What to do with old PC power supplies

[auraslip]

Got an old computer in the garage? You bet you do!

Half an hour of work and

There are a bunch of guides around, but the trick is loading the 5v rail to pull the 12v rail up to 12v. To do this I used 2 1ohm 10w resistors in series. Before this mod it ran at 11.5v, and now it runs at 12.5v.
Now... you should know before you do this that the damn think sags to 11.5v while the imax is only putting out 80w. So it's 250w power rating is pretty damn optimistic.
Maybe two or three in parallel would do the trick. Or you could just use a PSU that didn't come from a POS Dell!

 

[amberwolf]

The PSU rating is for the WHOLE PSU, every voltage at once.

Many (most?) PSUs have a label on there that shows the *max* wattage for each rail. Sometimes it only shows the max current, rather than wattage, but that particular math even I can do right most of the time.

But yes, this is a great way to get various voltages, and there's lots of sites out there with different ways to set them up as lab supplies, etc. Some even to make them adustable to a degree, and adding current limiters, etc.

 

[auraslip]

er yeah... this one is 250w for the 12v rail..

 

[amberwolf]

Ah, well, then they must include the potential 5% tolerance of voltage in that 250W as sag, or else it is not getting the right load on 3.3V or 5V or something else to give the 12V full output.

Or it just happens to be defective in the first place. :lol:

 

[Alan B]

Some folks rewire them to put all the power out at 12V. Not an easy job.

If you have a lead acid battery around you can use it and a small charger to produce lots of 12V for awhile. Works even when the power is out, or with a solar panel.

 

[auraslip]

If you have a lead acid battery around you can use it and a small charger to produce lots of 12V for awhile. Works even when the power is out, or with a solar panel.

I could attach the 12v psu to the battery and the battery to the charger? I was thinking of that, but since it's putting out just over 12v unloaded would that be enough to charge the battery?

else it is not getting the right load on 3.3V or 5V or something else to give the 12V full output.

Do you think I should load the 3.3v line as well?

 

[auraslip]

This is the best guide i've found for doing this project http://sites.google.com/site/tjinguytech/my-projects/convert-pc-ps

BTW - it seems that the voltage drop I'm seeing is normal and I can't do anything about it.
I don't know if the low voltage is the problem, but it takes for ever to balance my ping pack with the imax. Like All night long, and it's still not balanced in the morning.

 

[geetarboy]

This is the best guide i've found for doing this project http://sites.google.com/site/tjinguytech/my-projects/convert-pc-ps

BTW - it seems that the voltage drop I'm seeing is normal and I can't do anything about it.
I don't know if the low voltage is the problem, but it takes for ever to balance my ping pack with the imax. Like All night long, and it's still not balanced in the morning.

I just checked out that link & don't understand soldering in the resisors to shore up the 12V bus? I just got a PC power supply saturday 500 watt output w/ 336 watts on the 12V rail and am running 4 ECO6's off it - unmodified and seems to work fine, don't understand? I checked it w/ multimeter and it puts out 12.38V w/ no load.

 

[Alan B]

This is the best guide i've found for doing this project http://sites.google.com/site/tjinguytech/my-projects/convert-pc-ps

BTW - it seems that the voltage drop I'm seeing is normal and I can't do anything about it.
I don't know if the low voltage is the problem, but it takes for ever to balance my ping pack with the imax. Like All night long, and it's still not balanced in the morning.

Balancing takes very little power. The supply is not likely your issue with that.

 

[neptronix]

This is the best guide i've found for doing this project http://sites.google.com/site/tjinguytech/my-projects/convert-pc-ps

BTW - it seems that the voltage drop I'm seeing is normal and I can't do anything about it.
I don't know if the low voltage is the problem, but it takes for ever to balance my ping pack with the imax. Like All night long, and it's still not balanced in the morning.

Does the imax show the voltages of the various cells? how much off are they?
Usually chargers like this are meant for doing very small batteries, so they have an incredibly slow discharge rate on each cell.
Whaddya got.. a 15-20ah pack? it would take forever if they were real bad.. IE if one cell was discharged badly from the BMS, it would keep trying to discharge all the other 7 cells to get the cell 1 up to par, lol...

Or it's just a crap charger!!

If it's taking a very long time and one cell is taking eons, your best bet is to give that 1 cell a charge on it's own as a single cell. You should be able to do that with the imax.

 

[auraslip]

For example - right now I'm charging a 12v lead battery at like 1 amp. Cell-log shows 12v, imax shows 11.9v and as the output current increases the difference on the Imax end gets larger.
It seemed to balance half my pack with the SLA charged up and in parallel. We'll see how it works when it's charged in the morning.

 

[skeetab5780]

a 12v power supply doesnt put out enough voltage to fully charge a SLA battery, it also does'nt float the voltage higher like a SLA charger. you need 13.8volts to get that SLA topped off. but it wont hurt pushing 12volts into it with a PSU when you're in a pinch. my friends were amazed when they saw me parallel charge a cell phone battery with another battery and a few paper clips

 

[mdd0127]

I wouldn't use Dell power supplies. I have a Dell that runs my CNC and it really liked to pop power supplies. I quit ordering Dell specific power supplies and picked up a used, $10, no name, probably twenty year old unit with 250 watts instead of 350. I Spliced the wires and hacked up the case to get it to fit and it's worked flawlessly since. I've had two Dell laptops too and right after the warranty is over, their little internal charge controller circuit/power supplies popped. I think it's a planned obsolescence thing. My experiences with HP, Compaq, and Toshiba have been very similar. Ibm, Sony and Acer power supplies are tanks though, and better built overall.

If you use good ones it works well. I used to have one that I used as a bench supply to test customer provided car audio equipment before I installed it.

 

[auraslip]

I just got another power supply from school. It's a tiny 150w unit. I'm going to put it's 3.3v line in series with the 12v from the other for 15.3v. It should be good for 9a @ 15v ~ 150w The Imax charger can take up to 18v, so every thing should be good. More voltage seems to help the imax balance quicker and more accurately.

Found the image here. Any body have an idea of what diodes I should use across these? Radioshack only had diodes rated at 3 amps. Do I really need diodes? I'm not concerned about blowing the PSUs up since they're so cheap!

If there was a way to disable the low voltage shut off from these power supplies (as in they shut down if voltage dips too low), it would be fairly easy to make a large, powerful cc/cv charger for bulk charging for next to nothing. You can tune the voltage to whatever you want by using the correct 3.3v, 5v, and 12v lines, OR you can fine tune it by doing some voltage adjustment mods you can find online.

 

[auraslip]

 

[x88x]

No need to mess with the sense lines. Just hook them up like so, with diodes rated past the current and voltage limits you'll be expecting:


Another note, with some PSUs, especially newer ones, you may also have to put a load on the 3.3V (orange) line as well as the 5V (red) line to bring the 12V up to full voltage.

Oh, and don't be surprised if old and/or cheap PSUs don't perform to their advertised specs, both because of efficiency (rather low in consumer units older than ~5-8 years) of manufacturers just outright lying (especially problematic with cheap units, like the ones OEMs use in the low-end systems).

EDIT:
Oh, and you have to short the green wire to ground. That's what tells it to actually turn on. I didn't see that mentioned earlier.

 

[auraslip]

Thanks,
I'll order some 20 amp 100v diodes, and with hold my curiosity before I put them in series. Is this diode good?

I did research about disabling the low voltage cut out on pc and server power supplies so I could build a bulk cc/cv charger, but couldn't find anything. Liveforphysics has recommended using server power supplies in series many times, so there must be a way to do it.

I'm thinking I could get 10-20 power supplies from my college. They have tons of old computers. Figure out a way to make a giant 1kw charger out of all of them

 

[torqueon]

Re edited due to info left out and spelling

I have 9 Dell 3r160 power supplies for the old dell optiplex . red and brown are parralell + black and white are parralell - Green yellow parralell, a sense wire. i installed a 10k pot, center to the sense wire, I can vary voltage from 3.7 to 12 volts and the volts hold well while aplying different loads. With a large shunt I charge all 24 headways in paralell for initial balancing, success!!
Wire harness is set up to charge in groups of 12 volts time's 3, so I seriesed 2 dells, set one to 7 volts and varied the other with a 10k pots. This time without shunt, some fiddling it worked out great 1.5 hrs per bank. success!!
Now for bulk charging 36 volts, 4 dellls in series, set 3 at 10.5 volts and vari the 4th one, no shunt Here is when things get weird, as the volts on batts increased i keeped turning up the volt via the 100k pots installed, while watching the volts of each dell after adjustment. increase one and one or the other would drop, all the while total of the 4 units produces the same volts. Much like circus ballon being choke off at varing points and swelling elsewhere.

Notes : Forgot to mention the A C ground must be removed from all.
10k pots used on single PSU and on 2 seriesed PSU
100k pots used on 4 seriesed setup . I bet thats why things got weird, will try again with 10k
So far these things show good promise!! Would love hear from the E S Gods

 

[auraslip]

Oops
put the two in serial, and now the old 12v psu is dead.
guess I shoulda waited for the diodes

 

[auraslip]

Oops
put the two in serial, and now the old 12v psu is dead.
guess I shoulda waited for the diodes

torquen - how did the PSUs handle the voltage drop? Did they shutdown when the PSUs were in cc mode?

 

[torqueon]

There self protected and shut down if amps get to high I start the charge with volt below batteries and increase the volts until shut down and reset to .5 volt below last shut down. So far these things work good, just got work out some detail with 4 in serioes

 

[auraslip]

Oops... I should of saw this coming. I shorted out one of the PSUs when I connect it in series. Because both are connected to AC ground it shorted when I plugged it in. This is a good guide to running two in series.

 

[auraslip]

This is why my PSU busted:

derp derp right?

The solution is to obviously cut the ac ground, and to also isolate the power supply chassis from each other.

By adding a pot to the 3.3v sense line it's possible to adjust the voltage.
Although on the power supply I got yesterday from school it has adjustment pots inside it! It goes all the way up to 12.8v. I put that in parallel with my fully charged SLA, and my ping balanced in no time.

In another news I found two compaq proliant servers on craigslist for $30. I think they have 375w power supplies in them. They have an option for redundant power supplies, so maybe if the stars align I'll get 4 of them
I understand power supply ratings are up to spec(rather than the shitty ratings on normal pc ps), and they handle cc/cv much better. I'm still looking for a way to REALLY adjust the voltage output. My goal is to get them up to 14.5v each because 4 * 14.5 = 58v which would be perfect for 16s lifepo4.

Of course 375/14.5 = 20 some amps, so I'll have to implement some sort of current regulation. Unless ya'll think I can charge a ping at 1c
Really though current regulation should be as easy as throwing the new 150w unit I got yesterday in series with them because it's rated at 4amp @ 12v. This would keep all of them at 4 amps.

 

[x88x]

rather than the shitty ratings on normal pc ps

Just as a note, this is not true for all consumer PC PSUs any more than it is true for all of any other type of AC/DC power supply. If you get a shitty PSU, yes, it's probably not going to put out it's advertised max current (at least not for very long). If you get a good quality one, it will. So if you're like me and happen to have a bunch of high quality consumer units sitting around from previous builds, there's no reason not use them. That's where I got the PSU I'm using to power my 106B+ now. That being said, if you're buying units specifically for something like this, for our purposes here server PSUs will offer a better cost/benefit since they tend to be designed for a certain chassis, so enter obsolescence much sooner than consumer units, which have been standardized (in terms of hookups and physical size and shape) for the last 10 or so years with very few changes. So units for a previous generation of servers, that someone got a backstock of, can often be had for much less than a similar unit for a current generation.

 

[torqueon]

Mock up of dell supplies in series