AT PSU conversion

Discussion in 'General Electronics Chat' started by adam555, Sep 21, 2014.

  1. adam555

    Thread Starter Active Member

    Aug 17, 2013
    858
    39
    Hi, I'm about to convert an old AT PSU into a bench power supply, but there are a couple of issues that I need help with before starting.

    - The first is about the cable that you need to connect to ground to turn the PSU on (normally a green cable): well, it doesn't have one; the only cable that is not one of the outputs or ground is a yellow cable that connects to "Power Down" on the board. The PSU turns on without shorting any cable to ground -just with the main switch- so I'm wondering if this cable is to turn it off or something else, and if I should use it or leave it disconnected.

    - The second issue is the dummy load that most sites recommend. I turned the PSU on to test it, and it seems to work fine without any load; so, do I really need to add a dummy load, or is it just for ATX PSUs?

    - Most sites also say that you need to connect all cables of the same color together to output the maximum amperage, but this PSU is just 78W (the highest amperage is for the +5V with 7A, followed by the the +12V with 3A; the rest are all under 1A).

    - And last, do these PSUs have any protection, other than the fuses -like a current limiter or something-? I mean, what would happen if I exceed the maximum amperage -which for the -12V is as little as 0.3A- or if I accidentally short the outputs?

    Thanks in advance for your help...
     
  2. ian field

    Distinguished Member

    Oct 27, 2012
    4,413
    782
    On one cheap - low power ATX PSU I had on the bench, the 3.3V sense wire had a plug pin to itself rather than crimped into the same pin as one of the actual output wires - my dummy load didn't take that possibility into account and the PSU went bang when I switched it on! - since then all my dummy loads are wired to a section cut from a scrap board with the connector on it.

    The type of regulation error sensing circuit is difficult to predict - I've seen a few with a resistor network summing the 3.3 & 5V rails into a single TL431 that drives the feedback opto-coupler. In any event, you usually have to give the 3.3 & 5V outputs at least some load to get any current out of the others.

    If it doesn't have the on/off switching wire; it could be an old AT PSU - they're much easier to modify to a bench PSU but don't usually have 3.3V AFAICR.
     
  3. MaxHeadRoom

    Expert

    Jul 18, 2013
    10,500
    2,364
    Also if using this as a bench supply, keep in mind PC Power supply commons are usually connected to earth ground.
    Your reference to ground I assume is power common?
    Max.
     
  4. paulktreg

    Distinguished Member

    Jun 2, 2008
    609
    120
    The AT form factor power supply does not have the green power on signal. It switches on when mains is applied and issues a power good signal (+5V) to the motherboard that is needed before the PC boots up.
     
  5. adam555

    Thread Starter Active Member

    Aug 17, 2013
    858
    39
    This one is not an ATX PSU, it's a very old AT. Actually, I think it's not even AT, since I believe AT began with the 286 clone PCs; and this one is from an older 8088.

    It doesn't have the 3.3V, nor the "Power Good", "Power On" and "3.3v Sense"; just the other voltages and a yellow cable that connects in the board in a place labeled "Power Down". I tested it for a moment and it turns on with just the main switch; so I don't know that this cable is for, nor tested it before asking.

    I'm a bit confused with the dummy loads. Some sites say the dummy loads are needed to turn on the PSU; other say it's necessary to have current in the other voltages; and other advert that a PSU without them will be damaged or will even burn. Is there any safe way to test this, and if this PSU needs one and of how many watts?
     
  6. tom_s

    Member

    Jun 27, 2014
    285
    333
    at power supply with the p8 and p9 connectors (thanks google) with a mains power on switch.

    atx power supply sends voltage to the motherboard and the motherboard turns on the supply by pulling power good low.
    at power supply, its the power supply that tells the motherboard the power is good. pin 1 can be safely ignored.

    as the switch on works differently, i doubt it would need a load to switch on

    not sure what pin 2 is (key)

    [​IMG]

    edit: typo
     
  7. adam555

    Thread Starter Active Member

    Aug 17, 2013
    858
    39
    I saw those connections for AT and ATX formats yesterday, but they don't match my PSU; this one is older than the AT format, with 17 pins in one row instead of the 12 from the AT.

    There is no "Key" or "Power Good" on this PSU; the only cable that's not ground or any of the voltages is a yellow cable that is marked as "Power Down".

    I checked and it doesn't need any cable to ground nor a load to turn on, just the main switch is enough. What I don't know is if I can damage it without a load, and what the "Power Down" yellow cable is for.
     
  8. ian field

    Distinguished Member

    Oct 27, 2012
    4,413
    782
    They usually have a bleed resistor on all the outputs, so the rails might well come up with no external load - but the minus and 12V rails will collapse as soon as you load them if the +5V isn't loaded.

    The +5V rail is the one that is regulated - it controls the mark-space ratio of the power switching, if that isn't loaded there will only be narrow pulses in the transformer (very little energy).

    The early low powered ones had 78xx/79xx regulators for the other rails, as power ratings increased, it exceeded the ratings of a 3 terminal regulator and subsequent designs relied on very tight coupling between the secondary windings - balancing chokes also became common, but I'm not entirely clear on how they worked.
     
  9. adam555

    Thread Starter Active Member

    Aug 17, 2013
    858
    39
    So I will need a dummy load on the +5v anyway. How do I calculate the watts?

    Also, do you think it's true what they say about damaging the PSU without it. I'm asking because I was about to test how much power I can get out of the 12v without and with different loads on the +5v.
     
  10. ian field

    Distinguished Member

    Oct 27, 2012
    4,413
    782
    You can get a rough idea by loading it with a 12V 21W indicator bulb, that should light a bit less than half bright, if that doesn't hold up the load you put on the other rails - try a H4 bulb filament. You can guesstimate the equivalent resistance of whichever filament.

    You don't exactly have to calculate to 4 decimal places.
     
  11. adam555

    Thread Starter Active Member

    Aug 17, 2013
    858
    39
    I have hundreds of LEDs of all types and colors, but not a single bulb shorter than 220v 40W. :(

    I was planning instead to test the 12V with a motor -good for 12v and more than 200mA- and increase the current with a potentiometer; this ways see if it works without and with different loads on the +5v.

    If I understood correctly, the bigger the load on the +5v, the more current I can get out of the other rails; is that it?
     
    Last edited: Sep 22, 2014
  12. adam555

    Thread Starter Active Member

    Aug 17, 2013
    858
    39
    Got 1W out of the +12v without any load on any other rail...
     
  13. adam555

    Thread Starter Active Member

    Aug 17, 2013
    858
    39
    Went up to 200mA with the -12v without dummy load and I would probably could have gone all the way to the maximum 300mA for that rail; just that I didn't want to risk it, since 200mA is more than enough for what I'm planning to use it for.

    Anyway, have to postpone the test until I get another fuse for my multimeter. :p
     
  14. ian field

    Distinguished Member

    Oct 27, 2012
    4,413
    782
    There shouldn't be any risk to a decent design PC PSU, they usually have a bank of voltage comparators that trip a shutdown if any rails go out of spec.

    But there are a few cheap & nasty one's out there - not *THAT* big a loss if they do go bang.
     
  15. adam555

    Thread Starter Active Member

    Aug 17, 2013
    858
    39
    I think this one is a good brand. It came out of an Epson PC -an 8088- and these where way before cheap clone parts were made.

    So, supposing that I can get the full power out of each rail without any load on the +5v, should I still use a dummy load? If anyone thinks the answer is yes; then could you please explain why.

    Also, is there any safe way to test what the "Power Down" wire is for? I was thinking about putting it to to ground, or to +5v, and see what happens; but it seems rather risky.
     
  16. RamaD

    Active Member

    Dec 4, 2009
    254
    33
    The very initial PC power supplies required a minimum load on the +5V rail, without which the power supply could fail. The technicians usually connect a FDD as a load for checking the power supplies. Some even carried a 1Ohm 10W resistor connected to a FDD mating connector on the +5V rail to test out the power supply. Those days, the tantalum bead capacitor of 10uF in +5V motherboard used to go short, and the power supply would not start. That is why they needed to check the power supply without connecting the motherboard. Pretty soon, the power supplies improved and worked without the minimum load, though the regulation was not guaranteed. The tolerances are 5% on +5 and 10% on +12V power supplies.
    Even the first power supplies came with the power good signal and it is pin 1 of the connector that is closer to the rear end (expansion connector end). This signal was used inside to keep the CPU under reset till it became high. And there was no power down signal.
     
  17. ian field

    Distinguished Member

    Oct 27, 2012
    4,413
    782
    As I explained in an earlier post - usually only the 5V output controls the regulation circuit, once that rail is up to its specified voltage the control chip throttles back the mark/space ratio. If the chopper transformer is only getting tiny little pulses, there isn't the energy for a high current draw on any of the other rails. With the majority of PC PSUs the only cure is to draw some current at 5V to fatten up the mark space ratio.

    When I made my living repairing PSUs (amongst other things) a local trader who brought me work - he was in the habit of 'testing' PSUs with a spare floppy drive, and lumbered me with piles of PSUs that weren't faulty - it was a total PITA sorting through the pile to find the ones that actually did need repairing.

    A few top quality PSUs might be encountered that have separate transformers and switching transistors for different groups of outputs - I found one such PSU in an optical network hub, but it only supplies 12V & 5V.
     
  18. adam555

    Thread Starter Active Member

    Aug 17, 2013
    858
    39
    I haven't looked at the transformer or voltage regulators, but as I said above I tested the PSU without any load on the +5v and it seems to deliver the full power on all the other rails; which are -5v, +12v and -12v (doesn't have 3.3v).

    Also just tested the wire labeled as "Power Down" and it outputs +5v with respect to the ground wires.
     
Loading...