Power Supply PICO - ATX PSU Combo

[Schmit01]

Hey All

First time post.

I'm building a slim portable work PC with a nice CPU cooler design that I can carry in a backpack which I will post when complete.

Running an i7 4970K and GTX 760 at the moment and want to keep an external PSU at the office and home.
Although I love the HDPLEX 300W PSU, at my exchange rate and the cost of the Dell 330w power brick locally x2 I cant swing it passed my boss.

An Idea I have is running a 150/160W PICO PSU in parallel/series with a standard ATX PSU only running the 12v line.

So run the 12v line off the ATX PSU with #6 AWG wire to reduce voltage drop and using Anderson type 50Amp connectors (overkill I know). In parallel run the +5VSB and Power ON. So a 4 pin connector to the case.

I then connect the +5VSB and Power ON to the PICO corresponding pins. I will then run the CPU and the GPU directly from the ATX PSU while the lower voltages and the motherboard 24 Pin connector is handled by the PICO.

So what should happen
1. Motherboard Turns on the ATX PSU with Power ON pin.
2. 12V is sent to the PICO
3. With the Power ON already enabled PICO starts up.
4. Once 12v is stable PICO sends Power OK signal to Motherboard.
5. PC starts up.

Anybody see any reason why this shouldn't work?

 

[EdZ]

Probably OK, but having 12V turn up at the motherboard (via the ATX connector from the Pico and via the EPS connector from the ATX PSU) and GPU (via PCIe card edge from the motherboard and PEG from the ATX supply) from two different sources at two different times may cause Weirdness, even with the PS_ON chain. If the two differ in voltage slightly, there is also the potential for current to flow 'back' to one device from the other. This risk should be reduced (compared to the "two AC-DC bricks one DC-DC converter" scenario) because anything drawing a lot of power over 12V already has its own DC-DC stepdown solution with load balancing.
If you're already modding the Pico to redirect PS_ON and +5VSB, may as well stick some Schottky diodes in there too, same with the ATX supply. Could also be worth summing the PWR_OK from both the ATX PSU and the Pico to prevent one 'outrunning' the other.

::EDIT:: It will very likely work just fine the way you have it laid out, I'm just thinking in a belt-and-braces by-the-book way.

 

[Schmit01]

I get what you saying and appreciate the feedback.

The way I see it, the PICO doesn't regulate the 12v line, so there should be no difference in voltages on the 12v at each connector.
Yes I agree the power would turn up at the EPS and GPU connector first before the PICO turned on but should be ms difference.
To be safe I could run the EPS off the PICO, and the GPU off the ATX, but will see whats happens when testing.

I wouldn't connect the PICO PWR OK to the ATX as that is a voltage level signal. So if the one is High it would override the other.
However having the PICO run off the ATX PSU, if it detects a voltage drop on the 12v it will shut down the PC, though I'm not sure if the ATX would also shutdown as that it is supplying the +5VSB.

 

[Thehack]

I get what you saying and appreciate the feedback.

The way I see it, the PICO doesn't regulate the 12v line, so there should be no difference in voltages on the 12v at each connector.
Yes I agree the power would turn up at the EPS and GPU connector first before the PICO turned on but should be ms difference.
To be safe I could run the EPS off the PICO, and the GPU off the ATX, but will see whats happens when testing.

I wouldn't connect the PICO PWR OK to the ATX as that is a voltage level signal. So if the one is High it would override the other.
However having the PICO run off the ATX PSU, if it detects a voltage drop on the 12v it will shut down the PC, though I'm not sure if the ATX would also shutdown as that it is supplying the +5VSB.

+5vsb is on as soon as the power supply is connected and PSU breaker is on. It is usually a separate rail that controls the logic of the PSU.

The power on signal is a pull to low for the PSU. Issue is it usually pulls a 5v (from the standby) to a latching low by the motherboard. If your PSU is supplying 12V to a plugin DC board (PDCB) then it would need the 12V first to operate the board logic. It requires the board logic to generate a high to low signal so ...what do?

... In the end just short the PS_on signal on the ATX PSU. I'm not sure if I'm even correct on that but that is what I theorize.

 

[Schmit01]

I think you miss understood the wiring configuration. The +5vsb of the atx connects directly to the +5vsb of the pico powering the MB. It doesnt matter if there is no 12v supplied all the time to the PICO but i would have to check if the ground is isolated on the PICO.

The reason for not just shorting the PWR ON with a simple switch is 1. I want to hide the psu under the desk so dont want to climb under everytime 2. There is a good chance I would forget and leave the psu running with no load.

 

[EdZ]

The way I see it, the PICO doesn't regulate the 12v line, so there should be no difference in voltages on the 12v at each connector.

Whoops, you're correct there! Got mixed up with the 19V DC-DCs for a bit.

I wouldn't connect the PICO PWR OK to the ATX as that is a voltage level signal. So if the one is High it would override the other.

A discrete AND gate IC would work, rather than just tying the lines together.

However having the PICO run off the ATX PSU, if it detects a voltage drop on the 12v it will shut down the PC, though I'm not sure if the ATX would also shutdown as that it is supplying the +5VSB.

Having the PICO drop out the but ATX continue running could be an odd situation. It should drop the PWR_OK signal so the motherboard knows to shut things down though.

 

[Thehack]

I think you miss understood the wiring configuration. The +5vsb of the atx connects directly to the +5vsb of the pico powering the MB. It doesnt matter if there is no 12v supplied all the time to the PICO but i would have to check if the ground is isolated on the PICO.

The reason for not just shorting the PWR ON with a simple switch is 1. I want to hide the psu under the desk so dont want to climb under everytime 2. There is a good chance I would forget and leave the psu running with no load.

1. That may not be practical, as the voltage would mix with the PDCB voltage regulator once you turn it on, and it may cause some weird things to happen. It may actually work just fine. Current move based on potential. If the ATX is producing 5.01V and the PDCB regulates at 5.10V, then the current will flow towards (or away depending on which theory you use) from the ATX PSU. This may have an adverse effect. The voltages are produced from two separate circuits. You'd have to find a way to complete disconnect the +5VSB from the PDCB.

2. It's fine to leave your PSU on with no load.

3. An idea to just run the PS_on to a separate switch that you pull to low.

4. Make sure your connector can handle 16A at least if you're combining the 12V lines into one fat 6AWG cable.

 

[Schmit01]

Well this is what I intend on using http://www.andersonpower.com/_global-assets/downloads/pdf/ds-sb50.pdf

It is rated up to 120A and 50A as a hot plug at 41x35x15mm. With a 0.200 Milliohms resistance and a worst case load of around 350W/30A
I should see only a 0.06 Voltage drop at the connector by my calculations but will check when testing. Yes its is big but a lot better that a full size ATX PSU.

Yes I can just Pull the PS On with a switch, but then it wont work with I'm also trying to achieve of feeding the EPS & GPU from the ATX PSU. I don't want the chance of the 12V feeding the CPU and GPU for long periods while the motherboard is off. It comes down to being a safety feature for me with the benefit of performance/cost.

I will set up a test setup and let you guys know how it performs before proceeding with the case.

 

[Schmit01]

So I received my PICO PSU and started testing.

I found I had to isolate the +5VSB and the PWR_ON to the motherboard from the PICO. It functioned while both were bridged and would start and shutdown correctly however when the +5VSB was bridged the 12V rail would provide 5V when the PSU should be off and when the PWR_ON was bridged the 12V rail would provide 0.5V when the PSU should be off.

So this left me with the issue of the PICO PWR_OK pin. So I figured why not just have this on all the time with a bridge to ground as the PICO can only come on when the 12V from the ATX it activated from the motherboard and so far so good.

If anybody has a better understanding why I'm getting the lower voltages on the 12V rail please pitch in.

Otherwise I happy with the current configuration and cant find other issues, so have proceeded with my component orders and will post my build log soon.