Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!apple!usc!bloom-beacon!mit-eddie!uw-beaver!fluke!jat
From: jat@tc.fluke.COM (J. Andy Tiura)
Newsgroups: sci.electronics
Subject: Re: PCs and line power
Keywords: 3 phase balance, diode bridge supply
Message-ID: <9355@fluke.COM>
Date: 26 Jun 89 17:03:05 GMT
References: <4391@tekig5.PEN.TEK.COM>
Sender: news@tc.fluke.COM
Distribution: na
Organization: John Fluke Mfg. Co., Inc., Everett, WA
Lines: 48

In article <4391@tekig5.PEN.TEK.COM> pavelz@tekig5.PEN.TEK.COM (Pavel R. Zivny) writes:
>
>I just witnessed a little problem with power distribution in certain computer 
>'lab'; this is 3 phase, 208/120 V, 100A circuit loaded mainly with computers
>and such. It turned out to draw about 60 60 45 A from respective phases. AND,
>IT WAS TAKING ABOUT 100 A from Neutral wire.
>
>This is a problem, since the neutral is a 100 Amps wire only. 
>
>	Why do the phases balance out so poorly ?
>
>	Well, I have a guess. All these "diode bridge - filter cap"
>pwr supplies (most of them switching, but that makes no difference here) take
>most of the power at the moment of voltage peak on the line, right.  And since
>the peaks on different lines happen at different time (120 degrees shift) -
>they can't balance out - poor thing neutral has to carry every each of these
>peaks out of the door.

You are partially correct.  The 120 degree difference between phases would
allow a "summing to zero" for the neutral current with LINEAR balanced loads,
reactive or not.  Your model accurately describes a NON-LINEAR load that
creates significant harmonic components in the phase currents.  The third
harmonic component of the phase currents land right on top of each other on a
phasor diagram and seem to add up instead of cancel out.  (Yes, I know that's
not entirely accurate, but it's easy to mentally picture.)  

>	Well, I see three possible solutions:  
>a) Thicker neutral (not very elegant solution, since re-wiring of an existing
>    system is going to be a pain) 
>b) Better supplies (questionable; how much are willing to spend on clean
supply ? How are you even going to know which ones are the better ones ?
>c) _Some_ filter (balancer ?) or what. Like fly-wheel loaded, 3 Phase
>    asynchronous motor; or some cleverly switched bank of caps; or ....

b and c are viable options.  Look for power supplies that boast "active power
factor correction" or buy an outboard unit to perform the the same task.
These are available from several vendors, using both active and passive
techniques, to make the non-linear power supply input impedance appear
resistive.  The line current then becomes essentially sinusoidal, and the
neutral current problem you mentioned goes away.  (As an additional benefit,
you can draw more power from a given branch circuit if the load is power
factor corrected.)

Look for more power supply vendors to incorporate power factor correction
into their OEM supplies now that at least three IC vendors have introduced
chips that perform that function.

-jat@tc.fluke.com