Connect with us

DC current to computers' PSUs?

Discussion in 'Electrical Engineering' started by abu, May 1, 2008.

Scroll to continue with content
  1. abu

    abu Guest

    Hi all,
    would the PSU of a computer work if I'd provide 220V DC current to it?

    If yes: why everybody wants an expensive sinusoidal UPS, with a big
    expensive inverters in it, instead of using some 20 car batteries in a
    series (+ a rectifier to keep the charge), in parallel to the computers?
     
  2. James Sweet

    James Sweet Guest


    Most of them will work on DC, but it needs to be closer to 330V. Better ones
    with active power factor correction will not though, and smaller power
    supplies designed for 120V only often need more like 170V.
     
  3. Palindrome

    Palindrome Guest

    Computer power supplies do usually rectify the incoming supply and then
    apply that dc voltage to the rest of the supply. So, in theory, applying
    a suitable dc voltage, instead of an ac one, would work. However, in
    practice, there are usually extra components involved that rely on the
    input supply being ac, which makes your idea impractical.
    Something similar is done - many UPS use several 12v batteries in
    series. However, a bank of series connected batteries will still need
    some electronics to keep its output voltage constant. Once you have to
    have this electronics, other factors dominate the design. A few hundred
    volts dc from a bank of batteries is quite difficult to design for -
    especially when considering safety. Keeping it under 100v is a lot easier.
     
  4. There are a bunch of "maybes" involved. One typical design uses
    full-wave rectification on 240 VAC and voltage doubling on 120 VAC.
    It would likely work with 300-340 V applies, since that's the peak
    voltage of the 240 VAC line (though it might overstress an input diode,
    since half the diodes would carry all the current). It would *not* work
    on the 120 V setting with 170 VDC, because the voltage doubler
    configuration requires AC in.
    First, plenty of people do *not* use expensive sine-output UPSes, since
    the square wave (or so-called modified sine wave, which is really a
    modified square wave) type works just fine. Sine output is good for
    motors and other devices that care about harmonic content, but computer
    power supplies mostly don't.

    Second, how much would it cost you to get 300+ V worth of batteries,
    even if each battery is only a few Ah capacity? How much would it
    weigh? You'd need voltage equalizing resistors to keep the voltage
    correct across each battery when charging. You'd need a 300 V power
    supply as a charger. When all this is factored in, you might find that
    it's easier after all to use an inverter to convert a moderate DC
    voltage (12 or 24 V) to ~140 V stepped square wave.

    Dave
     
  5. Guest

    | There are a bunch of "maybes" involved. One typical design uses
    | full-wave rectification on 240 VAC and voltage doubling on 120 VAC.
    | It would likely work with 300-340 V applies, since that's the peak
    | voltage of the 240 VAC line (though it might overstress an input diode,
    | since half the diodes would carry all the current). It would *not* work
    | on the 120 V setting with 170 VDC, because the voltage doubler
    | configuration requires AC in.

    What happens in the PSUs that work on the full 100-240 volt range? There
    are a lot of those around. Do they still double the voltage at some point?


    |>If yes: why everybody wants an expensive sinusoidal UPS, with a big
    |>expensive inverters in it, instead of using some 20 car batteries in a
    |>series (+ a rectifier to keep the charge), in parallel to the computers?
    |
    | First, plenty of people do *not* use expensive sine-output UPSes, since
    | the square wave (or so-called modified sine wave, which is really a
    | modified square wave) type works just fine. Sine output is good for
    | motors and other devices that care about harmonic content, but computer
    | power supplies mostly don't.
    |
    | Second, how much would it cost you to get 300+ V worth of batteries,
    | even if each battery is only a few Ah capacity? How much would it
    | weigh? You'd need voltage equalizing resistors to keep the voltage
    | correct across each battery when charging. You'd need a 300 V power
    | supply as a charger. When all this is factored in, you might find that
    | it's easier after all to use an inverter to convert a moderate DC
    | voltage (12 or 24 V) to ~140 V stepped square wave.

    If a PSU can handle the full 100-240 volt range, AND if it can also handle DC,
    then if the DC voltage varies within that range, how is that a problem?
     
  6. Guest

    |
    |
    | |> Hi all,
    |> would the PSU of a computer work if I'd provide 220V DC current to it?
    |>
    |> If yes: why everybody wants an expensive sinusoidal UPS, with a big
    |> expensive inverters in it, instead of using some 20 car batteries in a
    |> series (+ a rectifier to keep the charge), in parallel to the computers?
    |
    |
    | Most of them will work on DC, but it needs to be closer to 330V. Better ones
    | with active power factor correction will not though, and smaller power
    | supplies designed for 120V only often need more like 170V.

    What about the continuous voltage range PSUs that handle 100-240 volts?
     
  7. Guest

    | would the PSU of a computer work if I'd provide 220V DC current to it?
    |
    | If yes: why everybody wants an expensive sinusoidal UPS, with a big
    | expensive inverters in it, instead of using some 20 car batteries in a
    | series (+ a rectifier to keep the charge), in parallel to the computers?

    It has been suggested that you might need as much as 340 volts DC to make a
    PSU work based on the typical internal designs they have. Be warned that
    DC has very different requirements for fuse or circuit breaker protection.
    An opening fuse or breaker contact that could extinguish the arc at some AC
    voltage will not be able to do so as easily with DC. The reason is that AC
    has a point in time where the voltage drops to zero while DC does not. So
    you can typically see fuses and breakers rated for both AC and DC, but with
    the maximum voltage being less, typically half as much, for DC than for AC.

    20 or more car batteries in series can create a very dangerous electrical
    setup, including possible issues inside the batteries themselves when one
    cell goes bad. It's not that this can't be done. Rather, it needs to be
    done with a substantial amount of engineering for reliability and safety,
    including batteries designed for such setups (not ordinary car batteries).
     
  8. Many electronic ballasts for fluorescent tubes are designed to run
    from both AC and DC. (These are basically switched mode PSU's too.)
    DC operation is generally provided so they can operate directly from
    emergency battery supply. One of the issues with this is as the
    battery voltage drops at end of charge, it was found the ballasts
    were burning out. The reason for this is that as the supply voltage
    decreases, the current in the primary of the high frequency
    transformer increases in order to transfer the same power through.
    When the voltage gets too low, the transformer overheats. Most
    electronic ballasts contain shutoff circuitry nowadays to avoid
    this.

    In the case of a 100-240 volt AC range, this maps onto a 140V-340V DC
    capacitor voltage, so I would caution against trying to run a 100V AC
    PSU on less than 140V DC, as you might again burn out the primary if
    you are drawing any significant load. I would also caution against
    operating at the 340V top end for the reason another poster gave -
    protection circuitry such as fuses and thermal trips may not be able
    to break a DC current.

    Also note that many SPMSU's now have more complex input stages to
    get the power factor up well above 0.9. These may not like DC at all,
    unless specifically designed for it. I haven't looked in to how they
    work.
     
  9. Guest

    | In article <>,
    | writes:
    |> If a PSU can handle the full 100-240 volt range, AND if it can also handle DC,
    |> then if the DC voltage varies within that range, how is that a problem?
    |
    | Many electronic ballasts for fluorescent tubes are designed to run
    | from both AC and DC. (These are basically switched mode PSU's too.)
    | DC operation is generally provided so they can operate directly from
    | emergency battery supply. One of the issues with this is as the
    | battery voltage drops at end of charge, it was found the ballasts
    | were burning out. The reason for this is that as the supply voltage
    | decreases, the current in the primary of the high frequency
    | transformer increases in order to transfer the same power through.
    | When the voltage gets too low, the transformer overheats. Most
    | electronic ballasts contain shutoff circuitry nowadays to avoid
    | this.
    |
    | In the case of a 100-240 volt AC range, this maps onto a 140V-340V DC
    | capacitor voltage, so I would caution against trying to run a 100V AC
    | PSU on less than 140V DC, as you might again burn out the primary if
    | you are drawing any significant load. I would also caution against
    | operating at the 340V top end for the reason another poster gave -
    | protection circuitry such as fuses and thermal trips may not be able
    | to break a DC current.
    |
    | Also note that many SPMSU's now have more complex input stages to
    | get the power factor up well above 0.9. These may not like DC at all,
    | unless specifically designed for it. I haven't looked in to how they
    | work.

    Then we might agree that the OP's safe choice is to do 48VDC and get the
    appropriate PSU.
     
  10. abu

    abu Guest


    Thank you all for your suggestions. I understand that it's not that
    easy: one big problem problem I see after your comments is that all PSUs
    are made differently and I would probably not be able to find a DC
    voltage that is ok for all (also some have suggested that active PFC
    ones wouldn't work with any DC voltage) and in most cases I cannot
    replace the PSUs that are embedded in the servers we have.

    I know 48v DC-to-DC PSUs exist, however 1) their cost is high ($130 at
    least) 2) I would need a lot of relatively thick wires for the high
    amperes due to the low voltage 3) I would need to replace all PSUs in
    our servers, and as I said most servers come bundled with a PSU and I
    don't think I can replace those 4) more importantly, I seem not able to
    find big 24v UPSs around, such as a 15kVA 24v UPS.

    It is interesting what Dave has said: square wave or modified sinewave
    UPSs should be cheaper (possibly much cheaper?) than sinewave UPSs,
    unfortunately I cannot find a >10kVA UPS with modified sine wave or
    square wave, all big ones are true sine wave and are very expensive. Or
    do you have a link?

    I kinda guess that modified sinewave ones are not so much cheaper,
    presumably because they still want to keep output voltages very precise,
    so there is expensive power electronics for that, after which making the
    true sinewave is not much more expensive. In reality I think computers
    don't need such precise voltages at their input, I think that a +-2v
    precision should be OK for computers. 2v is the voltage of a battery
    cell, so that means that a cheap & big UPS should just have a large
    array of 2v cells in a cascade and then add or remove a few cells from
    the cascade in real time so to keep an output of 310 (340?) +-2 volts.
    Then it could use this voltage to do the 3-levels stepped sine wave.
    Don't you think?
    A 15kVA UPS made like this I think doesn't exist, or does it? (link?)

    300 cells (car-like, like 50 car batteries) in a series could produce
    360V with cranking amperes (CA, according to car batteries specs that is
    about 600 CA each battery maintaining 7.2v per battery so 300x50 =
    30,000A at 7.2*50=360V), that is about 220*30000 = 6.6 MVA with little
    more than the cost of 50 car batteries (some $4000...)

    Wouldn't that work?
    I understand there would be some safety issues putting 50 car batteries
    together... probably this should be kept in a separate building

    Thanks for all your comments
     
  11. Guest

    | 300 cells (car-like, like 50 car batteries) in a series could produce
    | 360V with cranking amperes (CA, according to car batteries specs that is
    | about 600 CA each battery maintaining 7.2v per battery so 300x50 =
    | 30,000A at 7.2*50=360V), that is about 220*30000 = 6.6 MVA with little
    | more than the cost of 50 car batteries (some $4000...)
    |
    | Wouldn't that work?
    | I understand there would be some safety issues putting 50 car batteries
    | together... probably this should be kept in a separate building

    And keep people away. Maybe even yourself.

    When a cell gets weak, it gets "back charged" by the rest of the system under
    the forward current flow. That weak cell presents more of its own internal
    resistance instead of a forward potential in these cases. The higher the
    overall system voltage, the more it can continue to push the current through
    the dead cell. Backwards charging a lead-acid cell will damage it and make
    it even weaker still. Eventually, you get to a point with a high voltage
    system that the voltage is able to arc across between the plates inside the
    cell. The temperature would already be rising from the back charging, but
    the arc will now add to that. The cell could soon explode. If it did not
    have pressure release, it could be very violent (think: hot sulfuric acid
    rapidly spraying everywhere, along with toxic gasses). It can ruin your day.

    I would not put car batteries in series beyond 48 volts. And even then I have
    some worries. Marine batteries may be better designed for this. You might be
    better off getting surplus submarine batteries.
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-