[Beowulf] DC Power Dist. Yields 20%

[Jim Lux]

At 02:00 PM 8/15/2006, Geoff Jacobs wrote:
>Jim Lux wrote:
> > At 08:41 AM 8/11/2006, Geoff Jacobs wrote:
><snip>
> >
> > Sure you would.. Actually, these days, you might use an IGBT, depending
> > on the switch rate..
> >
> > Switching PSUs all rely on rectifying the AC supply to generate a DC bus
> > voltage, that is then converted to the DC voltage you want.  There's a
> > half dozen or so circuit topologies, but they all basically rely on
> > turning the DC on and off quickly and then doing something useful with
> > it.  The two basic strategies are: You turn it into high frequency AC,
> > then run that through a transformer, and rectify and filter it to low
> > voltage DC.  THis saves the iron and copper cost of a high power
> > transfomer (since the core mass is inversely proportional to
> > frequency... 20kHz or 100kHz saves a bunch over 60Hz)
>
>Okay, just like modern inverter-based TIG supplies. :) But aren't they
>already doing this at the prime rectifier to step up to 300-ish V DC?



No, you get 300VDC just from the offline rectifier.  either 230 into a 
bridge (for 1.414*230) or 115 into a doubler for 2.828*115.. both give you 
320V or so.  If you have three phase 480, the phase to neutral voltage is 
277V, which gives you 380ish out of the 6 pulse bridge.

Either way, you need some sort of transformer to get line/load isolation 
for safety.   A DC/DC converter (say a 24V or 48V bus regulated down to 5V) 
might not have any isolation ( a straight buck converter, for instance).


> > The other strategy is to take the HV DC, and pulsewidth modulate it,
> > then filter it, to make a lower voltage (the classic "buck" converter).
> > That is, take 100VDC, turn it on and off with a 5% duty cycle, low pass
> > filter, and get 5V.
>
>I see, and switching speed is dictated by factors like desired ripple in
>the output voltage for a given load while still avoiding the use of
>those coke-bottle caps in the LC filter (preserving such supplies for
>USAF rail-gun experiments).

Precisely.. In many conveters, the energy storage is in a combination of a 
series inductor and a shunt capacitor.  You might choose the switching 
frequency so that it matches some resonance (so you can switch at zero 
crossings, for instance), or to trade off the loss in the inductor and 
capacitor (both of which vary with frequency, but differently).

The basics of switching converter design are easy.  The subtlety is in all 
the tradeoffs to get the efficiency high.  Everything comes at "some" 
cost.. inductor loss (core and skin effect), capacitor loss(ESR), switching 
losses (during the switch transition time, the switch is "partly on"), 
switch device conduction losses (when it's on, it still has some resistance 
or voltage drop), rectifier losses (both switching, recovery, and forward 
drop). And, then, you have to deal with things like EMI/EMC (a 1 kW 
switcher can make a fine broadband RF jammer), various and sundry 
regulatory requirements (so it doesn't catch fire, cause the electric 
company to hate you for poor PF, etc.)


>Are power supplies using individual converters for each rail these days?

It depends.. mostly on the relative regulation and load currents.  If the 
loads go together, then you can basically put two secondaries with 
different voltages, driven by the same primary, and regulate the pulsewidth 
(etc) to keep one voltage steady, hoping that the other one will follow. 
WIth reasonably "stiff" voltage output designs, this works pretty well. If 
the loads are very different (or, more importantly, change differently) or 
there's a lot of loss in the output rectifier/filter, then you might need 
separate switchers/regulators for each voltage.

Imagine if you had 0.01 ohm winding resistance for each of the 5V and the 
3.3V outputs... and then you load the 5V supply down with 10 amps, while 
the 3.3V output is lightly loaded.  That's 0.5V drop, so the regulator 
increases the input voltage up to 5.5 to keep the output at 5V.  But now, 
the 3.3V is at 3.63 V, and probably too high.

So now you know way too much about switching power supply design....

And just to keep a bit of Beowulfery in the discussion.. in Switching 
supplies, everything hinges on the design of the magnetic components: the 
cores and windings. And those are designed (these days) with (moderately 
complex, but parallizable) FEM codes to model the magnetic flux in the core 
and conductors, and actually allow tradeoffs of conductor size and shape, etc.


Jim