5V/12V Molex adapter board, PCIe 6-pin input, 20A capable (GPU rigs and such)

[sidehack]

I have designed and successfully prototyped a small board which takes in 12V from 6-pin PCIe and outputs both 12V and 5V on three 4-pin jacks.



The board can handle outputting 20A on both 12V and 5V simultaneously. This is useful for powering GPU risers from a 12V-only server supply. Depending on your particular needs (most specifically the 12V current requirements), one board could power half a dozen risers. If your riser gets power from Molex only, 3 risers per board is more likely.

All materials have been ordered for the first batch of 100 boards, and I expect to start assembly in about two weeks. I'm taking in orders now so folks can lock in what they want and I can get a head start on cabling so full kit orders can ship immediately once boards are completed (a relatively quick process).



Boards are $15 apiece. 18" single-ended Molex cables are $1.75 apiece, or $2.50 with the sheathing.
I can sell a kit of one board with three cables for $20, or three sheathed cables for $22.30

I can also provide PCIe 6-pin cables and 6+2 cables of varying lengths. If you want something, shoot me a message and I'll run up a bid.

Shipping is going to depend on order size, but most orders should ship for $10 or less.

The new board's stock configuration will differ from the pictured board in two ways - the output jacks will be straight (like the 6-pins) instead of right-angle, and it'll have a couple small LEDs on there letting you know the 12V and 5V are powered and working.

[MarkAz]

Put me down for 10 more, along with the corresponding braided cables of course!  They've really simplified my builds and made everything much cleaner.

[sidehack]

Need any 6-pin cables for the 12V side, or do you have that taken care of?

[philipma1957]

I would like two boards.  Six pcie to molex. Cables.


 

[HerbPean]

I would like two boards.  Six pcie to molex. Cables.


 

Same here

Thanks

[MarkAz]

Need any 6-pin cables for the 12V side, or do you have that taken care of?

Yeah, I'll want 28x of the 13" 6-pin PCIe to 8-pin PCIe braided... And I'll need 20x of the 6-pin PCIe to dual 6-pin PCIe braided like before - were they 6" or 12"?  Thanks!

[sidehack]

Everyone who's asked about stuff I'll PM with prices today.

I just got word the PCBs should be shipping out October 10th, so factoring in transit time and assembly I should start shipping orders week of the 17th.

[HerbPean]

Everyone who's asked about stuff I'll PM with prices today.

I just got word the PCBs should be shipping out October 10th, so factoring in transit time and assembly I should start shipping orders week of the 17th.

Keep calm and sidehack on !

Payment sent.

Thanks

[MarkAz]

Got the quote and payment sent!

[Finksy]

Awesome product!  Does anyone have input on the merits of using risers requiring 12V & 5V input as opposed to the ribbon-cable-type risers that are powered strictly via 12V PCIe connectors? Or in other words, how much 5V current would a typical GPU require, if anyone could fathom a guess?

Edit:  Here is the type I'm referring to.



The 5V circuit from Molex connector does not power any circuits on the PCB's, and the 12V and ground circuits are shared between PCIe and molex, which leaves me to assume the GPU gets 5V through the ribbon cable.

[MarkAz]

Awesome product!  Does anyone have input on the merits of using risers requiring 12V & 5V input as opposed to the ribbon-cable-type risers that are powered strictly via 12V PCIe connectors? Or in other words, how much 5V current would a typical GPU require, if anyone could fathom a guess?

Edit:  Here is the type I'm referring to.
The 5V circuit from Molex connector does not power any circuits on the PCB's, and the 12V and ground circuits are shared between PCIe and molex, which leaves me to assume the GPU gets 5V through the ribbon cable.

When I was doing my testing of power draws, I would see an average draw of about 1a on the 5v side per card (this is a 6 card rig, those numbers are just for the molex side of the riser's power draw):



Keep in mind this was the average, it could peak up to 2x that, and this was with R9 390's... With the power demand of the RX480's I was pretty concerned about managing the power.  My big concern would be the sustained amount of power being pulled over a ribbon cable - from what I've seen, each wire on the ribbon is rated for between 100ma to peak load of 500ma, so unless they're using multiple wires to carry the 5v I suspect you'll run into issues.

Sidehack can probably provide more input, but for my money I'd rather have as much headroom on the power side of things, and some nice 16/18AWG wires where basically no load is going to pose an issue.

[alh]

With regard +5V for the GPU, I am about 99% certain that it does NOT come through the PCIe connector. My recollection is that the PCIe connector has NO +5v power pin on it. This is why you will regularly see PCIe USB cards which have a supplemental power connector (e.g. a 4-pin Molex) on them. That's to supply the required +5v that's needed on the USB connector. When you look at the pinout for PCIe socket, you'll see +12V and +3.3V, but no +5V.

Yes, I was completely flabbergasted when I heard that there is no +5V on the PCIe socket specification. I thought the guy telling me that pulling my leg.

[sidehack]

From what I understand, some risers take in 12V only and have an internal 3.3V regulator that steps down the 12V for the socket. These do not need my board. Risers which take in 12V from a Molex or SATA and step the 5V down to 3.3V, like MarkAz uses in his builds, could benefit from it. There probably is no actual 5V going into your GPU, just 12V and 3.3V

I don't know enough about ribbon risers to know what voltages (and at what allowed currents) they pull from the socket.

[adaseb]

The +3.3V usually never pulls more than 1Amp. Its 1.5A+ for you because there are huge losses in those voltage regulators on the USB Risers.

Each PCIe slot ribbon has 4 wires for the +3.3V.

[sidehack]

Not sure I follow that logic. For a switching regulator, power out equals power in times efficiency (usually at least 80%) and for a linear regulator current out equals current in. There should never be a condition where a voltage-reducing regulator intakes more current than it outputs (except for edge conditions like where output current is dwarfed by regulator supply current requirements or where the vout/vin ratio on a switcher is higher than conversion efficiency).

[methylminer]

I'll take 5 boards with the braided cables. Also a HP common slot breakout board if I can add that to this to ship all at once.

Thanks

[philipma1957]

I paid for my 2 sets with sheathed  cables


https://blockchain.info/tx/aa4fa96973e6fc688cd69ee66e656080dfcdc1229c306099fb89c641a1f0d8a4      $40
https://blockchain.info/tx/07e1b2ad9672679efb9836bb18122b1cfc8b8545a927623347678df3786e37cb   $11.60

total------------------------------------------------------------------------------------------------------------------$51.60


looking forward to them.

[sidehack]

PCBs and all components have arrived. I spent today verifying everything and setting up the pick-and-place for bulk manufacture. I ran one off to test calibrations, completed assembly on it and it's running right now with about 10A of output current on 5V with no trouble.

Tomorrow I'll do all the surface-mount for the batch and start on through-hole.

I got started cutting wires to run out cables last week but ran out of wire; my reorder has arrived. So tomorrow I'll also be finishing up cutting wire and start cable manufacture. I've got some better heatshrink than I had available for the test cables MarkAz got, which means the production batch should come out better. As of right now there doesn't look to be any reason I can't get initial orders shipped out Friday.

[HerbPean]

PCBs and all components have arrived. I spent today verifying everything and setting up the pick-and-place for bulk manufacture. I ran one off to test calibrations, completed assembly on it and it's running right now with about 10A of output current on 5V with no trouble.

Tomorrow I'll do all the surface-mount for the batch and start on through-hole.

I got started cutting wires to run out cables last week but ran out of wire; my reorder has arrived. So tomorrow I'll also be finishing up cutting wire and start cable manufacture. I've got some better heatshrink than I had available for the test cables MarkAz got, which means the production batch should come out better. As of right now there doesn't look to be any reason I can't get initial orders shipped out Friday.

I can wait to get my hands on them !

Thanks 

[philipma1957]

PCBs and all components have arrived. I spent today verifying everything and setting up the pick-and-place for bulk manufacture. I ran one off to test calibrations, completed assembly on it and it's running right now with about 10A of output current on 5V with no trouble.

Tomorrow I'll do all the surface-mount for the batch and start on through-hole.

I got started cutting wires to run out cables last week but ran out of wire; my reorder has arrived. So tomorrow I'll also be finishing up cutting wire and start cable manufacture. I've got some better heatshrink than I had available for the test cables MarkAz got, which means the production batch should come out better. As of right now there doesn't look to be any reason I can't get initial orders shipped out Friday.

good news thanks