PC Power Supply Mod.
 

Alrighty!

So reading through Brian's write up along with several others coversions for charging duty I decided it was definately the way to go. So I order a solid Corsair 400w single rail power supply capable of supplying 30A on it's 12v rail.

Fast forward, PS is open torn down here in front of me. I measured the 12v output and looks like about 12.17'ish solid. So should be good there.

So looking at making a few wiring mods to support my Hyperion 720i for charging duty. I'm planning on going to 12AWG DC out so looking at the board to find the taps and I see what says 12v1+12V2... WTF :no:

That said, I've got 2 sets of 4x 18AWG wires tapped into 12v1 on the board and a set of 3x 16AWG 4yellow with black stripes that I think are on 12v1 though I won't be able to tell until I get to the house and pull the board from the case to see.


Thoughts? I'll assume this doesn't have 2x 12v rails, no? I "believe" the yellow with blacks (had corresponding blacks with white) were from the PCIE 6 pin video card cable.

Thoughts?

Yeah, it sounds like it has two 12v rails.

Some people may say to just tie them together to get the sum of both rail current ratings. However, I've worked on a PS with dual 12v rails and the internal heatsink got quite hot when they were tied together. Each PS is different so you may be perfectly fine doing this, but I would run a test first; Temporarily hook the 12v rails together, and then hook the paralleled 12v wires to the charger, and charge a pack.

Monitor the 12v voltage and current to see how much the voltage sags at whatever current the charger is pulling. If the PS gets hot (and will happen withing a minute or two if it will) then there is an issue and each set should be used independently. If the PS stays cool, then you are fine with paralleling both 12v rails.

But be warned; certain components in the PS are running at dangerous voltages. Sometimes, even a heatsink can be floating at high voltage, so BE CAREFUL!!!!

Brian,

Here's the power supply...

http://www.newegg.com/Product/Produc...82E16817139008

Clearly states "powerful single rail"....


Let me get you a picture for a visual.

Hmm, it does say that in the specs and the reviews. Taking the board out will be the best way to tell then. I don't know why they would use different appearing wires if they go to the same electrical point - maybe they ran out of solid yellow? :neutral:

I looked at the pics on Newegg and I didn't see a sense wire on the mobo connector (would be a smaller gauge yellow wire that is tied to the main yellow wire on the mobo connector), but the pic isn't the best, so I'm not sure. There could also be a pot you can tweak in there as well.

BTW: Even though the PS is off/unplugged, there may be some caps with high voltage on them - so again, be careful. Don't mean to be repetitive, but don;t want to read about you in the obits...

Here's the pic (sorry Droid pic).

http://i23.photobucket.com/albums/b367/fast929/PS.jpg

NP on the warning. I've been working on 480v 1250A service on our new equipment here.

As I tell people "hands in pockets" when they are near the panels..... :yes:


Like I said, I need to pull the board to check.


And you are right, the only sense wires I seem to have with this supply are on the +3.3v stuff.

Hmm, don't see any pots to adjust. If they are there, they are generally pretty close to the FETs/coils. I think I might see one between the primary transformer and coil (roughly in the middle), but if it is, it is probably not to tweak the voltage.

I'm not overly concerned about upping the voltage as the charger will just pull more current.

Being the voltage is 12.17v, provided there isn't a lot of line drop, it should be ok. Seeing as I'm looking to charge at 2c for my 4200mah setup your talking ~10A so with 30A main feed, I should have some headroom.

I'm more concerned about the board listing 12v1 and 12v2 nearest the power wires. In the bottom of the image, you can see the yellow/blacks 3pair and the 2pair under those (to the left). The 12v2 is listed right there. 12v1 is listed nearest the 2x 4pairs above (to the right).

Going to need to pull the board to see/tell anything huh?

Also, as a neat aside... Looks like they are using this PCB for the 250-5000w power supplies. Makes me wonder if the 12v2 is a listing for the more powerful board options???? Just a though.

I also don't see any POT's in here. Appears voltage is fixed.


<-----(heading to the house to pull board).... Stay tuned.

Actually, upping the voltage will make the charger pull less current. Chargers have built-in switching supplies that turn 12v input into whatever output needed to charge the battery. So, for a given power output, higher PS voltage = less PS current required.

If you pull the board out, just follow the traces. It should be pretty obvious where they go since they are rather large.

The v1/v2 marking could very well be for larger supplies. Hard to tell.

Alright, here's the bottom of the pcb....

http://i23.photobucket.com/albums/b367/fast929/PS-1.jpg

12v marked with red circles.

Image without circles.

http://i23.photobucket.com/albums/b3...9/DSC_6656.jpg

Close up of area...

http://i23.photobucket.com/albums/b3...929/PSZoom.jpg

Did some more research...

Appears this IS a single rail power supply or at least, they are pulling power from a single rail. The actually supply puts out better than 450watts true power and closer to almost 500W before it shuts down. Not bad for $50...

Review.
http://www.hardwaresecrets.com/article/750/1


Still doesn't explain the single rail design but still has 12v1 and 12v2 which they tested independantly. How do I combine this? Just jumper the pads and wire them together???

OK, those two 12v areas probably have large coils going to another area. Circle the area where those coils go if you could. I'm thinking both those 12v lines come from the same output, but use seperate coils to avoid using one large coil,

Does this help? I've gone over it a few times and it seems correct...

Look at you! lol

I just spent an hour marking up the same drawing and I stop back in here to see you got me covered. lol

Clearly What's nitro has this better covered than I did. Seems to be right.

Coil tie in's.

The large coil has 5-wire tie in's denoted in purple.

The smaller 2-coil has it's 4-wire tie in's in blue.

12v power is in yellow.

http://i23.photobucket.com/albums/b3...t929/PSTop.jpg

http://i23.photobucket.com/albums/b3...9/PSZoom-1.jpg

Yeah some of my drawing needs to be shifted left a bit... Couldn't really tell based on the overhead view earlier in the thread. Close enough though! :lol:

Thanks again for the help here guys!

Much appreciated!

No problem, man!

Not my best image. lol

So this begs the question then,,,,


Where/what should I tap into for the full rail current?

Me again...with another awesome MS-Paint diagram! :rofl:

http://i52.photobucket.com/albums/g4...d/PSZoom-1.jpg

NICE!!!!

As you can see, I MS painted the crap out of mine too. lol

What your showing seems completely feasible to me. Provided it's all being powered off the same rail, I don't think there would be any issue bridging that pad.

Otherwise, I'm ditching all the 3.3 and 5v stuff anyway.

Looks simple enough!


Nice to see the east coast guys up late. lol I'm in CT.


Thanks again!

Yup. It is a single rail, but it splits into two leads through the choke coils. I think because in a PC there is a 12V ATX connector (for the Mobo) and the other 12V lead goes to the 6 pin PCIe and SATA connectors. I guess the choke coils arrest EMF feedback between the two leads??? That would suit their name... Since you only need one 12V+ lead to the charger I don't think the choke coils are necessary.

Perfect! Great information!

I'd also guess based on the ATX std, they generally keep the 12v feeders to below 18A I think. Maybe it's 15A but something like that. I found it in the white paper spec sheet for ATX power supplies.

The 12V lines are connected directly to the full power output. The smaller traces coming off of the main pads are for the regulation. I would guess that if either pad goes over 15A, or maybe 20A for ATX and 10A for Acc., then it would trip the overcurrent sensor and shut down. There must be a main OC sensor on the other side of the big Xformer, but at least if you bypass the chokes you wont have to deal with the unnecessary secondary OC circuits. Or maybe they will see equal load and not trip anyways... IDK.

In the review I posted earlier, they stated the OC protection on the 400cx worked well. Matter of fact, they stated at the limit, the power supply was closer to a 500w unit.

Well, hopefully Brian can verify so I can get this thing sorted tonight :intello:

Hmm. I posted a response last night, but it didn't save for some reason. Anyhoo...

Do NOT bypass the coils. They are probably sampling coils for regulation, and you might get crappy performance under load. I would simply bridge the solder areas after the coils.

Sounds like we have a plan!

Let me work this up and I'll keep you guys posted tonight on how it goes. Hopefully we'll be charging this evening!


Thanks again!

Don't forget; once you get the outputs paralleled, monitor heatsink temps after a few minutes of heavy operation. Shouldn't be a problem since it's really a single rail, but any regulation differences between the two coils/regulation circuit may cause an issue. It should be quite obvious after only a couple minutes, but I think you should be fine.

Will do.

I should only be using roughly 8-10A current for charging anyway so there should be some headroom available from the power supply. Hopefully the regulation between outputs should be similar seeing as they based off a common rail.

We'll see.....

Alrighty....

Got the power supply up and running and seems to be outputting power as per it's specs.

Hooked up the 720i and setup to charger my 5s 4200 Hyperions. Trying to charge at 2c so 8.4A @ 12v. Problem I'm seeing is some pretty significant voltage sag. Charger is reporting ~10.6v input and I'm reading 11.1v with my Fluke loaded. Unloaded it's running about ~12.2'ish

Not so hot. Matter of fact, seems as if the charger stops charging for ~4-5sec every so often. No error report and doesn't terminate the charging cycle but assuming a low input voltage issue....

Ok, charger just dropped out due to a low voltage issue. Bummer

Reset and dropped to 1c charge rate and charger is now pulling 11.197v. Still way on the low side.

Not sure what my options are as of now??? Any suggestions?

Issue 1: Did you attempt to load the 3.3v and/or 5v line(s)? Sometimes that makes the 12v line more stable under load. The PS mod thread explains this, and what values to use.

Issue 2: If the charger's input voltage is correct, there could be voltage drop on the input power leads. My old EOS7i came with really long (like 6') cables. Shortening them helps a LOT.

I haven't tried to load the 3.3/5v lines as of yet. Could be on the list of things to try as I'm a little surprised at the voltage drop. According to the reviews, the PS handled high load (max) during overload protection testing with very minimal main line drop. This even at nearly 500w.

Good info on the input cable length on the charger. I did find it interesting that I'm losing ~.6/.7v on the input side just between the PS cable 12v output and what the charger is reporting for line voltage. Seems significant.

Cable trimming to commence shortly and we'll see where we are.

Some supplies have totally independent 3.3v, 5v, and 12v regulation, so no loading is needed (and doesn't help at all). This is how the Antec TruePower units work. But, for most other supplies, one rail is monitored and regulated and affects the other two rails. After all, these units are meant for computers, so all the rails are used.

Yeah, if the wires are long, they can introduce a voltage drop. I created a thread about this a while back. I was losing a total of almost 0.5v just on the wiring, and that was at 9A (from the PS). This equates to a resistance of 0.056 ohms, so at a 15A draw from the PS, you can lose over 0.8v, which can make a difference.

Update:


So I shortened the length of the input leads on the 720i and it certainly helped. Originally I was losing roughly .6-.7v when measuring the bullets off the PS and then reading what the charger is/was seeing for input voltage. With the shortened leads on both PS and charger and addition of 6.5mm castle bullets, I'm seeing roughly .1-.2v loss now. Better for sure.

Still haven't tried loading the 3.3 or 5v lines. Next on the list. Yesterday I got to run the setup and get some batts charged at 1c and I was seeing about 11.4v on the input side(this at only 4.2A charge rate) No load on the ps/charger is about 12.1v.

I know there is more available from this PS so need to keep digging. At a minimum per ATX specs, I should be able to pull ~15A from one of those single pads. I've tied the 2x together via 12AWG so was hoping to have the full 30A available but clearly something isn't totally right here.

I just can't believe I'd see nearly a ~1V main line loss off the rail only pulling 4.2A @ 11.3v. We're only talking ~48watts here. Most processors alone draw more than that and any modern GPU draws a hell of a lot more than that......

PS doesn't seem to be heating up at all really. Nice and cool and quiet.

Still missing something...

"The +12 V output is produced by two SBR30A50CT (30 A, 15 A per internal diode at 110º C, maximum voltage drop of 0.55 V) connected in parallel. This gives us a maximum theoretical current of 43 A or 514 W for the +12 V output."

Chances are that loading one of the 5v or 3.3v lines will help. Unfortunately, since all supplies are different, it could be the 3.3v line OR the 5v line, no way to tell until you experiment unless you want to reverse-engineer the circuit to determine the current sampling coil. So, it's just easier to temporarily hook up your resistor to each rail, and load whichever rail produces the most benefit the 12v line. Usually, drawing around 1A from either of these rails is enough to stabilize the 12v line, which means 5 ohms for the 5v line and ~3 ohms for the 3.3v line.

BTW: Even at 1A, the resistor should be mounted on a heatsink somewhere.

Also make sure the resistor power rating is sufficient. So, let's say you find that you need to draw 1.5A from the 5v line for sufficient 12v stability; that's 3.33 ohms and 7.5w. You want a resistor that has at least double the power rating of what you calculated.