2016 New 8″ inch ATX 4 Pin Male to 4Pin Female PC CPU Power Supply Extension Cable Cord Connector Adapter 1PC Hot 20cm (WHL #14)
Religious holidays, you gotta love them. Short (pun intended) review today: A plain old power extension cable for your PC, one of the 12V supply cables to the motherboard. I don’t really know why I ordered it back in January, but I guess it was in preparation for upgrading one of the 19″ cases as a gaming rig. Which hasn’t been built yet, because the bloody display manufacturers only ship crappy 3440×1440 displays with ridiculous FreeSync ranges..or even without. In 2016! It’s not rocket science!
So, what do we have? A
High quality ATX power cable for motherboard
for the giant price tag of 40 Eurocents a piece. Label on the plastic bag says “SH-3C-1112”. I ordered two at once to at least justify the yellow envelope. Boy, that hurt my bank account. But I recall that there weren’t more expensive cables available that actually had some real tech specs listed. Same is true today, as I (spoiler!) just ordered a replacement one. I would really spend more money if anyone listed wire details. But as nobody does…buy the cheapest, file a complaint if is real Wan Hung Lo qualitty, get a refund, repeat the cycle until something usable arrives at your mailbox. F*ck th environment.
Doesn’t look too shabby, right? The connectors are actually very good quality. But if you have a closer look at the top of the image, the cable has an interesting kink. Why is that? Well, after measuring wire resistance I was curious if the wire really is THAT thin. It doesn’t feel very sturdy, but that’s always very dependent on the isolation material, the wire material at given diameter, and the type of stranding used. And at around 150 mΩ per 20 centimeters (end to end including 2x contact resistance) I was already sure the 18AWG printed on the cable was maybe 18CAWG (Chinese-American Wire Gauge), but certainly not to spec.
So I just hooked it up to my ATX power supply that takes over when my lab power supply hits its limits. Copper 18AWG should carry 16 Amps at 90°C according to Wikipedia, and as I suspected aluminium wiring, I derated to 10A. Which is the same as copper would do at 60°C. 10A at 3.3V, the lowest voltage on ATX power supplies, is reached at 0.33Ω load. Remember the 150mΩ figure of the single 20cm wire resistance? Very handy. So I soldered two contacts from a salvaged PCB connector of the same type together and had a go. 0.3Ω + two more contacts would certainly allow around 10A. The power supply can do more, but I will come back to that later.
Hooked it all up, added some more balancing load to the power supply (old unit that cannot take large cross-loads), and zeroed out the Uni-T clamp meter. Let’s go.
That’s the result of 6-7 Amps over about 20 seconds each (current rapidly decreasing). The cables warm up instantly, get very soft, and the one with the worst contact resistance released a little magic smoke. 6-7A also mean something like 0.5Ω total resistance. I thought the additional resistance was from the rest of the wiring (including the short circuit pins shown in the second picture), but I was wrong. I’ll discuss that in the next section.
150mΩ per 20cm sounds like 35AWG (ballpark figure). I don’t have a micrometer gauge yet, so I only can use a standard digital caliper. It tells me 0.33mm wire diameter of the slightly twisted entire strand or 0.09mm of the individual wire. Especially the last figure is repeatable, but surely down in the noise. However, calculating the area yields 0.0064mm² (have fun rounding this to a justifiable number of significant digits ). For 12 strands, that is 0.077mm² in total.
The twisted wire has 0.086mm² area but is not in tight packaging, so subtracting the usual 13% yields 0.075mm². Not contradicting each other, I think one can safely round to 28AWG which is 0.081mm². Google the absurd square-inch values yourself if you really need to I assume the real cable spec is somewhat around 30AWG. I wouldn’t use that for data lines if not absolutely necessary, but building power cables with that is just Wan Hung Lo gold. Excellent quality and certainly up for the task (if below one Amp or so..)
I wonder: Maybe I am wrong reading “18AWG YOU CHUANG” on that cable, interpreting it as “18 AWG wire diameter, produced by YOU CHUANG”? Maybe it says “I BAWG YOU CHUANG”, which is Wan Hung Lo factory slang for “I just sold you idiot the crappiest cable I could find on the Shenzhen market”? The Urban Dictionary doesn’t really help here, as who would contribute to such a project in their free time when working 20 hours a day, making excellent power cables for the entire world…
Anyway, I do have a splitter cable with the same connector type on one end (but for 6-pin graphics cards, not mainboards) for comparison. The other end is one of the old-style 4-Pin IDE plugs, so contact resistance might be a tad higher than on these high-current Molex ones. Cable length is between 20 and 25 centimeters. When shorting two wires just like in the other cable, I measured a total resistance that is basically the same as shorted measuring leads. My Half Ohm clearly explores the limits of 2-wire measurements…but this also means that total resistance of a very similar cable including the shorting jumper in between can actually be in the order of 10mΩ for ~45cm of cable.
Connecting this in the exact same setup as the previous cable yields a nice current of 27A! Oops again I even reckon this is limited by something else, as the power supply is rated at 32A at this rail, but certainly this isn’t to be delivered by a single cable. Also, the banana jacks from which it is drawn are quite cheap and not 32A-rated. The Hirschmann leads in between aren’t either (I thought so!), but are specced at 60V/16A and 22mΩ/1m. 27A @ 3.3V means 120mΩ total load, and that’s not much to include a lot of contact resistance of the overall setup and four meters of wire. As the cable under test and all others after the banana jack didn’t get warm at all, something else must have dissipated the ~90 Watts. Aaand I hope I didn’t damage anything in hard to reach places. You always kill the weakest link in the chain, and this time, it wasn’t the DUT
Which leads me to a conclusion for this article: I BAWG YOU CHUANG. Or: If you really intend to use such a cable and do not have time for multiple shipping attempts, buy at a local store where you can throw it in the face of your salesman if it is just Wan Hung Lo quality for a hefty price. Always test what you’ve got, and in this case, it’s not done by just measuring continuity, but actual resistance and/or ampacity tests are required. Those Molex connectors are rated to handle 6 to 10 Ampere per Pin, and you do not want to use some 30AWG cable to do the job. You really don’t.