|Feb 21, 2009, 06:27 PM|
A simple high quality 12Volt 100Amp Power Supply- Part1
With the growth in big LiPo packs and more capable chargers such as the Hyperion Duo2 and Hyperion EOS0720iNET there is a difficulty in finding a decent capacity bench power supply, particularly if you want to run two or more chargers at once. Power supplies for 12 or 13.8V up to 10Amps are reasonably cheap and available, up to 20-25 Amps are a bit harder to find and cost more, and over 25A are fairly rare.
There are plenty of articles on converting an ordinary PC power supply to a 12Volt high Amp supply and this has the advantage of low cost and many of us have done it. But there are a couple of disadvantages:
1) It involves opening a (potentially lethal) supply and modifying it, using external load resistors or changing the voltage sensing circuit, which is simple enough but outside many modeller's comfort zone.
2) The performance is generally woeful at high current because of poor regulation unless an expensive high-end supply is used thus defeating the low cost part. The most common way to do the conversion is to put a load resistor on the 5Volt output since these supplies are designed to maintain the voltage on the 3.3 and 5V lines which feed the motherboard. The bigger the load, the better the regulation but that also produces significant heat.
My typical converted cheap 550Watt PC supply drops to 10.5 Volts at 20 Amps. See picture PCsupply.pdf
An alternative that has been popular in the US is to use a second hand IBM Server Power supply which is of much higher quality, requires no internal modification and can deliver 12V at 30Amps. They are model IBM 235 Server Power Supply, part number 49P2020. Unfortunately they are becoming rare even in the US and I have never seen them available in this country. The trick is of course to get them to turn on at all since they are "hot-pluggable" and contain a lot of smarts to ensure they do not supply current until various logic is correct. Everydayflyer and teegreen have both promoted these on the forums and I understand from them that the solution originated 2 or 3 years ago, when the surplus electronic supply company called All Electronics was selling two of these power supplies in their server case for around $30 and someone included a sketch on their web page on how to do the connection. The thread is “What power supply should I get?” at http://www.rcgroups.com/forums/showt...=799781&page=7
I have just built the same thing using a currently obtainable HP server supply and figured out how to make it run. It delivers 12.77V under no load, and 12.27V at 118 Amps. That's a lot of power. This is a very high quality, compact, beautifully made power supply and dead easy to modify. It involved NO modifications to the supply at all apart from soldering on the output wires to the terminals at the back so it’s safe. I think I could have a finished working unit in 15 minutes now I know how, and the only parts needed (in addition to the connecting wires) is a servo lead and any small switch such as a Rx battery switch. The skill level needed is the same as that to solder a Deans Ultra or bullet connector to a battery lead. The cost will depend obviously on where you can get the power supply. It’s a HP PROLIANT DL580/ML570G3 RPS REDUNDANT POWER SUPPLY. P/N: 348114R-B21. These Proliant servers are very common and are designed to run continuously with high reliability. The power supplies are designed to run in pairs for redundancy and be hot pluggable (you can yank one of them out while the computer is running and the other one keeps on going).
I have seen the supplies advertised as a retail item from reputable suppliers on the Internet for about $100. (All prices are in US$ for comparison). In Australia, HP get rid of all of their excess stock, returns etc through a mainstream on-line auction house called GreysOnline. Almost every day there is a new listing of laptops, servers, printers and the like. Often there will be power supplies, sometimes as part of a bundle with other gear. They start the bid at $25 or less and depending who wants one at the time they can sell at that price to a single bidder. I paid $54 for mine but I saw someone get three of them and 10 CD drives for $12! You may find HP do something similar where you are. The cheapest comparable item I have seen available ready to go is a 43 Amp supply which sells for $100 and anything that offers 100A is much more expensive.
In the next post I’ll show how I did it with some pictures. You could run 4 chargers from one of these each drawing 25Amps, to simultaneously charge multiple packs and it wouldn't break a sweat!
|Feb 21, 2009, 06:33 PM|
A simple high quality 12Volt 100Amp Power Supply- Part2
I’ll describe how I converted a HP PROLIANT DL580/ML570G3 RPS Redundant Power Supply into a 12V 100A charger power supply.
But first a disclaimer: This is just a description of what I successfully did. I am not a qualified electrician and I am not offering any advice, representations or suggestions. If you choose to build one of these it is entirely your responsibility. If you burn down your garage, electrocute the cat or create a Black Hole that swallows up the State of Kansas and the Sydney Opera House, it’s entirely your responsibility (and those of us down-under are going to be a bit annoyed about the Opera House).
I bought this supply as a bit of a gamble since it is virtually impossible to find any information on what the pinouts are (or even what the output current rating is until you get one). All I knew was the input power spec given in the Proliant fact sheet. I was hoping it would deliver at least 45A on the 12V line as it had about twice the power rating of a similar Compaq supply I had seen. Turns out it was perfect for the job as it is designed to supply mainly 12V and the 3.3V and -12V lines are quite small. The supply has a connector on the back that consists of 8 high current blades and a 28 pin “logic” connection. It turns out that the blade connectors carry the 12V and some of the small pins the other voltages. By an empirical discovery-driven heuristic method (that means lots of trial and error) I got it to run by connecting two of the control pins to a third ground pin. I’ll do another post later if anyone is interested on how I figured it out which might be helpful to anyone who is trying to make a different server supply work. By stroke of good luck the three pins were in line so it was simple to use a servo connector to bridge them by just connecting the three ends of the wires together. In the final version I added a switch so that I can turn it on and off without using the wall switch.
UPDATE 1 xandrios discovered a diagram of the pinouts of this supply - see post #495. Also a data sheet with the pin logic explained. See DS1300-3 Series.pdf.
Read Post#7 if you have an unknown supply you want to turn on.
UPDATE 2 These guys are noisy and the only practical way to quieten this particular supply is to slow down the fans by adding a resistor in the supply lead.
This post shows how.
See the diagram and picture - Connections
The servo signal lead (Orange or White normally) connects to the one shorter pin of the 28 pins.
I used an old servo case to house the switch at the front and cut two little lugs off the front panel to allow the handle to swing right down and make a stand. The Tx gives you an idea of the size. I built it with two separate output connectors for convenience in using two chargers. I use a Y adapter if I want very high current as the current passes internally through the PC tracks to the blade connectors and it is a good idea to spread the load over all the tracks rather than take all the current from just one or two blades.
See the picture of the overall unit.
The supply comes with its own power cord which has a female connector on the other end that connects to a standard PC type power cord. I chose to shorten the cord and put a standard wall plug on it.
|Feb 21, 2009, 06:40 PM|
A simple high quality 12Volt 100Amp Power Supply- Part3
If I was building one again I would do two things:
1) Check that the power supply worked before I soldered on the output wires by plugging it in and turning it on. The fan starts and the green LED blinks to show it’s in standby. Then with a voltmeter across any of the two adjacent blade connectors, push on a servo lead and twist the three bare ends together. The fan speeds up, 12V will appear and the LED will glow solid Green if it is OK.
2) Solder the wires on differently. I used heavy gauge silicone covered wires and soldered them to the end of the blades which is fine but it is hard to get a solid join and there is always the risk one will get knocked and come adrift. The better solution would be to use 25A wire which can be pushed in between the two blades of each connector, then soldered. It’s easy because the blades are gold plated and they grip the wire while you solder. Makes a stronger, neater job. See the pictures (which are of different similar supply – hence the five blades not four). I carved away the top ledge of the plastic connector frame with a Dremel to let the wires go down far enough.
How well does it work?
The supply is rated on the case at 106A @ 12V on a 200-240Volt supply and 74A on a 100-127V one. In addition it supplies 1A@ -12V and 7A@ 3.3V. The lower rating for US voltages is presumably a function of the maximum current allowed in the input stages.
The attached .pdf file shows the regulation as I increased the load to 118 Amps. That’s when I ran out of 12V loads! The total voltage drop from no-load to 118A is just 0.5Volt which is pretty impressive. It will also not start if the load is too high. For example two 12Volt cup heater elements, 10x 50Watt and 4x 160Watt light globes in parallel have too low a cold resistance and the supply simply refuses to start. I had to connect them separately one after another to do the full load testing.
I used a Camlight 400 in parallel with the resistive loads to supply the variable load and an Emeter2 to drive the Camlight and log the results. Both are brilliant gadgets! Thanks John, Mark, and Dave.
Pros and Cons
Pros: This is a compact, extremely high quality (the MOSFETS inside are the biggest I’ve ever seen), high current supply with excellent regulation. It requires NO modification to the supply or disassembly. I’m pretty sure you could run two in parallel if you wanted 200A as they are designed to do this in the server. I haven’t tried this but Everydayflyer got two of the IBM’s to work just fine. They automatically shared the load.
Cons: They aren’t as cheap as a PC supply or the older IBM ones (but they do deliver a lot more current). The fans make a noise, although it is not an irritating high pitched one. You can’t run two of these in series if you want 24V. Because the case is connected to ground through the power plug connecting the -12V of one to the +12v of a second will just produce a dead short. UPDATE: There is now another thread I started on Running two 12V supplies in series which covers this. It starts here:
|Feb 21, 2009, 07:00 PM|
Update on part numbers
The part number I gave 348114R-B21 is the one that brings up the power supply if you do a Google search.
The label on the case has a couple of different numbers.
Part Number 337867-501
Spare number 406421-001
Which return nothing on the net but might be useful to know.
|Feb 21, 2009, 08:15 PM|
I ave also been searching for higher output supplies for charging. I have found some that are fairly easy to modify and bought several so I could drive shipping down and resell them to other modelers.
Note that the larger one does 84A at 240V
I have also been frustrated at the lack of available documentation.
RE the wires soldered to the outputs - As you can see from my pics the IBM X235 power supply has the same output plug as on the HP here. I first tinned my leads then flatten them with pliers. Then I push them into the connector on the supply and solder them in. Works great for over a year. You should also have some sort of strain relief as I see you ave done to prevent the solder joint from seeing and physical load.
Yes I'd like to know how you figured out how to get the supply to turn on.
|Feb 21, 2009, 10:26 PM|
Hi feathermerchant. They look a good solution and great value as well. The problem for those of us not in the US is that the shipping on any of your local solutions is a killer because of the weight of these little beasts. That's really what prompted me to look for an alternative locally and I figured being HP they are going to be available pretty much worldwide.
I think I prefer the method of soldering the cable inside the two blades for a number of reasons. It's easy to fit a wire to every blade pair and join them at the connector which spreads the current evenly across the PC Board tracks, it's very convenient to add the strain relief with a couple of cable ties through the back plate, and it's really simple as the blades hold the wire while you solder.
There's no magic secret to figuring out how to turn them on.
1) If you don't have any documentation then you have to hope it's your lucky day since it is a Catch 22 situation. Until you get it to run you can't measure voltages and until you can measure voltages you can't logically work out the likely right answer to get it to run!
2) You have to hope that a simple pin combination will do it and they don't rely on any timing or order of pin pullup to do it. If they do you are out of luck. My guess was that these hot pluggable devices would be pretty conservatively designed and as simple and fail safe as possible. I was banking on simple high/low logic to tell them what to do.
3) You take one apart and have a look at the PC board and figure out what the various pins might do. (Not recommended unless you know what you are doing though as these things have a couple of very healthy capacitors in them that may or may not discharge when it is turned off. I wouldn't want to find out which the hard way!)
That's what helped me find that the 3.3V was on the logic plug not the high current blades which was the obvious first assumption. It also eliminated a number of the smaller pins which were joined together as the 3.3V line.
4) A multimeter identifies all the grounds and whether they connect to the case.
5) It's then just a matter of trying all the pin combinations in a logical exhaustive sequence. Only a fool and a Darwin Award candidate would go round shorting out the outputs of a 1300Watt device at randomn so I used a 1k resistor between pins on the assumption the worst I could get was 12mA of current.
One clue was that there was 1 pin shorter than all the others. I figured this last to make/first to break pin would be the one that said it was finally OK to start up and so I started with that grounded and tried all the others. Fortunately it worked - and no, I have no idea why. Maybe someone with genuine PS knowledge of these things can tell us.
Just for the record. If you don't apply some logic and observation then the only alternative is to try all the pins, 2 at a time, then 3 at a time and if that doesn't work 4, 5 .... at a time.
The number of combinations of 2 pins out of 28 is 378, for 3 pins it's 9,828 and if it needed 4 pins connected to start there are 245,700 different combinations you would have to try. And that's assuming that they were all connected to ground or V+ not spread around.
UPDATE: Two years on and we have a lot more info on these supplies. xandrios has posted a lot of detail on how these work in later posts, including the pin out and logic. Starts around post #495
Here's his summary from Post #570 on how to solve an unknown supply
Bringing that up, this is the method I used to find the correct pins to power up this supply. It also works for a majority of PS units out there.
With power off and testing each pin to ground.
1. Exclude any pins that are common to each other(including ground pins). Usually these are the 3v and 5v rail pins. They also show the same resistance.
2. Exclude any open pins(pins with no resistance that don't connect to anything).
3. Exclude any pins with a value below 1k ohms and above 10k ohms. From my experience, I've found that the pson and pskill resistance usually falls between a 1k and 10k range.
With power on.
4. Exclude any pins that show no voltage.(pson and pskill are held partially TTL high or just not grounded. So they show some voltage on them).
This will usually leave between 4-6 pins.
Use a .5k ohm resistor on each of the individual remaining pins and connect each to ground. The power supply will usually power up at this point.
Disconnect one resistor at a time from ground.
If the PS remains on after a you disconnect a resistor from ground, then the remaining pins contain the pson and pskill. So keep it disconnected from ground.
If the power supply turns off, then the disconnected pin is either the pson or pskill. So reconnect it to ground.
Repeat this process until you find the pson and pskill pins.
In some cases the PS will turn on with a fault.
If this happens then disconnect one resistor(pin) at a time from ground to find the one that is causing the fault. Then continue with the process above to find the pskill and pson pins.
Even with the 32 pins on the Poweredge 6800 PS, I was able to narrow it down to just 5 pins before I even tried to power it up. It took just under 20 minutes.
Posts #598 599 here:
Contain useful info on finding the pins which turn a supply on and control the regulation if you have an unknown supply or want to raise the output voltage a little.
|Feb 22, 2009, 08:01 AM|
Thank you jj. I just found one that hopefully will work. For high current 12v but not RC. This is a very well written thread. Thanks again.
|Feb 22, 2009, 03:23 PM|
Don't do this!
Charles, this solution will of course work but please don't encourage anyone to do it.
These are metal cased supplies with the -ve connected to the case. The whole idea of the earth (ground) connection is to protect you if anything goes wrong inside the supply. As far as I can tell this is how all these bigger supplies are made and it is the standard way of wiring anything that is not "double insulated". The solution you show floats the case and if anything should fail inside there is the possibility of it becoming live at mains potential. You could enclose the PS TOTALLY in an insulated case and effectively create a double insulated device but ordinary folks shouldn't be mucking around with well established mains wiring practice.
The statement "Obviously, precautions must be taken to ensure that the cases of the power supplies do not come in contact with one another" is true but misses the critical point. It's ensuring they don't come in contact with you that matters!
This is an absolute NO in my book.
UPDATE: There is now a whole thread I started on the issues to watch and how to do this. Consider the comment in the line above as outdated. It can be done if you know what you are doing. Plenty of people have.
|Feb 22, 2009, 04:11 PM|
great stuff, thanks for the heads up - being down under does suck sometimes with regards to getting hold of gear that is available elsewhere around the world. This could be an excellent alternative as a hot-wire PS, I'm running my charger through a PC PS but run out of amps for bigger cuts.
I'll be heading into my nearest supplier to see what I can find.
|Feb 22, 2009, 05:25 PM|
Steve, (and anyone else who tries this) I would be interested in how it works for you. Mine is a sample of one and it worked fine but I'm conscious that we don't know exactly what I have done here. Without any documentation or service info it is not possible to tell what these pins are designed to actually do.
I did test the current between all pairs of the 3 pins when joined, and know it was less than 1mA.
|Feb 26, 2009, 03:21 PM|
Joined Jan 2009
My converted ATX :)
When i bought my multiplex easy star, it came with a 12v battery charger, i had a wall type psu rated at 4amps (what the charger needed) but it is to small the charger stopped the charging process as soon as it started.
Works nice but i did not use any load resistor on the +5v line but it works !
top posts: -5v 0.5a | -12v 0.8a
bottom posts: Ground | +3.3v 14a | +5v 30a | +12v 12a
top led is green for ON, bottom is red for Standby
|Mar 31, 2009, 08:40 PM|
Joined May 2005
Quit while I'm ahead??
I just yanked the 3 power supplies from a Dell Power Edge 4300 server that we were scrapping at work. I pulled the wiring harnesses from the supply to the boards so I had the colored wires. I got it to come on by connecting only two gray wires and I had 12.08 volts on the yellow wires. I connected my Triton with a big Pb gel cell and 3 lipos on my Polycharge 4 and cranked them all up each at about 1000mah. The PS only dropped to 12.04 volts. Is there any reason to go any further? Am I done? I tried loading the 5 volt line with a resistor and the voltage actually dropped. So is there anything left to do or just use it and be happy? The PS says 16amps @ 12 volts which is plenty for me. But I thought I would offer to sell the other 2 to some guys in my local club. I already had 2 converted PC PS, one 8 amps, one 2 amps. Let me know, Thanks,
|Category||Thread||Thread Starter||Forum||Replies||Last Post|
|Alert||Need advice on high grade power supply for ISL-8||Byron Betterly||Batteries and Chargers||7||Dec 07, 2003 08:31 PM|
|12volt, 9amp power supply for sale||Paul Susbauer||Aircraft - Electric - Airplanes (FS/W)||1||Dec 03, 2003 12:44 PM|
|12volt wallwart power supply reading 14.6 ? (help)||power||Batteries and Chargers||4||Sep 09, 2003 04:21 PM|
|Where to Get A Quality Power Supply?||Joe Elston||Batteries and Chargers||4||Apr 29, 2003 02:09 PM|
|High end charger and power supply for sale! Very new.||ted616||Aircraft - Electric - Airplanes (FS/W)||6||Dec 02, 2002 11:31 PM|