How to braid cables

My handiwork

I have also done it with several other items (most often 3-wire fans), but here are some specific examples in (roughly) chronological order:

FDD adapter cable from Seasonic G-360

As previously shown in my 2017 forum thread on cable bundling methods.
At the time, I didn’t strictly use the proper method for a square braid; since then I’ve redone it properly, and also plaited the other half (top in the picture) for a comparison.

Old server wiring harness

This is from a 2012 Intel server, among a few other items I salvaged (including the redundant 750W Delta PSUs).
The wire here is of noticeably high quality and more-flexible than average (manufactured, or at least branded and assembled, by BizLink Technology). Unfortunately, the original black tape left some dark stains where it was wrapped around.

On the group harness (powering most items except the mainboard), I square-braided both of the 4-pin drive power cables (plus the tail from the second SATA to another 4-pin connector). You can see that I did the two in different pitches (the tighter one first), not that this makes much difference overall to flexibility. One has 11 locks in 48cm (about 44mm pitch), the other has 5 locks in 39cm (about 78mm pitch). I since did them over with the starting exception, also going down to 6 locks in the longer cable so they match well.

The PMBus cable got a simple plait. I have not done the main ATX cable (not shown) yet, but if I get a terminal extractor, I’ll plait it too (and probably square-braid the EPS12V cables on the group harness).

Both the group harness and main ATX cable use a mix of 18AWG and 20AWG wires; this is because some of the source pins are shared, so they get 20AWG wires to fit (Molex Mini-Fit Jr. terminals being designed for 16AWG at the largest; two 18AWG conductors equal 15AWG).
The main ATX cable (which uses an 18-pin source connector for the positive rails, and a small 6-pin with one unused for the control/sense signals and −12V) shares 3 source pins among the 5 ATX pins for +5V, leaving it with 4×20AWG and 1×18AWG; the others are all individual, so 18AWG for the rest with 24AWG for the low-current lines.
The two CPU connectors draw from six +12V and seven ground pins; with pins 6 and 7 sharing a source pin per connector, and pin 2 of both connectors sharing a source pin between them.
The 4-pin drive cables are grounded with one source pin each (so +12V and +5V there are 18AWG, the grounds 20AWG); only the SATA cable gets a full pin complement and all 18AWG.

Although not shown, there was a separate 4-pin ATX12V cable (same at both ends) with green +12V wires; the Power Board Port Module includes connectors for 5 of those, labeled “ADPT PWR”.


An SFX PSU (in the original narrower, longer style) from 2001, which I chose for this effort because every wire is soldered to the PCB individually (unlike in many other units where groups of wires are crimped into ferrules together, then soldered into a few larger holes); so I could free them up at the supply end, and braid the main ATX cable (20-pin at the time, and also 20 wires given that SFX never required the −5V) without restrictions.
This is my first (and probably only…) 20-core square braid, and it was hard work to make. For so many cores, I recommend just plaiting; the asymmetry should hardly be noticeable with 20 or 24 wires (versus 4).

The ATX12V connector here was very much an afterthought (even more so than on most units of the time), being daisy-chained off a drive/peripheral cable.
(Not that this really mattered with only 4.5A on the +12V…) Since the two +12V wires were also crimped together at the drive/peripheral connector, I didn’t get its braid section quite right either (although it’s still theoretically possible if the +12V cores are twisted and grounds threaded through just right).

Macron MPT-400

(I also recapped this unit years ago with Rubycon ZLH, Panasonic FR and TS-ED primaries; operating Casing Macron PSUs with their original capacitors is a bad idea)

I plaited the SATA cable (while unsoldered from the unit, of course) and square-braided the others. Note the difference between the 4-core cables with and without applying the rule exception at the start. (I did over the single peripheral cable after the photos, forgive its lackluster appearance here.)
Wire length tolerances can still deflect the braid somewhat, but applying the exception clearly makes an improvement.

Dual adapter cable

I applied an extra touch here – braiding the two parts in opposite directions.

Note also that if you want these adapters to remain usable with ATX12VO (+12V only) PSUs, you must ensure not to swap the grounds either, as the ATX12VO specification only connects the ground adjacent (pin 2) to +12V.
(I’ve kept their respective places in all of these examples anyway.)


Discuss in our forums!

Support Hardware Insights!

Select Payment Method
Personal Info

Donation Total: 10$

Or select from our donation goals!

Pages: 1 2 3 4