone it will very likely heat up, smoke off it's insulation, and turn into a light bulb filament, or just vaporize with a bang. If the wires are marked you should be able to calculate the safe load each can conduct alone with a little internet searching. Remember that the ground is sinking the current from all the positive supplies that are drawing load, and that accounts for the larger number of wires (or thicker wires/traces) that are present on the ground side.

This means that you cannot, for example, just drop one of the +5V wires to your output and then draw 10A through it without damaging that poor lone wire, and if this should happen you should open the supply back up and verify that nothing was damaged. It might seem to work fine, but you want to be sure that the wire insulation isn't compromised and the connections are still good. I've seen a wire melt off it's insulation in tiny spots everywhere it was under stress, which turned out to be the outside curve of all the bends it made. The supply worked fine, but would have been very dangerous to operate with all that naked wire inside. If the wire looks at all odd, replace it.

The solution is to determine the safe capacity of a single wire and use as many in parallel as you need when connecting them to your output terminal. This goes for the ground connection too, and even more so if you plan on using multiple voltages at the same time. You may notice that the supply can produce -5V and -12V with respect to ground at very small currents, usually less than one amp. These wires will likely be single leads from the board because the supply will current limit them before the wire itself is in danger.

The last bit to be sure of are your output terminals. Make sure you have good solder connections, and if you use screw terminals make sure the resistance through them is low. I've had screw terminals heat up quite a bit because of a loose connection. If your terminal (or any part of the circuit really) feels warm under a normal load then you've found a weak spot that could use shoring up.

That covers the inside of the supply, now for the outside. Every part of the circuit counts when a high current is drawn. Whatever you use to connect your supply to your project, hookup wire on a breadboard, wire connection to the breadboard buss, and the breadboard buss itself. For example, a cheap breadboard will probably start to have problems at about one amp, but I haven't seen ratings listed for many of them.

The short version to all this is that you should be as certain as you can that everything from the solder connection of the wires to the power supply board all the way to your project and back again is capable of handling the max current the supply can produce. Adding a fuse reduces the effective max current capacity, but you have to be aware of what else is in the circuit and ensure that the fuse is the weakest link. If the weakest link is not the fuse, you may be in store for heat, smoke, and fire.

I ran on about this ad nauseam because I think the frame of mind is more important than a simple answer. My apologies if it was a bit excessive. :)

Keep in mind as well that you can use more than one method for current protection. You can use an inline fuse in your power supply to protect your wiring, and a separate fuse on your breadboard input to protect the buss strips. You can try other forms of current protection too, like breakers and polyswitches (which are likely already present in the power supply elsewhere), just be aware of their limitations and cost.

From what I have read, most people suggest a 10W 10 Ohm resistor on the 5V line.

I was blessed with several older AT power supplies which switch on with a mains voltage loaded toggle (!). I solved this by bypassing the toggle internally and turning the supply on and off with an old power strip that has a toggle for each output, the old sort you used to see under the monitor of old computers with a switch for the computer, monitor, printer, etc. all separate. Some of the ATX power supplies do require a load as you suggest, but you may find yours doesn't. Some have reported success simply using a LED and matched resistor as an indicator light, which draws a great deal less current and produces less heat. I don't have a good answer for you, but you can apply the same methods to be sure that whatever solution you come up with doesn't contain a dangerously weak conductor.

All told, this is a great way to get a cheap bench supply and learn about power delivery. I'm still using the old AT supply that once powered my first IBM clone to power my new projects, and that's satisfying. Just be sure to avoid electrocution, fire, and that sort of thing.

Hope this helps some. :)
 

Offline fsleeman

  • Regular Contributor
  • *
  • Posts: 54
Re: Power Supply Conversion
« Reply #6 on: March 14, 2010, 01:33:32 am »
Thank you for that detailed information, I will keep it in mind as I complete this project.
 

Online Simon

  • Global Moderator
  • *****
  • Posts: 13124
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Power Supply Conversion
« Reply #7 on: March 14, 2010, 05:24:21 am »
surely it is not very difficult to make a current limiter or an electronic over current cut out ? you can use something like a 1-10 mohm resistor to sense the current and so loose very little, a mosfet such as the IRF9540N can carry up to 23 amps and has an on resistance of 0.117 ohms
https://www.simonselectronics.co.uk/shop
Varied stock of test instruments and components including EEVblog gear.
Also, if you want to get ripped off: https://www.ebay.co.uk/usr/simons_electronics?_trksid=p2047675.l2559
 

Offline DJPhil

  • Frequent Contributor
  • **
  • Posts: 511
Re: Power Supply Conversion
« Reply #8 on: March 14, 2010, 06:00:22 am »
surely it is not very difficult to make a current limiter or an electronic over current cut out ? you can use something like a 1-10 mohm resistor to sense the current and so loose very little, a mosfet such as the IRF9540N can carry up to 23 amps and has an on resistance of 0.117 ohms

Very true, and I should have mentioned that you could use circuitry to get the job done. I imagine it would be relatively simple, probably laughably so, but my personal experience with circuit design is very limited. Looks like I've got some homework to do. :)
 

Online Simon

  • Global Moderator
  • *****
  • Posts: 13124
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Power Supply Conversion
« Reply #9 on: March 14, 2010, 06:07:30 am »
well my experience in design is limited too but I'm addressing this particular project in another post as it's something I might build for other things like powering leds
https://www.simonselectronics.co.uk/shop
Varied stock of test instruments and components including EEVblog gear.
Also, if you want to get ripped off: https://www.ebay.co.uk/usr/simons_electronics?_trksid=p2047675.l2559
 

Offline mkissin

  • Regular Contributor
  • *
  • Posts: 93
Re: Power Supply Conversion
« Reply #10 on: March 14, 2010, 07:55:10 am »
You could add an electronic current limiter to the output of the power supply (on all three rails), but you'd need to be careful about what you want to do with it. Depending on what you decide, you've essentially made yourself an entire linear power supply anyway.

For example, what happens if you overcurrent the 5V rail? Do you leave the 3.3V and 12V rails running and just trip out the 5V rail? Do you trip them all at once?

Instead of tripping on overcurrent conditions, would you rather have a limit such that you can use the supply as a constant current source?

Finally, if you were to add the circuit, you might want to consider using the voltage drop of the internal FET resistance to detect the output current. Very accurate low value resistors can be expensive, and might not work very well inside the noisy environment of the SMPS. For example, a 1mR resistor would only develop a 0.01V drop with a 10A output. You'll possibly get that much voltage across it from switching noise in normal operation.

You'd need to determine the FET on resistance by hand for the exact FET you use, but if you're only making one supply, it's not very difficult. As long as the FET doesn't heat terribly much then the rds(on) will remain fairly constant.

That brings me to another point though. At 5A, that 0.117Ohm FET resistance is going to knock half a volt off your 5V output! That will bring it out of spec for all 5V ICs, which are generally +/- 0.5V max. The computer PSU has no way of compensating for this, as it would traditionally be done via remote sense leads (if the liit circuit is external to the main PSU).  Just something else to consider.
 

Offline VladKEasternTiger

  • Regular Contributor
  • *
  • Posts: 143
Re: Power Supply Conversion
« Reply #11 on: March 14, 2010, 09:41:17 am »
I am in the process of converting an unused PC power supply to a bench type supply as mentioned in many places online. Most of the time I will be using the LM317T circuit I am adding to run off the +12V line but will also be using the 3.3/5/12 lines directly. My only concern is short circuit protection for the three voltage lines. The variable output should shut down if shorted but I am not sure what will happen if, say the 5V line, is shorted, and not sure I want to find out. Does anybody have experience doing this or ideas how to make this a little safer? What dose a commercial bench power supply do when it is shorted?

See my latest thread, mabie you can help me out on this
 

Offline Smd

  • Newbie
  • Posts: 1
Re: Power Supply Conversion
« Reply #12 on: April 04, 2010, 06:03:34 am »
 

Offline DJPhil

  • Frequent Contributor
  • **
  • Posts: 511
Re: Power Supply Conversion
« Reply #13 on: April 04, 2010, 10:02:55 am »
Check this idea : http://www.chirio.com/switching_power_supply_atx.htm

Good find! I suspected that with enough experience it would be possible to modify the output.
I'm going to dig in to the details and see if I can figure this out.
Thanks for pointing this out. :)
 

Online Simon

  • Global Moderator
  • *****
  • Posts: 13124
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Power Supply Conversion
« Reply #14 on: April 04, 2010, 10:45:37 am »
Check this idea : http://www.chirio.com/switching_power_supply_atx.htm

hm looks interesting, anyone have trouble wading through the italian give me a shout: I'm fluent in the lingo
https://www.simonselectronics.co.uk/shop
Varied stock of test instruments and components including EEVblog gear.
Also, if you want to get ripped off: https://www.ebay.co.uk/usr/simons_electronics?_trksid=p2047675.l2559
 

Offline Ed.Kloonk

  • Frequent Contributor
  • **
  • Posts: 719
  • Country: au
Re: Power Supply Conversion
« Reply #15 on: April 11, 2010, 11:19:37 am »
Check this idea : http://www.chirio.com/switching_power_supply_atx.htm

hm looks interesting, anyone have trouble wading through the italian give me a shout: I'm fluent in the lingo

Thanks, but translate.google.com does a great job. It's actually a great article.  :)
 

Online Simon

  • Global Moderator
  • *****
  • Posts: 13124
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: Power Supply Conversion
« Reply #16 on: April 11, 2010, 05:17:38 pm »
yes it is, I'm not sure if the old PSU's i have use that chip though, are there any clones/equivalent chips that may have been used and can be modded in the same way ?
https://www.simonselectronics.co.uk/shop
Varied stock of test instruments and components including EEVblog gear.
Also, if you want to get ripped off: https://www.ebay.co.uk/usr/simons_electronics?_trksid=p2047675.l2559
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf