Here is my take on the server power supply.

I removed the breakout board, figured out which pins need to be shorted to make it turn on, then mounted a bind post to some angle aluminum I had laying around then mounted it on the power supply so that everything looks nice.

I also added (2) 1/4 watt 100ohm resistors between the fan and the psu so that the fan turns at a lower speed and isn't so noisy.

All in all it's a great setup for under $25.

I'm going to start modifying my second power supply and I'll take pictures and document. I was going to make them into a 24v ps, but after reviewing the process I decided that removing the safety features of one of the power supplies wasn't worth the risk so I'm going to live with 600 watts on my PL8.

schu

Well, after playing with these power supplies and my powerlab 8 charger I decided to go ahead and run in series for 24v. At first I had decided that I wouldn't do that and instead ran the power supplies in parallel for a while, but seeing my powerlab8 pull 65 amps and having the 10awg wire get really hot, I decided that 24v is just the way to go.

I didn't want to remove the ac ground since I just didn't like that idea. If we have ac ground on one side and dc ground on the other then why remove the ac ground and play the insulation game. Wouldn't it be better to keep the ac ground and remove the DC ground? I would much rather have a problem on the DC side and besides it solves the physical isolation problem.

So I went to work on this and found that the DL380G4 power supplies I have (same as feather merchant) have 4 posts that mount the circuit board. The two rear ones have a cap between the chassis and ac ground, and the front two are tied directly to DC-. On the left one there are two wire jumpers soldered on the board which where easy to cut, on the right one I cut the corner of the circuit board off and pulled up the trace.

If I had to do it again, I would probably just remove the right post and put a rubber foot under the board since 3 screws will hold the board in just fine, but it doesn't matter, I left enough of the circuit board to fasten with a screw, but not enough of the traces left to make a connection.

The result:

Very nice work. This I think I could do as opposed to the other. What did you have to do or remove to mount the aluminum with binding posts? Did you put the resistors in series? Don't suppose you took any pics of the modification to the board?

Also, what about using a fusible link on the DC output side, so if there were ever an issue it wouldn't be a total melt down?

http://www.amazon.com/Beck-Arnley-17.../dp/B000CRVM12

Any comments on this mod as opposed to clipping the AC ground off one of the cords? When doing this mod do you cut the DC ground connections on both power supplies or just one? Sorry for the dumb questions. I am certainly no electrician.

You don't need a fusable link on the dc side since the only thing that can short is the negative to the positive post which causes the power supply to shutdown due to it's internal fault protection.

Basically, to make the power supplies work in series you need to float the DC side or AC side from ground. Most people float the AC side because it's simple to remove the ground pin from one of the cords then isolate the units, but I thought it better to float the DC side because I didn't like having the AC side floating, didn't want to deal with case isolation, and because it's a lot more convenient to carry around when it's not installed in a box or case or something.

In my case I only floated the dc gound on one power supply and made sure that it was the one that provides the positive for 24vdc. This means that the entire unit is grounded to dc- for 12v off of the first power supply and 24v off of both.

I am thinking about building two more of these for sale to other modelers but they are expensive to build because I use quality AMP connectors (with the nice crimper), good quality binding posts, and Y power cables. Also, the time to modify these is extensive because I have to remove the existing breakout board, jumper the pins to make it turn on, add a resistor to the fan, fabricate the front cover (this time I'll make it one piece so that it mechanically joins the power supplies together), install the binding posts, wire up the binding posts to the studs on the inside, then make the two front mounting posts out of plastic.

Cost of materials and a little labor to put it together would put these at $100 each and that wouldn't include shipping.

On the flip side, it's a really nice package and works very well. If anyone is interested in having me build one of these PM me. I'm not wanting to get in the business or nothing, but I was thinking about building myself a second unit and it wouldn't really be that much more work to toss together a few more while I'm at it.

schu

Originally Posted by akschu

I didn't want to remove the ac ground since I just didn't like that idea. If we have ac ground on one side and dc ground on the other then why remove the ac ground and play the insulation game. Wouldn't it be better to keep the ac ground and remove the DC ground? I would much rather have a problem on the DC side and besides it solves the physical isolation problem.

Schu,

That looks well and neatly done. I don't want to intoduce a problem unnecessarily, but there is still a potential risk of a transformer insulation failure, ie primary - secondary which effectively would connect the output to the mains. That is probably why the outputs are normally earthed; I see no other reason as mutiple earthing often causes noise and signal difficulties.
I would connect some heavy zeners from output to earth as I suggested in the main thread as a safety backup.

Wayne

Wayne,

That sounds like a good idea. I'm assuming that they would be put in like this:

GND -----i>| ----- DC -

How about this one?
1N5364BRLG
http://www.onsemi.com/pub_link/Collateral/1N5333B-D.PDF

I'm not sure how to size these things, I'm thinking it would probably burn out of there was a transformer failure, but perhaps not before the power supply shuts down.

Can you recommend one that will work?

I am going to order a few power supplies and build a new updated version for myself and a few extras for others that want them. But before I place my parts order, I would like to nail down exactly which diode to use.

schu

Originally Posted by akschu

Wayne,

That sounds like a good idea. I'm assuming that they would be put in like this:

GND -----i>| ----- DC -

How about this one?
1N5364BRLG
http://www.onsemi.com/pub_link/Collateral/1N5333B-D.PDF

I'm not sure how to size these things, I'm thinking it would probably burn out of there was a transformer failure, but perhaps not before the power supply shuts down.

Can you recommend one that will work?

I am going to order a few power supplies and build a new updated version for myself and a few extras for others that want them. But before I place my parts order, I would like to nail down exactly which diode to use.

schu

Schu,

Yes that one (33V) or the 28v version should be OK and I agree with connecting it across the 24V rail and retaining the output earth connection on the lower unit. Or you could put a 15V one across each 12V output.
You obviously understand the situation, so even this is not a 100% fix, it must depend on the resistance of any fault path. The lower the resistance, the bigger the zener capacity you need.
If you have an earth leakage trip on the mains input it does not need to be larger, but with a hard short and no trip who knows? We are talking about remote chances but that's what safety is about. There is no guarantee; specifically from me! But it would make it safer.
Just retaining the output earth on the lower unit would help; it's having completely floating outputs that would make me uneasy.

Wayne

Well, I'm going to use a 15v diode on the floating power supply since that is all of the potential it will ever see before we are in some kind of failure.

One thing I'm wondering about is how to test to see if the diode is even doing it's job. As long as nothing blows up when I turn it on I guess I would need to have a real fault to know since that is the only way we are going to see over 15v where the diode is.

I think this is a reasonable step to try and add back the protection removed to make them run in series.

I think this solution is much better than getting rid of the ac ground and isolating the cases. It looks nice too.

schu

Originally Posted by akschu

Well, I'm going to use a 15v diode on the floating power supply since that is all of the potential it will ever see before we are in some kind of failure.


schu

Agree completely provided the lower unit DC earth is retained.
I think you are correct about testing as well. If you use a minimum size mains fuse commensurate with the full load current, then it has the max chance of clearing on a fault without blowing the zener open.
Overall looks best solution.

Wayne

Gents, I have posted a link to this thread in the "How to convert Server Power Supplies" sticky. Interesting alternative approach to the multiple PS solution. I like it. Mind you I don't need any more of these things.

John

Agreed - slickest and most thorough solution yet.

I must admit, I am more than a little concerned these days about folks connecting different power supplies in series without a full understanding of the dangers. High powered chargers demand more power, and to get it many people having heard of this practice are haphazardly connecting power supplies in configurations for which they were not designed for and have not been tested or CE certified.

All we can do is fight the good fight and continue to educate people. Every week now I get a call or two with a story about series connecting two random, disparate power supplies in an attempt to get more power - with varying success. I sincerely hope the chances of disaster are infinitesimally small and that most all modern supplies have adequate protection built in to avoid anything catastrophic. My concern is for the people playing around modified industrial or hobby grade equipment who don't understand the physics involved and that their actions may be potentially dangerous.

Cheers,
David

Dave, that's always been a concern and the reason I started the "Two supplies - safety issues" thread. The problem is I suspect that many of the regular contributors tend to be reasonably knowledgeable while the occasional Google driven searcher who comes across the idea and neither knows nor cares about the whole story may not be.

Not much we can do about it. I would be happy to add your warning to the Sticky thread. Any objections?

John"That

Originally Posted by write2dgray

Agreed - slickest and most thorough solution yet.

I must admit, I am more than a little concerned these days about folks connecting different power supplies in series without a full understanding of the dangers. High powered chargers demand more power, and to get it many people having heard of this practice are haphazardly connecting power supplies in configurations for which they were not designed for and have not been tested or CE certified.

All we can do is fight the good fight and continue to educate people. Every week now I get a call or two with a story about series connecting two random, disparate power supplies in an attempt to get more power - with varying success. I sincerely hope the chances of disaster are infinitesimally small and that most all modern supplies have adequate protection built in to avoid anything catastrophic. My concern is for the people playing around modified industrial or hobby grade equipment who don't understand the physics involved and that their actions may be potentially dangerous.

Cheers,
David

No objections at all .

Done - thanks.

John

Originally Posted by write2dgray

No objections at all .