3-Phase Power Converter Help

[faultywarrior]

Don't know if this goes here or in general chat, so mods - feel free to move it if needed.

Anyhow - I'm planning to obtain a new vintage computer for my collection next year (yeah, it needs a bit of planning!).  IBM S/390 Gen 6 - i.e. a huge mainframe.  The problem is, it needs 3-phase power at anything over 208 volts.  Here in the US, the easiest way to get 3-phase in a residential environment is to use a 3-phase converter to take to the 240v single-phase (well, split-phase really, 'cause we're a bit dumb here and thought 120v was a good standard to start with) and push that to 240v 3-phase.  That works fine for single loads - motors, heaters, etc. - but IBM decided to do some weird things that I can't wrap my head around.

The first (and this is quoting their installation manual): "The S/390 G5 and G6 Enterprise Server models are designed with a fully redundant power system. Each computer has two line cords attached to two power input ports which, in turn, power a fully redundant power distribution system within the computer." - OK, no big deal - pretty much any server has redundant power.

Where it starts getting weird is here (again, directly quoted): "The S/390 G5 and G6 Enterprise Server models require three phase power. Under normal operating conditions, current is drawn only from the first two phases. The third phase only draws current in the event of power loss to one of the other two phases." - uhh, ok - so we've now got redundant-redundant power...

The second weird thing is this (also directly quoted from the manual) "If several of these power cords are fed from the same power panel, the load on that panel will be unbalanced." - this I don't understand, and this is where I need help.

My original idea was to just tie the two power feeds together to the output of a rotary 3-phase converter.  I don't know if this will work, or if it would damage the rotary converter from being unbalanced.  It also brings to light the question - Since each feed is inherently unbalanced, would a rotary converter be in danger if powering the machine AT ALL?  Do I need to find some other way of getting 3-phase power?  A static converter seems to do just this - only make use of the first 2 phases.  They're also cheaper, especially given the fact that I need something that can handle 5kW of resistive (or inductive...not sure what you'd categorize a computer as?) load.  Other options I have is to use a digital phase converter (expensive) or a motor-generator package (expensive, loud, space consuming).

Any input from some electrical gurus would be appreciated!

[senso]

Its unbalanced because if you wire everything up to code/regulations all your power supplies will be only putting load in two of the three phases from each distribution panel, so you either swap two phases in half the cords you will load two phases and one will see zero load, thats what they are saying.

[Siwastaja]

It's very well possible it won't need 3-phase power at all, but it's just using 3-phase wiring to get enough power, and for redundancy - this makes sense, since power distribution is 3-phase, and large establishments using these mainframes had 3-phase supply already.

I suspect this because they are talking about the third phase being a redundancy thing, and about the imbalance. This means the remaining two phases are used in single-phase configuration, just to power standard single phase transformers, single phase motors, or switch mode converters that are happy with single-phase rectified voltage.

You should do some extra homework to find out if you can just power one of the "phases" (you can actually just try this out), or connect the inputs in parallel to your 240V line. If this is the case, a rotary converter would be a massive waste of money, space and energy for no reason whatsoever.

[faultywarrior]

Its unbalanced because if you wire everything up to code/regulations all your power supplies will be only putting load in two of the three phases from each distribution panel, so you either swap two phases in half the cords you will load two phases and one will see zero load, thats what they are saying.

Yeah, that makes sense.

It's very well possible it won't need 3-phase power at all, but it's just using 3-phase wiring to get enough power, and for redundancy - this makes sense, since power distribution is 3-phase, and large establishments using these mainframes had 3-phase supply already.

I suspect this because they are talking about the third phase being a redundancy thing, and about the imbalance. This means the remaining two phases are used in single-phase configuration, just to power standard single phase transformers, single phase motors, or switch mode converters that are happy with single-phase rectified voltage.

You should do some extra homework to find out if you can just power one of the "phases" (you can actually just try this out), or connect the inputs in parallel to your 240V line. If this is the case, a rotary converter would be a massive waste of money, space and energy for no reason whatsoever.

I was bouncing this idea off a guy who's got a generation newer machine (z890), and he did just that - wire one of the 120v legs of the split-phase 240v to each of the first two phases and left the third phase unconnected.  I'll have to wait until I get the machine to try it...I can't imagine anything bad would happen - it'd probably just refuse to power on and would give me a PSU input error.

[richard.cs]

It's very well possible it won't need 3-phase power at all, but it's just using 3-phase wiring to get enough power, and for redundancy - this makes sense, since power distribution is 3-phase, and large establishments using these mainframes had 3-phase supply already.

I suspect this because they are talking about the third phase being a redundancy thing, and about the imbalance. This means the remaining two phases are used in single-phase configuration, just to power standard single phase transformers, single phase motors, or switch mode converters that are happy with single-phase rectified voltage.

You should do some extra homework to find out if you can just power one of the "phases" (you can actually just try this out), or connect the inputs in parallel to your 240V line. If this is the case, a rotary converter would be a massive waste of money, space and energy for no reason whatsoever.

I was bouncing this idea off a guy who's got a generation newer machine (z890), and he did just that - wire one of the 120v legs of the split-phase 240v to each of the first two phases and left the third phase unconnected.  I'll have to wait until I get the machine to try it...I can't imagine anything bad would happen - it'd probably just refuse to power on and would give me a PSU input error.

Based on what you've said I would expect that to work, but that it might give some kind of warning as it has no redundancy in that configuration.

[faultywarrior]

Based on what you've said I would expect that to work, but that it might give some kind of warning as it has no redundancy in that configuration.

Yeah, the other option I have is to use a 3-pole breaker and run 3 lines to the machine.  Since it appears that the machine doesn't care about the actual phase of the input power, this should silence any complaints of no 3rd leg.