PSU Tech Look: Cable Length and Voltage Regulation

Thanks so much for this Paul. A lot of really dang good information here. Good shiznit to know.
 
Good article. I'd agree that the quality of the connectors plays a bigger role than wire length - there are some shoddy/loose connectors out there.
 
I mentioned this in the comments of the last PSU review, but it is probably more appropriate here.


Never just assume the extension cables you get on Amazon or at Microcenter are safe.

You might have them burst into flames like my 8 pin EPS extension cable did...

1571350436784.png
 
I mentioned this in the comments of the last PSU review, but it is probably more appropriate here.


Never just assume the extension cables you get on Amazon or at Microcenter are safe.

You might have them burst into flames like my 8 pin EPS extension cable did...

View attachment 150
Your previous post was fresh in my mind when I was reading this article from Paul. **** was crazy. Almost fell outta my chair when I saw dat pic, dang near had a heart attack.
 
Last edited:
I kind of knew the outcome before even reading the article. At the lengths and gauges used in a PC, at least for quality cables, the differences are going to be negligible.

If you want nice sleeved extensions just buy quality and know that the weak point will be the connection and not the wire itself.
 
  • Like
Reactions: _k_
I kind of knew the outcome before even reading the article. At the lengths and gauges used in a PC, at least for quality cables, the differences are going to be negligible.

If you want nice sleeved extensions just buy quality and know that the weak point will be the connection and not the wire itself.

Yeah, in general I'd just avoid older extensions and splitters. Not because cables don't age well, but because computers used to draw less power and a lot of the old ones just weren't made to withstand modern requirements.

They probably didn't meet spec when they were new, but back then no one cared because they worked.

This should be something in the minds of all of us who have been doing this for a long time, as if you are anything like me, you have spare parts (and screws, and wires) bins in the basement that span decades.
 
Great article, more investigations like this pls :)

I figured it wouldnt be a large difference but am surprised the efficiency changed as much as that.
It would be interesting to see how a silver plated cable performs for both efficiency and stability.
 
As you've shown, cable length is not likely to be a big difference; the biggest difference is what pin is crimped on the end of the cables.

These guys do a good job of listing the connectors, With Pin Numbers.


Most of you know that Molex made the connectors for PC's, and still own the basic designs, even though the patents expired long ago.

Most of the 12V pins are rated at 6 or 7A; the 4 pins in the 8 pin PCIe connector comes up to 336W, a comfortable margin for the stock design.

This says in smaller housings, the brass terminals can go 9A.

However; few of us are at stock levels; I've seen my CPU draw 250W, and that's not counting the draw from the RAM, which also runs off that plug.

These ratings are from the High Quality, Molex brand connector pins; what's actually in that noname cable extension you bought off Amazon?

That likely has a bigger bearing on performance than the cable length; pin retention pressure, gauge of metal, is it beryllium copper, brass, or steel?

I've seen steel pins in those cheap connectors; if it sticks to a magnet, throw it away, lol.


Thanks for the article; it makes me want to measure some actual drops in my systems, especially the ones with splitters on drives.


It would be interesting to get a bunch of the cheapest connectors, and watch them burn. :)
 
I mentioned this in the comments of the last PSU review, but it is probably more appropriate here.


Never just assume the extension cables you get on Amazon or at Microcenter are safe.

You might have them burst into flames like my 8 pin EPS extension cable did...

View attachment 150

I know, right?

I actually had a person on LTT forums say this when I told them to compare apples to apples when comparing a $5 eBay cable kit versus something from BitFenix, CableMod, etc.: "dont understand your apples to apples point. Wires of copper and plastic are still wires of plastic whether sold on eBay or amazon."

I think the article would have been better if Corsair cables weren't used. The Corsair cables are thicker gauge, have better pins, have flammability treatment (melamine) and plastic connectors with lower melting point (PA6 vs. PA12).

Let's flip this on it's head and replace the Corsair cables with something bought on DealExtreme! :D
 
Approaching this with the eyes of someone who's gone through EE classes, that's a great review/comparison.

No difference at these voltage and amperage levels, as expected.
 
One thing I've always found interesting.

You've got all these little pins on a CPU package, small traces on a motherboard, and small gage wire inside a computer. That stuff is pulling tens, sometimes upwards of hundreds, of amps.

A molex connector, if I'm not mistaken, is only rated for about 6A per pin. So I guess that explains why we have all these aux connectors to supply additional power to a motherboard, but the traces on a motherboard still aren't all that big.

Looking at a high wattage PSUs... I'll just pick one to have some specific numbers: https://seasonic.com/prime-ultra-titanium
83A on the 12V rail.

If I wanted to run 83A in a commercial electric setup, I would need a lot of copper. Probably a #4, which is about 1/4" diameter of straight copper. An 18GA conductor like is commonly used in wiring, can carry only carry 7A per NFPA rules (NFPA mostly cares about amps, not volts, when sizing cables. Volts determines insulation class).

I know consumer electronics like this don't follow NFPA - that's for mainline power wiring in homes and buildings -- consumer electronics will follow UL and other standards. I also know they get split into a lot of parallel connections, but comparing a #4 wire to the pins on a CPU socket is ... crazy.
 
The parts that are drawing the big currents use tons of pins to get that power thru.

The individual pins on a CPU are running a few amps.
100A at 1V is 100W, so at 2A/pin it needs 50 pins; and 12V at 100W is 8.4A on 6A pins, so it needs 2 pins.

I've ran 250W to a CPU designed for 150W, so the pins have some headroom, lol. (OC mistake)
 
Become a Patron!
Back
Top