Server power supplies in series!

[ZOMGVTEK]

ZOMGVTEK - very nice! Can you elaborate on how you isolated the DC negative from the chassis? I picked up a few Dell DPS-500CB's (12V 41A) and I'm getting ready to assemble them in series. I like the idea of keeping the cases grounded and isolating the DC negative.

These supplies happen to be fairly well documented, but I just opened one up and its fairly obvious. The two screws closest to the DC side had pads that contact the case, and the AC ground comes in and goes right to the case. Clearly this would be an issue if you threw the supplies in series with a common ground. It happened to be fairly easy to just disconnect the DC ground, but removing the AC ground internally works fine. You would then need to physically isolate the cases if you went that route.

It just depends on your supply. Take a meter to the case and a the DC ground, start removing screws and lifting sections of the board until it goes open. You can just shove tape under the board and not put the screw in if you want. It all just depends on how the thing is designed. It might even have a single wire you can just cut.

 

[auraslip]

Changed the name of the thread.

Wish I went the route of breaking the chassis ground. Probably could of, but it was a really tight fit and I didn't like the risk of a component touching the chassis and shorting out.

Where did you get a 60a sorenson for $300?!?!

 

[ZOMGVTEK]

By $300, I meant $300 total, including $100 for the server supplies.

I'm still working on finding the exact model I want, but assorted older units can be found for about $200-400 no problem.

http://www.ebay.com/itm/SORENSON-DCR-20-25B-POWER-SUPPLY-/290609821144?pt=LH_DefaultDomain_0&hash=item43a9b1f5d8#ht_629wt_1180

This is a good price, and he might be willing to take a bit less, since its been on there quite a while.
It's ONLY 25A, but that might do for some people. I need a 30A breaker at 240V to get all of these supplies fully loaded anyways!
There was a lower voltage 60A one that sold for <$200 shipped a while ago that I missed.

Just need to be on the lookout, but wait for me to get mine first.
I don't really need it, so i'm waiting for a great deal.

 

[auraslip]

Yeah I was checking out those units. The new prices are insane. All I could find in that power range for around $100 was old analog units. Not sure if that's much better than a modded meanwell clone for $30 since I don't really think I'll need a voltage I can't get with the meanwell I have.

A digital one with a built in ahmeter would be rad though.

 

[ZOMGVTEK]

The new units are too damn expensive. I have found lower power <10A units for cheap, but they're not that useful to me. I have lower current supplies, I want something like 50-100A... Voltage isn't the big deal, that can be added on with some server boosters in series, but a common cheaper unit that does the job is 10V 50A, which would make gaps between the 12.5V server supplies.

While a meanwell would 'work' you wouldn't be able to adjust the voltage down to 0, and the current window is limited as well. They're cheap, but just not made to make a true adjustable 0-100+V high power supply. And even the units that are, tend to take ONLY 220V when you get into the several KW range. At least with this setup, I can run on 120 or 240 for as much power as the breaker can hold. Decent gauges, and very fine current control is a must. This setup wont monitor the output voltage all by itself, but it will tell you the exact current, and let you regulate it very accurately. Thats good enough for me, and these old supplies are reasonably priced.

 

[auraslip]

Why do you feel accurate current limiting is important? I mean, I'm able to get it down to the .1a range with 5k 10 turn pot. Or are you talking about the range? I didn't know there was a range (besides from 0 to it's upper limit)

For a volt meter, you could get a panel meter off of ebay. $10 up to 200v dc.... run it off a $2 5v cell phone PSU.

 

[ZOMGVTEK]

It depends on what you wanted to use it for. I would like very accurate current regulation, down to a mA. All of the modded meanwell's I have used, do not like to run very low current limits. I'm not sure a lot of people have tried, since it really wont apply to charging LiPo, but can you get a 50A Meanwell down to .1A without it buzzing like its going to burst into flames? Also, how would you be able to adjust the voltage down to 0? Can you extend the range for that easily?

Since the solution would require connection the appropriate number of server supplies, theres no real need for a 'guesstimate' voltage panel meter. I'd just run a quality DMM on the output for voltage, a panel meter wouldn't be able to easily read the total output voltage anyways.

 

[auraslip]

Don't know about the voltage range... especially on the 12v ones. Probably 20%?

I thought you were using these as battery chargers. Didn't realize you wanted a high voltage low current supply. I'm all like, "Pshh, anywhere between 10a and 15a is good for me!"

 

[ZOMGVTEK]

Ah, that makes sense. I want a high power variable supply for any purpose. I have <50V models up to a few amps, but I want something a bit more capable. I'm using 4 series Meanwell SP-500-24 for bulk charging, they will do 20A at 100+V, with ample voltage and current control to charge 24S at 5-22A. It works great and I'm running it at 15A to play nice with 15A breakers. It was under $200, which is reasonable.

This new Server supply setup is just a way to make a very high power regulated supply on the cheap. If I need more voltage, I can just buy 8 more of these HP supplies and make another brick! Having a nice lab supply to regulate this just makes sense.

 

[auraslip]

btw,
how dangerous are, say, 82v of power supplies. I mean, we're looking at like 30a here....

 

[dennyt]

ZOMGVTEK - very nice! Can you elaborate on how you isolated the DC negative from the chassis? I picked up a few Dell DPS-500CB's (12V 41A) and I'm getting ready to assemble them in series. I like the idea of keeping the cases grounded and isolating the DC negative.

These supplies happen to be fairly well documented, but I just opened one up and its fairly obvious. The two screws closest to the DC side had pads that contact the case, and the AC ground comes in and goes right to the case. Clearly this would be an issue if you threw the supplies in series with a common ground. It happened to be fairly easy to just disconnect the DC ground, but removing the AC ground internally works fine. You would then need to physically isolate the cases if you went that route.

It just depends on your supply. Take a meter to the case and a the DC ground, start removing screws and lifting sections of the board until it goes open. You can just shove tape under the board and not put the screw in if you want. It all just depends on how the thing is designed. It might even have a single wire you can just cut.

I opened these DPS-500CB's last night and successfully clipped the jumpers on the boards that bridge -12V to chassis ground. Really quick work once you know what to do. I'll write it up once I finish everything, I'm using these as 24V power supplies for my 1420i's. Thanks for your help!

 

[ZOMGVTEK]

btw,
how dangerous are, say, 82v of power supplies. I mean, we're looking at like 30a here....

It's really not that big of a deal in the right hands. I was abusing my supply that runs about 99V @ 47A, and it sure is capable of putting out quite of bit of juice, but nothing terribly scary. It's not like dealing with 200A mains or something. It has a ton of output capacitance, and sure burns up small wires quick on a dead short. However, you can touch the wires without much worry. Just don't go and puncture your skin while grabbing + in one hand, and - in the other. It's starting to get to be enough voltage to consider worrying about, since it can kill you in the right circumstances. You should try to avoid contacting exposed terminals, and if you do, make sure there is no path across your chest and you should be OK.

 

[amberwolf]

Grabbing the terminals with sweaty hands is generally a bad idea, but if you are careful and treat it like anything else with voltage on it, with respect, then you'll be ok.

Someday maybe I'll relate a story about leaning across a 72V pack with my bare forearm....

 

[auraslip]

Honestly, I'm mostly worried about my cat getting herself fried more than anything. I guess fur makes a good insulator though, and even skin isn't too good unless it's sweaty right?

 

[auraslip]

Hey guys, what do you think about this variable power supply as a current limiter? http://www.ebay.com/itm/KEPCO-VINTAGE-KS8-15-REGULATED-8V-15A-DC-POWER-SUPPLY-/120611691614?pt=LH_DefaultDomain_0&hash=item1c15045c5e

 

[ZOMGVTEK]

That supply should work fine, but it of course would not allow for seamless adjustment. Ideally you would want a supply rated for a touch above the output voltage of the server supplies you are using. That way you can use the main supply up to 25V or whatever, then turn it down to 0V and plug in a 24V server supply and dial it up 25V more. So long as you don't need this adjustment, and 15A is enough, that should do just fine.

Skin is a very good insulator when dry. Typical resistances are in the MOhm. Wet skin can be much, much lower, and make 100V semi-dangerous. You cat should be fine. The major issue isn't really the voltage, its how you touch it. If you give it a path across your chest, and its sufficient voltage to pump 50mA+ into your body, it will not end well. However, you can run a ton of power from one finger to an adjacent finger and its not generally life threatening, just sucks.

 

[Jay64]

First off I found these puppies for $5 each SHIPPED. I now have eight of them. They are rated at 585w and ~60a on the 12v bus. That's a WHOLE lot of power for $40. Basically, I'll be using this thing to charge what ever EV I can build for the next few years.

Is this link still pointing to the same units? I don't see anything there for $5, so I'm trying to find out if the price went up, or these are different units.

 

[auraslip]

sucks. Price went up I guess.

 

[hillzofvalp]

if you live in the us.. search esp 120 on ebay. a guy sells them for $34 + 18 shipping.. they are each 2950W and 51.4V. I'm thinking of buying one

220 only

 

[auraslip]

Finally got a genuine meanwell that I can mod. An s-150-5 - I should be able to do 84v @ 30a!

I replaced the R33 with a 1k 10 turn, heat shrinked it and glued it to the front of the board.

I also unsoldered the OVP zener diode ZD1 as I had problems on my series group of 2x s-150-27s with the OVP triggering and trying to burn out the board and the SCR instead of shutting down (because it's lower current limit I guess)

But this made me wonder; the output caps are rated at 10v. Is there any risk of of them popping if something goes wrong and the unit is exposed to the voltages from the other power supplies?

 

[ZOMGVTEK]

In series it won't see any more voltage than what its putting out. So long as the thing never kicks in OVP or OCP types of protection, all should be well.

It's probably a good idea to make sure this supply is on before applying a load.

 

[auraslip]

Thanks man, that's what I thought but wanted someone more knowledge to reassure me. You've quite a help in that dept. !!

I'm gonna rewire my battery discharge to provide a 12v dummy load for the 5v meanwell and 1x of the server PSU units to see if the current limiting is good. I'm not sure if anyone has ever reduced the levels that much. I had to go from 560 ohms to 150 ohms, and I think I recall some difficulties with my s-150-27s powering up when the R33 adjustment pot was out of range.

 

[auraslip]

Ok... I ran into difficulties just like I thought I might.

Instead of limiting current, the meanwell just shuts down. With two power supplies and the meanwell putting out 30v into 4 bulbs in 2s2p it's reading about 9a. As soon as I turn the pot down, the meanwell shuts off. It won't even power up with the pot down below a certain point.

 

[auraslip]

IT LIVES


Charging 8s at 31.5v - I can adjust the voltage from 3a all the way up to the sky.

what the hell is the difference between bulbs and batteries?

 

[torqueon]

From your last post I gather that you were trying to reduce the amps.

Something to keep in mind is that a light bulb will lose resistance as it cools, thus increasing the load it draws.

Keep up the good work