Server power supplies in series!

[ZOMGVTEK]

The current is limited by dropping the voltage. If its only a 5V? supply, it only has 5V of adjustment at the MOST. In real life as low as it can go is probably closer to 2-3V... If you are ONLY using the Meanwell, it might 'appear' to work fine, since it can drop the 5V down to 2V, which would allow for good control over current. Once you throw a few supplies in series with it, its few volt adjustment range makes up less of the total output voltage, making it not as capable of adjusting current on such a resistive load, like a lightbulb.

So long as you finely tune the system, it should work with only 5V of swing. It basically will ONLY be a limited battery charger. You get it working on your pack, and thats all it can really charge as is.

 

[auraslip]

You get it working on your pack, and thats all it can really charge as is.

That was the plan all along. If I want 24s some other voltage I'll need a different voltage meanwell..... but who cares.... they're only like $40 ebay.... that's the price of one lipo brick

 

[auraslip]

After reading what you said, I don't think this is going to work with a 5v unit. What a damn shame.

The reason is that the server PSUs put out a hard 13v and won't start sagging until god knows how many amps. The meanwell OTOH have 4.15v-3.7v * 20s = 9v worth of sag to work with to reduce current. Well beneath the under voltage protection.

arhg.

 

[auraslip]

I suppose I could un-mod the server power supplies to remove the sense line. That way they'll sag a bunch more.

 

[ZOMGVTEK]

That MIGHT help, but it would still be really finicky.

What you want, is a power supply with ample headroom. The different Meanwell series have different voltage adjustment windows, and you might be able to mod it to get an extra few volts out of it. Something like a 15V supply might be able to do it, but if you want to run 50A, you need a 15V 50A supply that can adjust current, and those are NOT cheap.

This is why a lot of people just get a big ol' lab power supply. They go from 0-X Volts, and 0-X Amps... But they are generally large, heavy, and expensive. Overkill for a dedicated charging solution, but for very high power chargers its one of the few ways to get it done. The better bet is to simply use supplies that can EACH limit current. If every supply in the series chain can drop its voltage to help limit the current, the end result should be a CC supply that can sag tons of volts, and tapers off at the preset voltage. Just what we want to charge LiPo.

A lot of people do appear to get away with having one Meanwell that can limit current, in series with 2 or 3 more. Generally something like a 24V unit in series with more 24V units. With two supplies, only one being CC should be fine for LiPo, but if you have 3 or 4 in series you might run into problems. It really would depend on your exact setup as to what voltage range is required, but shoot to have a bit more than you 'need' and it should work out. My Meanwell Bulk charger has 4 CC modded SP-500-24's in series. It cost $200 for 2KW, so it is substantially more costly compared to server supplies, but it does the job affordably.

 

[ohzee]

I am sure this has been asked/answered , but could not find a adequate answer.

I have 2 server power supplies I have modified both similar but a little bit different 12v 30amp vs 12v 50 amps.

can I run these in series or is that just a silly question ? thanks in advance.

One is a esp135 other esp115.

 

[ZOMGVTEK]

You can put any power supply in series, as long as the output is isolated. The ESP135 does not have an isolated output, and I would imagine the ESP115 does not as well. You only 'need' to isolate the output on one of them. You can do this by cutting the AC ground and isolating the two chassis, or opening the ESP135 and following instructions available online to isolate the DC ground. The ESP115 might have a well documented way to isolate the DC ground as well, or you might find that its already isolated.

 

[ohzee]

thanks again ZOMGVTEK - you gave me a good project for tonight.

 

[ZOMGVTEK]

Anyone looking to drop some cash on a probably perfectly functional Sorrenson?

http://www.ebay.com/itm/Sorensen-DCR-20-50B-0-20V-50A-DC-Power-Supply-Rackmount-DCR-B-Series-PARTS-/200666281890?pt=LH_DefaultDomain_0&hash=item2eb8a477a2#ht_4268wt_1180

Buy it before I do. This is the best price on probably the most desirable version I have found in several months. It's looking like it just needs a power switch.
Most of the 50A+ supplies are 10V (Not enough for a seamless transition using 12V supplies) or 40V (Requires 220VAC)... This is the perfect voltage, and 50A is enough to tap out the ESP135's...

 

[amberwolf]

Heck yeah--if I had money I'd buy it. (but I already have a 55V 55A on loan to me, so I don't really need it yet).

Anybody that needs a current limiter for fast charging a big pack--this is your chance to get a good one!

 

[auraslip]

I was about 3 seconds from hitting the buy it now button till I saw the shipping. Anyways it doesn't matter now. I have a s-350-48 on the way I got for $40 (2 more still available!) I should be able to use this to charge my 16s ping, freeing up my 2 s-150-27 units to be put in parallel for running the server stack. That should put me right where I want to be stock for my now 22s pack.... at 12a-14a. now I just need to figure out a mobile charging solution for up to 92v cause this 10ah pack ain't gonna cut it for daily use with out mobile charging.

http://www.ebay.com/itm/380377165777?ssPageName=STRK:MEWAX:IT&_trksid=p3984.m1438.l2649 here is another good deal hp 20v 20a in great condition for $150 shipped

I wonder about these psu's; how do they do current limiting? I mean, would a 5v unit work, or does it drop voltage like a meanwell (this is pwm right?)

 

[ZOMGVTEK]

You simply can NOT limit current while maintaing voltage. All power supplies limit current by dropping the voltage to the voltage at which the load requires that pre-set current.

That's why something like a big lab supply is great. They start from 0V and 0A.

Those HP supplies aren't bad, but 20A is nothing like 50A. You can never get more current, but you can always stack on more supplies in series for more voltage!
That Sorenson is basically $30 over the cost of shipping. It's HEAVY, so shipping aint cheap. None of this silly switch mode supply, this guy runs on transformers and linear regs.

It really depends on what you want to do. If you ONLY ever want to charge batteries, a Meanwell or similar will be cheapest.

 

[amberwolf]

All current-limiting supplies I am aware of must drop voltage during the limiting portion; it's ohm's law. But these do it smoothly (or at least, my Sorenson's do, and these should), rather than hiccuping, or PWMing the output line. I dn't know how the MW do it as I don't have one to scope out.

The linked one by HP has both adjustable voltage and current limits, which makes it like my Sorensons.


Basically you'd have the MW or whatever main voltage supply you want to use, plus the HP set up in series, no load.
Adjust the HP for nearly zero current output (knobs for current limit all the way counterclockwise, then just a touch clockwise).
Adjust the HP so the total series voltage is exactly what you want the top voltage of your charging to be at, at the end of charge.
Plug in your pack to the setup, and it will drop the output voltage to just about what the pack is at.
Adjust the current limit on the HP upward until the maximum current you would ever want to flow during charging is flowing; the voltage will rise a bit once this is done.
Now the pack will charge at a maximum of the current limit you set in the last step, up to a maximum of the voltage you'd set in the third step.

The final part of charging may take a while, since the voltage is exactly at the final pack voltage rather than above it, but it will never overcharge the pack, and the current flow near end of charge is very low.

You can shorten the last part of charge by increasing the voltage to something higher than final charge voltage, and using cell- or pack- level HVC to turn off the charger system or disconnect the pack from it. Keep in mind that some lab supplies (like my big Sorenson) don't have any way to prevent power from the pack going back into the supply, and may even trip their overvoltage protection (which will then shunt current from the battery at an unlimited rate, or rather, limited only by the resistance of the components in the OVP circuit, which may smoke). So disconnecting the pack at HVC is better than turning the charger off.

 

[ZOMGVTEK]

I've used a few old HP supplies, and they all throw a resistor on the output when you turn the power off. This was intended to discharge the caps rapidly, but it ends up smoking a whole lot with a big battery on it.

With switching supplies, most of them will be fine with a battery around the rated output voltage on them. Still you should try to avoid putting any battery on a supply that is not powered on. Depending on how much you care, it might be worth looking into how the supply reacts if it looses power, should a breaker trip, or the power otherwise go out while charging. Carefully monitor the current, paying attention to any possible reverse current. If there is any real significant reverse current, a diode would be a good idea.

Meanwell supplies without hiccup mode current limiting do tend to limit current within reasonable range well. They generally do NOT like to limit current much lower than 25% of the rated output, and going all the way down to 0 is not a good idea. You need to make sure the voltage range your pack operates within is inside the adjustment envelope for your charger. This means the voltage the pack is at when dead, must be within the range of adjustment of your CC power supply. So long as the CC power supply used can drop voltage significantly enough to match or go below the minimum battery voltage, it can limit current effectively. Obviously you would want the high end of the adjustment range to go up enough so that the pack will be charged as well. If you are charging a higher voltage pack and have a bunch of non CC supplies in series with a single CC supply, you probably will need a higher voltage main supply that you can limit the current on, so it has sufficient voltage range to be able to charge both a dead pack as well as top it off. It's not really complicated, its just logical. You can't shove 10V and 10A into any load you want. You throw some volts at it and the load determines how much current it wants. Your supply monitors this current, and can limit the output voltage to drop the output current, but it has no real control over the current. Its just limiting the voltage to limit the current. You can't force current into anything without sufficient voltage.

For testing things out, and assorted experiments, a true adjustable supply like that Sorenson is absolutely critical. For charging a known voltage pack, its overkill.

 

[auraslip]

I love you guys. I ask whether the big lab supplies limit current via pwm and I get a whole dissection of using them in action. This thread is officially note worthy. hopefully my mistakes, and ya'lls descriptions of what went wrong, will guide others in the right direction.

Those things MUST be huge; i was trying to figure out what it'd take to build a 2a-3a 92v linear power supply and the damn transformer alone weighed something like 10lbs. I can't imagine a 50a setup. Although the hardest part was just finding the right transformer....

You can shorten the last part of charge by increasing the voltage to something higher than final charge voltage, and using cell- or pack- level HVC to turn off the charger system or disconnect the pack from it. Keep in mind that some lab supplies (like my big Sorenson) don't have any way to prevent power from the pack going back into the supply, and may even trip their overvoltage protection (which will then shunt current from the battery at an unlimited rate, or rather, limited only by the resistance of the components in the OVP circuit, which may smoke). So disconnecting the pack at HVC is better than turning the charger off.

This is what LFP said he does. Well, he said he just watches it and pulls the plug, but a hvc relay would be trivial and cut the last 30 minutes or so of charge time down to just a few minutes. As it is now, I use a $3 ebay dmm to check to see if my 16 ping is charged. If current is below .2a, it's charged.

Should we be using diodes to protect the power supplies from reverse current? Or do most of the newer ones have them on the outputs. I would hate to have the power go out, as it's wont to do here, and the battery pack fries all my hard work.

 

[ZOMGVTEK]

I prefer having the charge trickle off towards HVC, its easier on the pack that way.

Current is meaningless without voltage. You can make THOUSANDS of amps off a small microwave transformer at very little voltage. However, 50A at 20V is 1000W, which sure is a whole lot of iron. Good transformers are not cheap. Those 1000W Sorenson's are around 70 lbs. There is just something about the laminations buzzing under load thats nice. Switchers are snazzy and efficient and whatnot, but I still prefer a big ol' transformer for a lab supply.

 

[amberwolf]

I love you guys. I ask whether the big lab supplies limit current via pwm and I get a whole dissection of using them in action.

Hey, if ES is good for anything, it's for overdoing things. Be it ebikes or answers to simple questions, we ARE overkill.

Those things MUST be huge; i was trying to figure out what it'd take to build a 2a-3a 92v linear power supply and the damn transformer alone weighed something like 10lbs. I can't imagine a 50a setup.

The little 40V 0.75A ones I have weigh about 15lbs. The 40V (60V? I forget) 1.5A version is close to 20lbs or more. THey're all linear types. The 55V 55A is over 50lbs easy, and I'm pretty sure it's an SMPS type!

Should we be using diodes to protect the power supplies from reverse current? Or do most of the newer ones have them on the outputs.

Diode can't hurt, as long as you account for it's voltage drop, and have ones that are rated for high enough current, or parallel enough smaller ones.

 

[voicecoils]

99.8V @ 47A with all of them in series. They can be paralleled without complaint as well.
I'm really happy with these supplies. The voltage is somewhat adjustable, but its basically just a 12.45V supply.
I isolated the DC ground from the chassis, all of them have earth ground connected and are in physical contact with each other. The tape just holds them together.

Those are nice units. I just bought 7 in total on eBay-Australia and am looking for a few more. Even though I had to pay $25 each plus too much for shipping, I still consider them a good deal.

I haven't seen many other server power supplies in the flesh but the DPS-600PB are seriously compact.

This is a nice guide for the PS too: https://sites.google.com/site/tjinguytech/my-projects/HP47A

My project today was taking 2 of them and setting them up to use in series for the iCharger 3010b. The ones I have both put out 12.66v without any load. I'm on the fence as to whether I should open them up and isolate DC ground. For the first 2 I've just physically separated them and cut AC ground pin off one of the pair.

If I can get at least 1 more, I'm looking at using 6 of them for an 18S lipo setup. That would give me 76v (4.22v per cell) and 3.45kW total charging power. It's handy having 240v in Australia and 10A or 15A sockets are common so decent charging power can be had from just one outlet.

Have you taken any measurements of the supply voltage at different current? Do you know how far it gets dragged down?

Cheers, the info in the thread has been really useful!

 

[ZOMGVTEK]

Those are nice units. I just bought 7 in total on eBay-Australia and am looking for a few more. Even though I had to pay $25 each plus too much for shipping, I still consider them a good deal.

I haven't seen many other server power supplies in the flesh but the DPS-600PB are seriously compact.

This is a nice guide for the PS too: https://sites.google.com/site/tjinguytech/my-projects/HP47A

My project today was taking 2 of them and setting them up to use in series for the iCharger 3010b. The ones I have both put out 12.66v without any load. I'm on the fence as to whether I should open them up and isolate DC ground. For the first 2 I've just physically separated them and cut AC ground pin off one of the pair.

If I can get at least 1 more, I'm looking at using 6 of them for an 18S lipo setup. That would give me 76v (4.22v per cell) and 3.45kW total charging power. It's handy having 240v in Australia and 10A or 15A sockets are common so decent charging power can be had from just one outlet.

Have you taken any measurements of the supply voltage at different current? Do you know how far it gets dragged down?

Cheers, the info in the thread has been really useful!

They are very well designed relatively high end switch mode supplies. Very compact, no random holes on the side that limits stacking, APFC, easy IEC in, and CHEAP.

I picked up two more for $30 shipped a few days ago, and reduced the fan speed way down, and increased the output voltage for 12V charger use. Mine right out of the box put out 12.49 - 12.50V, it might be due to the 240V that yours are slightly higher? All of mine were damn near identical. You can pull it up to about 13.6V before OVP kicks in, I wouldn't go higher than 13.5 to make sure it wont be randomly shutting down. When viewing the unit from the back, + on the bottom right, - on the bottom left, the voltage adjustment pin is the second from the bottom on the right side. Pull it up with a 1.1K+ resistor to +V and the output goes up. When you do this, the voltage DOES drop under load quite notably, but its still higher than it was and its easy enough to do. The fan pin is one down from the top, on the left side. Pull that one down with a 5K+ resistor to ground. Whats neat is when I tossed a 5K resistor on it, the fan speed was very low, but ramped under load. It definitely runs warm like this at light loads, but appears to be fine.

Without touching the output voltage, which I would imagine is simply some sort of sense adjust not designed to react to load, the output is fairly stable. It starts off at 12.49V and drops to 14.2x under a 50A load. Not a big deal.

Definitely isolate the DC ground. I can mod these units in <5 minutes now. I take all visible phillips screws out, remove the side cover, drill the rivets holding that locking tab in, and remove it. Pry the top open and wiggle the adhesive, set the top aside with the fan connected. Remove the screw in the visible corner with the pad going to DC ground. Then, clip the board away with cutters JUST outside of the screw. Theres a trace on the bottom of the board you might nick, but it looks like its fine since I nicked it on most of the ones I modded and they dont care. Then I just shove a screwdriver under the jumpers on the other side, break them in half, and screw it back together. Its a few minute process and you can series/parallel, with stock cables, and have the chassis's touch.

 

[ohzee]

Just ordered me a 2nd one of these so I can get myself 24 volts. I tried using 2 different type power supplies in series and
one of them just didnt seem happy.. Bad smells it almost seemed like one was working against the other. So I will just go with 2 of
the same type and hope for better results. Thanks for your writeup I had some confusion , but you spelled it out pretty nice.

if you do anymore you mind some close up pics of the changes to isolate the dc ground ? Regardless thanks for the good info.

 

[ZOMGVTEK]

I have two more I need to mod sometime, but theyre not at home right now.

I have one laying around that I can easily crack open and take pictures of, but its being used to balance a whole bunch of LiPo right now... Could take days.

It's a lot easier to do than it even sounds. It's really quite easy. I'm not exactly sure what that one trace does on the bottom RIGHT next to the ground pad, but cutting it didn't affect my charger one bit. You should still try to avoid cutting it by cutting away the board right next to the screw only, and NOT the entire ground pad area. The two little jumpers can be snipped in a few seconds quite obviously, and then thats it.

 

[auraslip]

I just tested it to see if it cares about the balance plugs. Not in fast charge mode. That's good.

 

[hillzofvalp]

Besides isolation and only hooking up earth ground on the positive ended psu in the string (positive right?), what should I be watching out for safety wise when I enclose all 4 of my hp esp 114s in an enclosure that is going to end up in a sheltered, open-air outdoor location subject to Midwest weather? I am going to use some scrap 8020 aluminum extrusion to make a cube to enclose them in. The hope is that I can easily expand it when I go to a higher voltage. Btw, this is going to be sitting in a parking garage under my apt. (chained to something), so that's why I'm trying to get it reasonably fool proof so some drunk idiot doesn't fool around and electrocute himself.. And in case you're wondering, I'm too lazy to to carry my 60-80 lb 29er up 4 flights of stairs everyday..

there will be high voltage signs.. I will make vent covers that don't allow fingers through.

Is there a good way to keep snow or rain out (that makes its way inside the garage) without obstructing air flow? Maybe something similar to the downward facing vents on most houses. also, what about operating temperatures? could I operate this thing below freezing without too many problems (the parking garage is generally dry where I'm putting it). What about safety concerning the Anderson's Charge connectors that may be somewhat moist or have dew on them? Btw, I'm not leaving this thing on all the time.. I will walk away from it for 30-60min until arduino posts twitter message and cell text that it's complete.

As a safety feature, I'd like to put neon indicators between all of them, but they're somewhat expensive. I have one though. I guess if AC was live between two enclosures, it would also be live between the others, right? ---eliminating the need for more than one neon indicator.

 

[ZOMGVTEK]

I'm not sure how long supplies would last outdoors, even if its mostly covered. Below freezing is probably OK, most parts of the supply wont care. Really hot would be more of a concern, and most of these supplies throw lots of air you wouldn't want to simply recirculate. You'd want some airflow through the enclosure, thats up to you as to how it gets done. Leaving it powered 24/7 isn't the best idea, but if its easier go for it and hope for the best. If the supplies are cheap enough see how it works out.

I wouldn't be worried about moist Power Poles, avoid giving the electricity a path across your chest and you should be fine. Hold plugs one handed, and avoid direct contact with >100V. Generally its not a remote issue, but its possible things can go wrong with reckless use. Don't worry about AC on the enclosures. Isolate them, and make it so its hard to touch them. Cover them in tape, separate them with plastic, something like that. No need for an indicator that theres voltage on the case, big deal. If the supply fails, only then should you be concerned. It wont make much of a difference as to which one, if any, has the AC ground. You can cut all of them or leave one. It's not going to be any safer. I'd probably run all 4 floating.

 

[hillzofvalp]

I was going to have a small battery for the arduino (to keep security streaming to twitter available if someone unplugs it), and I was going to put a 15A/30A relay on the hot AC line AND contactor so that when charging is complete the arduino disconnects battery followed by unplugging the 3kW machine from the wall (probably not going above 1500W). The security sense was going to bypass the contactor and just measure for present voltage. cool?