Where to start with diagnosing custom cables that dont work

[Voxels-Box]

Hi guys,

I've been working on a set of custom cables for a coworker, and we installed them last week, but there's no signs of life other than the red led on the motherboard. 

Hardware: 

AMD TR 1920X

MSI x399 Carbon AC

Corsair Vengeance RGB 32gb

1080ti sli

EVGA Supernova 1200 P2

Custom MDPC-X cables made by me, using 16-18AWG wire

I used a pinout found online, transferred it to a spreadsheet, then recreated it in 3D (OBJ and blend file here) to help visualize wire placement, and got the lengths as close as I could, but I must have some sort of fundamental misinterpretation of the pinout. I thought I was so meticulous,  yet it didn't work, and I'm lost at where best to start with troubleshooting. 

 

We tested each cable individually, and determined that the 24 pin, 2x 8+6 bridged GPU cables, 6pin supplemental PCIE, and the shorter drive power Sata cable are defective.  We tested every combination of stock and custom cables, and only the CPU powers and multi port sata are working.

 

I have a Thermaltake PSU tester and a cheap multimeter that Ive yet to learn how to use,  but I'm not sure what angle to tackle this, or what I should be trying to isolate. 

 

I currently have both the custom cables, stock cables and the PSU on hand, so I can test as effectively as possible. 

 

Does anyone have any advice for troubleshooting a messed up pinout?

 

 

Thank you in advance!

Edited March 21, 2019 by Voxels-Box
Added links and photos

[Jurrunio]

So all the factory cables of the PSU work?

[Voxels-Box]

3 minutes ago, Jurrunio said:

So all the factory cables of the PSU work?

Affirmative. It's most likely errors that I made in the final pinout process.  The custom 2x 8pin CPU and main sata cables also work when the rest of the cables are stock. 

[Jurrunio]

4 minutes ago, Voxels-Box said:

Affirmative. It's most likely errors that I made in the final pinout process.  The custom 2x 8pin CPU and main sata cables also work when the rest of the cables are stock. 

Then it's a matter of measuring how the wiring on the factory cables are done and whether your custom cables match their sequence then

[0x1e]

@Voxels-Box

 

I'm no electrician but I like the structure of this post and the system that was built that I'm willing to help. Even if I look the fool.

 

Is your 6/8 pin GPU correct on the GPU side?
You have shown G/12v/12v. It should be 12v/G/12v.

 

Are you able to test with the stock GPU cables?

The pin layout seems to be correct, at least on the spreadsheet.
Have you been able to successfully boot the system with all the stock cables?

[Voxels-Box]

On 3/21/2019 at 8:43 PM, 0x1e said:

Are you able to test with the stock GPU cables?

The pin layout seems to be correct, at least on the spreadsheet.
Have you been able to successfully boot the system with all the stock cables?

Yes, I have all the stock cables, and the PC works normally when using them. 

 

Ive used my PSU tester in all the combinations, and only the custom 24pin returns as an error. All GPU custom cables seem to clear the test.

 

Ive used a multi meter to test all the voltages on the custom 24 pin's mobo side connector. It looks like I have several ground or sense cables returning as around 5V,  instead of 0V. Attached is the comparison, with the bad wires marked with red and !!!!.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

@W-L Do you have any insight?  Appreciated in advance!

[W-L]

5 minutes ago, Voxels-Box said:

Yes, I have all the stock cables, and the PC works normally when using them. 

 

Ive used my PSU tester in all the combinations, and only the custom 24pin returns as an error. All GPU custom cables seem to clear the test.

 

Ive used a multi meter to test all the voltages on the custom 24 pin's mobo side connector. It looks like I have several ground or sense cables returning as around 5V,  instead of 0V. Attached is the comparison, with the bad wires marked with red and !!!!.

@W-L Do you have any insight?  Appreciated in advance!

Sounds like a sens cable is not connected correct or returning the proper signal back to the unit. I would suggest to probe the pinout and trace the original modular cables to determine any difference between that and your current cable.

 

This is what your pinout should be for the 24 pin:

[Voxels-Box]

11 minutes ago, W-L said:

Sounds like a sens cable is not connected correct or returning the proper signal back to the unit. I would suggest to probe the pinout and trace the original modular cables to determine any difference between that and your current cable.

 

 

Thanks! Ive now tested against the stock cable. Pins A-5, A-8, A-9 are reading as 0V on stock, but 5V on the stock cables.  I traced the custom cables, and it looks like I has A-5 and A-6's split ends go to the opposite targets on the PSU side. I fixed that, and now 5/6 are reading correctly on the custom cable.  Now all thats left is 8 which is reading at 5V but should be ground.  Should I relocate that pin on the PSU end to go into an open ground? ?(if theres an open one?)

 

[W-L]

4 minutes ago, Voxels-Box said:

Thanks! Ive now tested against the stock cable. Pins A-5, A-8, A-9 are reading as 0V on stock, but 5V on the stock cables.  I traced the custom cables, and it looks like I has A-5 and A-6's split ends go to the opposite targets on the PSU side. I fixed that, and now 5/6 are reading correctly on the custom cable.  Now all thats left is 8 which is reading at 5V but should be ground.  Should I relocate that pin on the PSU end to go into an open ground? ?(if theres an open one?)

 

Pin 8 or PWR OK should have voltage through it, as that's what is used to trigger the PSU to turn on. Power OK is what shorts to ground when one uses the paperclip method to power a PSU without a motherboard. 

 

Try testing with the PSU tester and if it gives the green light you should be good to go on the board. Hopefully the wrong voltages earlier didn't do anything bad to the board. 

[Voxels-Box]

1 minute ago, W-L said:

Pin 8 or PWR OK should have voltage through it, as that's what is used to trigger the PSU to turn on. Power OK is what shorts to ground when one uses the paperclip method to power a PSU without a motherboard. 

 

Try testing with the PSU tester and if it gives the green light you should be good to go on the board. Hopefully the wrong voltages earlier didn't do anything bad to the board. 

Much appreciated. Ive been using pins 16 and 17 with the paperclip trick. Is 8 the sense wire for 16? 

 

The tester failed the cable. Heres what the tester's displays shows.

[W-L]

7 minutes ago, Voxels-Box said:

Much appreciated. Ive been using pins 16 and 17 with the paperclip trick. Is 8 the sense wire for 16? 

 

The tester failed the cable. Heres what the tester's displays shows.

Oh sorry, yes you are right power-ok is just for the PSU to do an internal check, it's pin 16 to ground that is for the paperclip trick. From what I see on that tester it looks like it's not actually thinking that the PSU-On (pin 8 ) and ground are shorted to turn on the PSU. Check and compare the original cables to where pin 8 goes on the PSU end. That connection is critical for the PSU to function and can't be mixed up.

 

Quote

8

PWR_OK

Gray

Power Ok is a status signal generated by the power supply to notify the computer that the DC operating voltages are within the ranges required for proper computer operation (+5 VDC when power is Ok)

 

[Voxels-Box]

A24

1 hour ago, W-L said:

Oh sorry, yes you are right power-ok is just for the PSU to do an internal check, it's pin 16 to ground that is for the paperclip trick. From what I see on that tester it looks like it's not actually thinking that the PSU-On (pin 8 ) and ground are shorted to turn on the PSU. Check and compare the original cables to where pin 8 goes on the PSU end. That connection is critical for the PSU to function and can't be mixed up.

Ahh makes sense.

I used continuity mode to trace 8 on both cables, and and it seems to be in the correct spot on both cables.  I marked down the exact voltages of the mobo side connector on the custom cable, and all thats wonky now is that A-24 is 5V instead of ground.  The strange thing is that A-24 and A-21 are supposed to join and both terminate at C-5. Is that the expected voltage? Should I replace those wires, and move A-24 to a free ground on the PSU side?

[W-L]

2 hours ago, Voxels-Box said:

A24

Ahh makes sense.

I used continuity mode to trace 8 on both cables, and and it seems to be in the correct spot on both cables.  I marked down the exact voltages of the mobo side connector on the custom cable, and all thats wonky now is that A-24 is 5V instead of ground.  The strange thing is that A-24 and A-21 are supposed to join and both terminate at C-5. Is that the expected voltage? Should I replace those wires, and move A-24 to a free ground on the PSU side?

Make sure pin 16 (PSU_ON) is also in the exact same location as the originals. As for A-24 (grd) and A-21 (5v) I'm not sure why those two would be joined together since that would short 5V to ground. Usually sense wires will have two wires go into a single pinout on the 24 pin side and on the PSU side into two separate pinouts, one for the dedicated voltage and the other into a sense pinout. Sounds like it's backwards. 

[Voxels-Box]

6 minutes ago, W-L said:

Make sure pin 16 (PSU_ON) is also in the exact same location as the originals. As for A-24 (grd) and A-21 (5v) I'm not sure why those two would be joined together since that would short 5V to ground. Usually sense wires will have two wires go into a single pinout on the 24 pin side and on the PSU side into two separate pinouts, one for the dedicated voltage and the other into a sense pinout. Sounds like it's backwards. 

Yes, very strange. I also wondered what the purpose was for having a reverse split wire from the beginning, but now that I started making my own pinout with the stock cable and continuity mode, and Im realizing that my original source may have been incorrect. For example, my pinout orginally from evga's forums, says that A-24 goes to C-5 (with A-21), but the stock cable puts it as a single wire to B-17.  Ive checked A-24 to A-19, and they all go to different places than my pinout.  

 

A-16 must be working, since thats how Im turning it on. Since both spots either side of it should be ground, I moved my bridge to short it into turning on from both positions, so Im assuming its the same and isnt part of the issue.

 

Also, A-20 is marked as not in use, but i see that the -5V on the standard may be deprecated now?

20    NC    White    -5 VDC (Optional - Removed in ATX12V v2.01)

So that should be okay?

 

While the original pinout I started from may have been flawed, currently, that 5V reading on A-24 is the only pin that differs from ATX spec. So I shouldnt need to repin all connectors, right? If I can get A-24 into a free ground spot, that should work, I think.

 

[W-L]

2 minutes ago, Voxels-Box said:

-SNIP-

Yeah I would trace back and double check the diagram with the original modular cables to see if there are any odd discrepancies since we know the original cables are what the manufacture specifies. In theory you can switch around all the pinouts on the PSU side as long as they are providing the correct voltage to the 24 pin output, other that the specific ones such as PSU-ON, PWR-OK, 5VUSB, -12V. While I really do not recommend doing that I've tried that before and it does work. 

 

 For -5V that's obsolete at this point it was for the old ISA type expansion slots. 

[Voxels-Box]

7 minutes ago, W-L said:

Yeah I would trace back and double check the diagram with the original modular cables to see if there are any odd discrepancies since we know the original cables are what the manufacture specifies. In theory you can switch around all the pinouts on the PSU side as long as they are providing the correct voltage to the 24 pin output, other that the specific ones such as PSU-ON, PWR-OK, 5VUSB, -12V. While I really do not recommend doing that I've tried that before and it does work. 

 

 For -5V that's obsolete at this point it was for the old ISA type expansion slots. 

Thanks again for all the help! I'll quintuple check the voltages, and trace back those special wires. My findings so far are positive. 

I think that perhaps the OG pinout was either mislabeled, or modified for some reason (maybe for a straighter pinout, which knows)

But except for the 5V/ground issue, all the voltages seem to check out. 

 

I'm going to try changing A24 to a free ground, and try the tester again. 

I'm in a pickle though, because that split wire uses MDPC-X Carbon for the A21, and I'm 100% out of that color. Gonna have to get creative haha 

[W-L]

7 hours ago, Voxels-Box said:

Thanks again for all the help! I'll quintuple check the voltages, and trace back those special wires. My findings so far are positive. 

I think that perhaps the OG pinout was either mislabeled, or modified for some reason (maybe for a straighter pinout, which knows)

But except for the 5V/ground issue, all the voltages seem to check out. 

 

I'm going to try changing A24 to a free ground, and try the tester again. 

I'm in a pickle though, because that split wire uses MDPC-X Carbon for the A21, and I'm 100% out of that color. Gonna have to get creative haha 

Sounds like a plan, once it passes the tester you should be good to test with the motherboard. I guess you can use the slit cable and have the Y end only connected to one pin and hide that extra leg, that or another order of sleeving  

[Voxels-Box]

4 hours ago, W-L said:

Sounds like a plan, once it passes the tester you should be good to test with the motherboard. I guess you can use the slit cable and have the Y end only connected to one pin and hide that extra leg, that or another order of sleeving  

Ive confirmed the correct positions for all the special wires, and all the other voltages check out. I chopped off the A-24 side of the split wire and tidied up the area. The carbon side has a heatshrink bulge but oh well.  Itll be hidden at least.

 

I tested the open pins on the PSU side with a small scrap wire, and the two open spots read as 0V. I made a new wire to go from A-24 to B-15. It cleared the PSU tester!!  

 

I also checked the voltages of all five PCIE cables, and they all check out. 

 

It looks like everything is all green!

 

Side note, I went to Mod-One to place an order, and found out some bad news from their homepage:

"Unfortunately, the rental unit next to our shop experienced a severe explosion on Wednesday March, 20th, which has caused significant damage to our unit. Due to the damage, we are unable to process orders at this time"

Theyre my main source for MDPC-X sleeving materials, but its unclear when theyll open up again.  Do you know any other US resellers?  I cant imagine what a shock that mustve been, I hope they recover OK.