I agree with #1 and #2, so I'm gonna drop the idea. As for #3 (the load sharing), I thought it could eliminate the MOSFET alltogether, but because of #1 and #2 (improper startup voltage, and fluctuations), it's not feasible, and I agree.

Oh well :( Thanks for the input though, I really appreciate it.

Received the Maxim 1614 sample today. I should be able to drive the 12v MOSFET, no matter what. Now to find the time to build the darn thing :( :mad:

Damn dude, by the time you get around to doing this you will be able to buy a 600watt supply for $20. They already have 550watters for $22ish + $15 shipping on e-bay. :D

http://cgi.ebay.com/ws/eBayISAPI.dll...&category=3670

I need some help!


I've got the Maxim 1614 high-side MOSFET driver, but it's too small to use.

I want to use an adapter from http://aprilog.com/perl/main.pl , but I can't figure out which one!

Any help would be greatly appreciated!

Here are the Maxim package specs:

Oh, otherwise, I'm going to build the circuit, hopefully this weekend, and compare the "forward" and "reverse" MOSFET configurations, with a dummy load, on the +5v line.

edit:
It occured to me that the reason that current does not flow from one PSU to another, while they're on, is because there's a load on the rails. Not a significant revelation, but it got me thinking about how there should be an internal resistance to the supplies, as there is with any supply, and that there ought to be a minimal load on the array, to keep things balanced: the lower the load, the more out of balance the load would be on the PSUs, until there's no load, at which point the PSUs just feed each other.

So I'm wondering if a simple 10K resistor to ground would take care of it, meaning, making sure that there's always a load, and allowing the balancing circuit to actually keep running, while avoiding the feedback problem. Extended, I figure that this load resistor would actually have to be sized relative to the PSU's internal resistance.

Does anyone know if I'm on the right track here, or if I'm just blowing smoke?

That is why I had the resistors on the "final vesion."

I just threw in a 1k resistor on each side, if I remember correctly. I'm not sure about being porportinal to the internal resistance of the supply, though :shrug:

Yeah, I remember going through a procedure to determine a battery's internal resistance, but I'm not sure if it applies the same way with a PSU, since the lines are regulated. But I'd still like to have an idea.

Maybe 1K is plenty:shrug:

I think I got the adapter right, but the drawing doesn't match the pic.:shrug:

http://smt-adapter.com/perl/part.pl/...6D3-SMT-S.htm#

So I e-mailed their sales dept: I guess I'll get a reply Monday.:rolleyes:

1st build: failure!

The driver is outputting only 4 volts, with a 5 volt supply.

The MOSFET is outputting 1 volt.

:(

I think that the driver was a bad choice. I'm not convinced that this "pump-charge" was designed to output a steady voltage.

I'll try again, with another assembly, with the MOSFET in reverse.

Build#2: failure.

Strangely, the driver is actually outputting a steady 13.7 volts, but the MOSFET in reverse, outputs nothing, nada, zip...

:mad: :shrug:

edit: it was a broken connection!

Attempt #3: it works!

Voltage across the MOSFET was 0.12 volts.

Turning off the driver left a load ON, but I didn't measure the potential there...

Starting up...

It won't start!?!?

All outputs are ~14v, except the one I'm using, which is at zero. Blown?:shrug: (using a 16v zener, to ground, on the driver line)

I found the Maxim chip adapter: it's p/n 8SSO-D3-SMT-S.

After having had a chance to think about yesterday, I thought I'd share these thoughts:

-It's difficult to solder the small ICs, and I may have not soldered them well enough. I'm guessing that pre-tinning the adapter is best.

-I don't know if I blew a driver output: I'll have to fire it up again, sometime. I might order the 50 Ohm resistor first, since I didn't use any. I'm aware that they're static sensitive, but I don't think I zapped anything.

-The first failure may have been caused by a damaged MOSFET, which I twisted severely, when mounting the heatsink. I had marked that MOSFET with a red marker.

-If one of the driver lines is blown, it's ok, 'cause I'll be using the Maxim chip to drive the 12v line.

-I an NOT looking forward to soldering the Maxim chip to the adapter, once I've ordered it.

-Blowing the line seems to have occured as I left the driver on, when powering it up.

As posted in the Johnson pump thread:

Ok. Because I'm building a PSU array, and that I don't have room for a 4th power supply (!), I'm forced to rethink the idea of putting this Johnson pump on a seperate PSU. Note also, that I plan on using TWO of these pumps on the 12v line, and an automotive blower, on the +5 line.

So I need to have a better idea of how running this pump might affect the overclockability of a system. Opinions?

Ok, I have to admit I haven't read through the past 8 (!) pages of posts, so I'm not sure if this has come up. TI has some power supply load share controllers. Basically you hook one up to each output, and it senses and balances out each of the outputs. Here's a quote from the product page:
here are all of TI's offerings in this category.

Thanks Rufus. I actually did look into it (and linked earlier). The problem is that the TI chip requires a + and - "remote sense" output from the power supply, which is infinitely difficult to implement. In fact, I opted to drop it, in favor of YoDuh's solution.

I managed to put together a tray, from one of those dead CD-ROM drives, and by trimming off the back of the old PSU case.

It was a long cutting job (hack saw), but three rivets later, and minor adjustment/trimming :D and everything lines up.

Perfect fit!:)

I'll have pics next week.


The next step is to mount some kind of bracket, to support the rear connectors. Then I'll re-use the CD-drive's plastic sheeting, to make sure that the PSU PCB is completely protected from the tray (preventing shorts). I ordered some mounting posts to make it all happen.


I'm going to have to put the watercooling aspect aside, until I can source the perfect connectors, or decide to make that compromise, and make it all permanent.:shrug:

It's too late for me now, but someone else might be interested. I found the ideal power connectors:

http://www.andersonpower.com/product...pd_family.html

Here's an early pic of the circuit board:

Here's a pic of the 1163 drivers, mounted. The board is an SMT to DIP converter.

Here are the MOSFETs, pre mounted. Tip: the pins don't have to be in a straight line, it's easier to split them up over two rows.

Here's the part of the PSU case that I trimmed off:
(by hand, thank you very much!)

...which fits (essentially) perfectly in 2 * 5 1/4 drive bays:

Damn, that does fit nice. Looking good!

Thanks!

It'll need a bit of polishing up, but I plan to add a lit fan, a 6" blue CCFL, a few status leds, a handle, and a lit power switch. Otherwise, the remaining holes will be plugged, and the front plate will be painted.

That fan hole is going to be opened up, and I'll put a grill there instead, for maximum flow.

Did this get finished BB2K? Made for an interesting read (although over my head in numerous places!) :)