Open Compute WindMill v2 Motherboard

[s2600gl]

Here's a thought:

What if pin D1 is polled on both the MB and the MP ? I initially thought that only the MP was interested in if
anything is seated in the socket (so it could properly delay the start up) Maybe there's a "double check" on the
MB to make sure that the board really is seated.

Missing the return path to GND via D2 would mean that the circuitry attached to D1 on the MB would be pulled high
thus preventing a boot. So could this scenario perhaps work:

P1/P2 => +12.5V
P3/P4 => GND
B4 => +5V
D2 => GND (assuming that D1 indeed gets pulled up on the MB itself)

 

[novegg4]

Here's a thought:

What if pin D1 is polled on both the MB and the MP ? I initially thought that only the MP was interested in if
anything is seated in the socket (so it could properly delay the start up) Maybe there's a "double check" on the
MB to make sure that the board really is seated.

Missing the return path to GND via D2 would mean that the circuitry attached to D1 on the MB would be pulled high
thus preventing a boot. So could this scenario perhaps work:

P1/P2 => +12.5V
P3/P4 => GND
B4 => +5V
D2 => GND (assuming that D1 indeed gets pulled up on the MB itself)

Have tried that - no go.
Did some measurements (disconnected boards)
MP connector - between D1 and D2 is 10KOhm, MB connector - between D1 and D2 is 1 Ohm
however it might be meaningless to measure pins

 

[s2600gl]

Not necessarily meaningless. The low resistance on the MB tells that it's probably not much more than a wire trace between
D1 and D2.

What's puzzling is with such a low resistance between D1 and D2 on the MB any voltage present on the D1 will be pulled to ground.
So why doesn't grounding D2 without the MP work while grounding D2 with the MP works ???

 

[novegg4]

Not necessarily meaningless. The low resistance on the MB tells that it's probably not much more than a wire trace between
D1 and D2.

What's puzzling is with such a low resistance between D1 and D2 on the MB any voltage present on the D1 will be pulled to ground.
So why doesn't grounding D2 without the MP work while grounding D2 with the MP works ???

S U C C E S S

As soon as I took MP off the loop and connected ATX PSU directly, it went through... at least booted into Win7 64
Summary:

P1/P2 => +12.5V
P3/P4 => Ground
B4 => +5V



All credit goes to s2600gl - knowledge is power!

 

[macrules34]

That's great to hear!!! We should put this all in a document (guide) to share with everyone.

 

[novegg4]

However, using non-standardized power most likely eliminates some quite necessary features - for instance Wake On LAN.

 

[s2600gl]

SWEET !!

Give yourself some credit, though, for meticulously testing all the different combinations.

Now if only Digikey/Mouser could be persuaded to sell us smaller quantities of the FCI connector

 

[gigatexal]

can someone post pics of an ATX 12v 2.1 connector connecting to one of these mobos with a diagram of what wires go where?

 

[macrules34]

I called Digikey and inquired if it would be possible to get less than their minimum purchase amount and I was told that they do not have them in stock and get them directly from the manufacture so you have to buy a quantity of 528.

I also called Amphenol FCI (the manufacture) and they said that you can request a sample to get the low quantity. Link

 

[novegg4]

can someone post pics of an ATX 12v 2.1 connector connecting to one of these mobos with a diagram of what wires go where?

I have wired power through PCI-E connectors on my Corsair HX750 while having the main motherboard cable just shortened because I plan to use this HP PSU eventually:

HP Proliant DL380 G6 G7 ML370 G6 750W Power Supply 511778-001 506821-001

Still have to figure out how to get +5V off that one which output is +12V only

to s2600gl

Please suggest whether this one should be enough for 12V=>5V conversion to feed the signal line of B4 pin:

DC To DC Converter Step-down Voltage LED Power Module 3A 12V - 5V 3.3V

or this one:

23mm 12V to 5V 3A 15W Car Charger DC Voltage Regulator Converter

 

[novegg4]

I called Digikey and inquired if it would be possible to get less than their minimum purchase amount and I was told that they do have them in stock and get them directly from the manufacture so you have to buy a quantity of 528.

I also called Amphenol FCI (the manufacture) and they said that you can request a sample to get the low quantity. Link

Thank you, requested some samples through the link provided.

 

[s2600gl]

@novegg4 : Both of those VRM:s are more than adequate for the power good signal (as it's logic signaling
we're probably talking sub 50mA currents).

But if you're planning on using the HP PSU there's a very very good chance that it's has a +5v output too
(even if it doesn't get mentioned in the specs).

 

[novegg4]

@novegg4 : Both of those VRM:s are more than adequate for the power good signal (as it's logic signaling
we're probably talking sub 50mA currents).

But if you're planning on using the HP PSU there's a very very good chance that it's has a +5v output too
(even if it doesn't get mentioned in the specs).

Good to know, thank you.
I'll check it out as soon as ones arrive in the mail.

 

[s2600gl]

With the adaptations to make these boards run off of 120v by DonJon in the "Potential Deal: 2x Dual 2011 nodes @ $199, Quanta Openrack" thread there must be a way to make these fit in or design a standard 19" rack-mount case.

Ok, so I have some disturbingly crazy ideas bouncing around in my head regarding this. Any chance @macrules34 , @novegg4 or anybody else lucky enough to own these could measure the total width
of 2 nodes side-by-side ?

 

[novegg4]

Ok, so I have some disturbingly crazy ideas bouncing around in my head regarding this. Any chance @macrules34 , @novegg4 or anybody else lucky enough to own these could measure the total width
of 2 nodes side-by-side ?

As for me, I have only one board mounted on the metal tray - see this ebay listing for details:

QUANTA DA0F03MBCD0 REV D DUAL FCLGA2011 SYSTEMBOARD FOR INTEL XEON E5 SERIES

Please let me know what particular dimensions you are looking for.

By the way, I've just got my HP power supplies - would be great if you give me a clue where to check ones regarding +5V output

 

[s2600gl]

Your best bet is the narrow "fingers" on the left (top) and right (bottom). If there's a +5V it will be on one/some of them.

Btw, I found info on how to power up those HP PSU:s when they're not in a HP server over at RC Groups.com:
RC Groups - View Single Post - A simple high quality 12Volt 100Amp Power Supply- Part1

And if you don't mind whipping out the tape measure cold you check the width and length of the metal tray the MB is mounted on.

 

[novegg4]

Your best bet is the narrow "fingers" on the left (top) and right (bottom). If there's a +5V it will be on one/some of them.

Btw, I found info on how to power up those HP PSU:s when they're not in a HP server over at RC Groups.com:
RC Groups - View Single Post - A simple high quality 12Volt 100Amp Power Supply- Part1

And if you don't mind whipping out the tape measure cold you check the width and length of the metal tray the MB is mounted on.

The metal tray is:

174.5 mm x 522.5 mm

Regarding +5V output: you were right, there are several pads in the left/right area - #32 and couple below.

 

[s2600gl]

Thanks for the measurements. Now I got something to work with.

And good to know about the +5V output, one of those might actually be the PG signal.

 

[novegg4]

Thanks for the measurements. Now I got something to work with.

And good to know about the +5V output, one of those might actually be the PG signal.

You are welcome.
Are you going to build a modded true 19" case for those?

 

[s2600gl]

Something along those lines yes, but with a twist. At the moment just looking into the plausibility of a
few ideas.