This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ATX power supply supervisor

Hello everyone.
I've got some problem related to ATX power supply.

I've got NUC type PC (with rather weak power supply) and I wanted to use external GPU which need to be powered on by ATX power supply.

So I will try to describe what I wanted to do ;)
In normal case I need to turn on my PC and ATX power supply at the same time (with minimal delay) - so there will be two switches, one on my PC and one tied to PS_ON pin and GND.

What I wanted to do was to remove manual power switch from ATX power supply for automatic startup at the same time as my NUC does.
So I'm looking for some IC which will be connected as a bridge between both devices :)

Could you help me with this? I saw some designs on internet so I know it's possible.

Thank you.

  • Hello Mateusz, thank you for posting in our e2e Forum.

    NUC seems to refer to the Intel Industrial grade form factor.

    It is powered by an external adapter, 19V, 65W, which is always ON.

    When you activate the ON/OFF on the NUC, it just sends a logic signal to the system to enable the processors, much like the PS_ON signal of the ATX.

     

    From an ATX spec I have in my archives:

    PS_ON# is an active low signal that turns on all of the main power rails including 3.3VDC, 5VDC,

    -5VDC, 12VDC, and -12VDC power rails. When this signal is held high by the PC board or left open

    circuited, outputs of the power rails should not deliver current and should be held at a zero potential with

    respect to ground. Power should be delivered to the rails only if the PS_ON# signal is held at ground

    potential. This signal should be held at +5VDC by a pullup resistor internal to the power supply.

     

    Depending on which NUC platform you have, there will be some connectors to the outside world.

    USB, Video, audio, etc.

    If you want to go inside the NUC you might find other signals to use, power good, reset, but I do not know what is inside the NUC.

    I am assuming you will apply some external circuit.

     

    One way to do this is to monitor the Vbus of a USB3.0 Standard-A or B connector.

    When the NUC turns on there will be some delay before Vbus is applied, but is usually just a second or two.

    Vbus will go high 5V, and you want PS_ON to go low, this is an inverting function

    A transistor as shown below functions as a discrete inverter.

    N Channel Mosfet or NPN BiPolar.

    Drain or Collector would connect to PS_ON

    Gate or Base connect to 5Vbus, might put some capacitance after the resistor for noise filtering.

    1. Input low, transistor is off, internal pullup on PS_ON keeps the ATX off
      PS_ON Internal pullup represented by Rp and R2 below.
    2. Input goes high, transistor is ON, output is low pulling down PS_ON and turning the ATX ON.
    3. R1 of NPN can be 1k, must limit the base emitter current.
      Ic=Beta x Ib, Suppose you design for T1 to sink 50mA (PS_ON)
      And suppose the Beta gain is ~50. That means you need 1mA into the T1 base.
      5Vin - 0.7Vbe = 4.3V across R1.
      R=E/I = 4.3/.001 = 4.3K,   something close to this for R1 will work.

    4. The mosfet is a voltage driven device, 
      Vgs threshold is 0.8 to 2.4V, mosfet will turn on somewhere in between.

     

     

    This is one suggestion but one I think would work.

  • Alright :)

    That is really nice post man ;)
    May I use diffrent voltage? Like 3.3V or signal PWRGD from PCI-E line?

  • You can use any signal you can find, either outside or inside, that will come up when the NUC is turned on.
    PWRGD from PCIe would be perfect.
    Usually a power good is active low, goes high when power is good, you need the inverted function of that.
    A NPN or Nfet makes a simple inverter.

    3.3V input would require a mosfet with a Vgs threshold below that.
    3.3V would work fine with a NPN.
  • May I use BC546 for that?

    If I'm going to use it which values of resistors would be best for 5V and 3.3V ?

    BTW if I turn off my NUc then PS-ON will be pulled high again?

  • Happy Friday Mateusz,

    BC546 could be used. It has a lot of gain, but as an inverter switch it will not matter.

    PS_ON# Signal Characteristics
    Min. Max.
    VIL, Input Low Voltage 0.0 V 0.8 V
    IIL, Input Low Current (Vin = 0.4 V) -1.6 mA

    VIH, Input High Voltage (Iin = -200 µA) 2.0 V
    VIH open circuit, Iin = 0 5.25 V

    You need to pull PS_ON below 0.8V.
    And you have to sink 1.6mA, (provided by the ATX pullup on PS_ON)
    VCEsat of the BC546 is 250mV which will be below 0.8V when fully on, saturated.

     

    As I previously discussed:
    5Vin - 0.7Vbe = 4.3V across R1.
    Assume 1mA base current.

    R=E/I = 4.3A/.001A = 4.3K,   something close to this for R1 will work.
    R2 is the pullup on PS_ON inside the ATX supply.

     

    If 3.3Vin, 3.3V -0.7=2.6V
    2.6V across 4.3K = 600uA.
    The BC546 has a minimum gain of 100, so 600uA into the base times 100 = 60mA.

    You only need to sink 1.6mV

     

    With the NUC off, the transistor is off, the ATX internal pullup drives PS_ON high and keeps the ATX off.

  • So R2 from schematics above is not needed?

    I've got pleny of BC546B at home and some resistors :)

    Ok it's working, thank you.

    But I'm curious how about using darlington or IC? Do know any IC that will do the trick?

  • R2 is the pullup resistor inside the ATX supply.
    Regarding the darlington, a darlington stage is designed for current gain in bipolar transistor circuits.
    You only need to sink a few mA so a darlington is not needed.

    Regarding the IC, you can't get much simpler than the basic NPN inverter.
    TI makes supervisor IC's; www.ti.com/.../supervisor-reset-ic-overview.page
    These can be used to monitor a voltage and give a RESET or POWERGOOD signal.
    However:
    1. there are very few that will go LOW when the input threshold is satisfied.
    2. there will likely be reverse current from the reset output back into the supervisor IC, because the ATX pullup is always pulling up.
    this could cause latchup in the IC.
    I looked at supervisors but for these reasons did not recommend them.