# Convert ATX PSU to run on 192V DC rail

Discussion in 'Computer Information' started by WiseIndian, Jun 10, 2004.

1. ### WiseIndianGuest

I am wondering if a normal 200V AC input ATX PSU can be converted to
run on a 192V DC Rail, considering the enormous electricity my 16
battery online UPS is using it must be more efficient running the
computers directly an the 192V DC rail and running the monitors on a
seperate offline inverter line . I am not very familiar with how the
ATX PSU works , but it understand the first stage involves converting
the 200V AC Input to 200V rectified DC. anyone guide me if it is
possible?

WiseIndian, Jun 10, 2004

2. ### Ian StirlingGuest

Normal 200V AC input?
where are you?
Do you perhaps mean 220/240?

Ian Stirling, Jun 10, 2004

3. ### philoGuest

yes, you are right

"switching" type power supplies first recity the A.C.
the D.C. is then inverted to high frequency A.C. where it is again
rectified... so in theory if your battery bank equaled the D.C. buss of
your supply, you could skip the first step...

however the power savings would probably be too small to bother with
unless you happen to be running a pretty niced sized "server farm"

philo, Jun 10, 2004
4. ### VWWallGuest

Where do you get the value for "the 192V DC rail"? 16 12V, (6 cell),
batteries in series will give this value. Is it then rectified and
converted to 120/240 V AC? This must be a very large UPS!

The ATX PSU input is nominally 240V AC, which when rectified results in:
240 x 1.44 = ~345V DC. (For 120V AC input a doubler rectifier circuit
is used, resulting in about the same DC voltage.) This DC voltage is
supplied to the switching transistors, resulting in an output to the
final transformer at an equivalent of 40-60 K Hz. This high frequency
allows for the use of a much smaller transformer.

One could design a PSU to run from 192V DC with a somewhat lower
efficiency, but this would be a completely non-standard unit.

Virg Wall

VWWall, Jun 10, 2004
5. ### ricGuest

Yes, but the OP's 192VDC is too low. A SMPS peak charges the input caps
to 1.414 times the input voltage. So, 200VAC would result in about 283VDC
on the caps. The OP's DC value of 192V corresponds to about the same as
134VAC. Much too low.

ric, Jun 10, 2004
6. ### N. ThorntonGuest

There are supplies that will run happily on anything from 110 ac to
240 ac. Not the ones with voltage switches, but ones that will cover
the whole range. One of those should do 192dc ok, though really it
depends on the rectifier/reservoir arrangement. I couldnt guarantee
all will work.

Regards, NT

N. Thornton, Jun 11, 2004
7. ### Tam/WB2TTGuest

There is no guarantee that a 120VAC supply will have a voltage doubler. For
the same power (~300W), I suspect a 170V input supply is cheaper to build
than a 350 V supply. For the voltage doubler, you about doubled the price of
the filter caps, used more expensive transistors, and transformers with more
turns. If you make a million power supplies, and save a dollar on each one,
you start approaching serious money.

Tam

Tam/WB2TT, Jun 11, 2004
8. ### WiseIndianGuest

yes i did mean 220/240V
yes it is a 35 comp plus gaming center with the monitors on a seperate
offline inverter line. the online UPS has a 16 battery bank and seems
to
working at 65-70% overall efficiancy, since we are running 24 hrs a
day
the saving potential is huge.
So, >200VAC would result in about 283VDC on the caps.
true but considering that we run on clean battery power is it
necessary to
have them caps fully charged ?
batteries in series will give this value. Is it then rectified and
it is a 6KVA UPS
actually i dont know too much about smps supplies, but since the smps
is
suppose to work in the range of 90-150V or 190-250V wouldnt that fall
in the band on 192VDC?
sometimes i wonder if the computer manufacturers have a deal with UPS
manufactureres and power utlity companies in keeping us stuck with
costly power guzzling online UPS's?

WiseIndian, Jun 11, 2004
9. ### VWWallGuest

Typo: This should be 240 x 1.414 = ~340 V DC.
You are forgetting the 1.41 factor in converting RMS AC to peak DC.

At nominal 115/230 V AC input the switching circuits run at ~325 V.
The ATX PSU is specified to run at 90V AC. This results in 254 V DC at
the output of the doubler. 192V DC is far too little for the switching
circuit to provide the correct DC outputs, without changing the turns
ratio in the output transformer. Even then, the switching transistors
would be running at about 35% higher current for the same output current.

Even if a current design ATX PSU would work at this voltage, one would
have to provide a non standard input for the 192V DC to the "front end"
DC circuitry in the ATX supply.

VWWall, Jun 11, 2004
10. ### VWWallGuest

I've never seen one without, have you? Actually the cost of switching
transistors is much more dependant on their current capacity than their
voltage ratings. Also, with a fixed "on" voltage drop, the heat will
increase with the current being passed, requiring larger heat sinks, and
resulting in lowered efficency.

For a 240 V DC input , there is no choice but to use the full peak
rectified DC! Check the price of filter caps. You'll find no where
near a double price! The output transformer primary turns are double,
at the higher voltage, but with half the wire size . The output
windings have fewer turns at the same wire size.

It may be serendipity, but the voltage doubling power supply represents
about the most economical design possible. When you consider that a
single SPST switch makes it a 115/130 V unit, you can see why the design
is so universal!

Virg Wall

VWWall, Jun 11, 2004
11. ### VWWallGuest

The DC input to the switching circuitry has to be within design limits.
For 90-140 V this is 255-396 V DC. Your 292 V DC is far too low.
You're forgetting the 1.414 factor between AC and peak rectified DC voltage.
Switching all the ATX supplies and the UPS to 220V might gain a few
points in efficiency, but the re-wiring might not be worth it.

Virg Wall

VWWall, Jun 11, 2004
12. ### Paul Hovnanian P.E.Guest

Its possible but non-standard. Do you already have the 192 VDC rail
available or are you free to change the design?

There are ATX (and other) supplies already built to operate from 48 VDC
busses, this being a standard telecom system voltage derived from
batteries. Also available in 12 and 24 VDC I believe.

Paul Hovnanian P.E., Jun 11, 2004
13. ### ricGuest

At a 230vac input rating, a SMPS will "breakout" at about 170vac, at
which time about 240vdc will be on the HV buss. This is at full load.
Whether or not it will go much lower depends on the load and on the
design of the PSU.
Actually, it is 90-132 vac. 132vac would result in about 370VDC on
the buss (due to doubling), and 180vac would result in about 254VDC.

You would have to modify the supply to stop the doubling, and use
the 90-132 setting (and pray.)

ric, Jun 11, 2004
14. ### N. ThorntonGuest

so youre saying more or less all universal supplies should work on 192 IIUC.

Regards, NT

N. Thornton, Jun 11, 2004
15. ### Ian StirlingGuest

Why not simply stick in a offline UPS?

Ian Stirling, Jun 11, 2004
16. ### glen herrmannsfeldtGuest

VWWall wrote:

(snip regarding ATX power supplies run on 192VDC)
In 230V mode it is a bridge rectifier into the filter capacitors.
In that case 300VDC should pass through just fine. I have
never had a 300VDC power source to test it on, but have believe
that it would work, at least for AT power supplies.

ATX power supplies have the stand by power to supply the
turn on circuits. I don't know how that one works,
but it may have a transformer.

It is convenient that a SPST switch can convert a bridge and
two capacitors into a voltage doubler.

-- glen

glen herrmannsfeldt, Jun 11, 2004
17. ### Robert BaerGuest

The ATX supply is a switcher and if the switch in the back is in the
240V position (not the 129V position), i believe it would not care DC or
AC (25Hz, 50Hz, or 400Hz).
All of these designs rectify the input voltage first, producing a DC
voltage for the switcher.
The "trick" of allowing 120VAC or 240VAC, i think, is switching in a
voltage doubler for the 120V capability.
I would have to dig into one to see exactly what is done.
In any event the efficency is almost exactly the same if run from a DC
source or an AC source, so there will not be any observable gain or
merit.
Also, the efficency will not observably change between a 120VAC source
and a 240V source.
Now, if that inverter is running from 192V DC, thenone would bypass
the conversion loss of the inverter, if the ATX PSUs were connected
directly; make damn sure the switch (for each one) is in the 240V
position!
I think it would be safe, but since i have not investigated these
supplies, i cannot bs absolutely certain.

Robert Baer, Jun 11, 2004
18. ### Robert BaerGuest

*RECTIFY*???? a DC voltage??? NUTS! cannot be done!

Robert Baer, Jun 11, 2004
19. ### Keith WoottenGuest

In message <>, WiseIndian

Just in case it's not clear to the OP, a 'voltage doubler' as used on
the 90-150V can only work with an AC input.

Cheers

Keith Wootten, Jun 11, 2004
20. ### Tam/WB2TTGuest

I am saying this is a marketing issue, not an engineering issue. Also, for
multi voltage use, having to change a strap from bridge rectifier to voltage
doubler is not the preferred way. Preferred by whom? depends on your
management and marketing types.

Tam

Tam/WB2TT, Jun 11, 2004