I deliberately did not use the arms on my 1U servers - 2 mains cables and 11 network cables[1] - that would not have worked well!
[1] This is VMWare - so we spread the load over a number of aggregated gig links (iSCSI, VMWare Management, vMotion, Core uplinks).This would be nicer with 10gig but that was expensive at the time (still is, but less so).
Pah.
Waxed linen thread is the way to go.
Thomas Prufer
Nice - thanks for the link
J^n
En el artículo , Andy Burns escribió:
Fitted dozens of Proliant DL360 and DL380 (1u and 4u respectively) and always chucked the arms out. They were more trouble than they were worth. The servers were so reliable they hardly ever came out of the rack and when it was needed, it was a moment's work to disconnect the cables.
En el artículo , Thomas Prufer escribió:
I actually learnt to do that years ago. Oddly satisfying.
Got some friends that learned and did that, in step-switch exchanges, way back when. Apparently it was important that the wires ran level and square, in nice radii, branching off in regular and pretty patterns. And the wires were soldered in place with a huge iron that had a copper tip like a bent screwdriver. There were various methods of tuning said tip to perfection, first filing it, and then rubbing the sides with pencil graphite, and the copper dissolve dint e solder and had to be filed regularly.
(They said the waxed string used for tying the of wires usually broke in the penultimate knot. Or just ended up 2" short, from being cut off too short to begin with.)
This was replaced by punch-down wiring (ISTR "LSA"), and pre-configured wiring with plugs, though I do not know when or in what order, along with the move from less solidly electromechanical exchanges. All servers and terminals now, nothing that needs to be greased or oiled.
Thomas Prufer
En el artículo , Thomas Prufer escribió:
Wire lacing was used in the early electromechanical (Strowger) exchanges. It was a work of art, from a time when tradesmen took pride in their work.
(third picture down)
Also seen extensively in early radio transceivers.
Not tried the "unplug and see who screams" technique, then?
Also seen proably to this day in military and avionics kit, where you want the reliability that comes from NOT having wires flapping around and breaking from fatigue..though mebbe heatshrink or some other methods are used these days.
En el artículo , Huge escribió:
I have in the past, but had other things going on at the time and decided to let sleeping dogs lie and leave it for another day. Would have been interesting to see if the black helicopters hoved into view outside my office window :)
There was a knackered old 12-port 10Mbps + BNC hub (yes, hub) in the rack for those lusers who pissed me off sufficiently. With the uplink forced to half-duplex. Allied Telesis, IIRC.
As did I when I worked for BT (or 'The Post Office' as it was then). ;-)
Yup. With 'flexibility points' allowing you to make changes without wrecking the whole thing.
Yup, and each termination had to be between 1 to 1.5 turns *then* soldered 'correctly' and no run-backs or burns on the insulation.
I don't remember doing that but it was a long time ago now. ;-)
You were allowed to join it and a trick was to hide the joint at the back (remembering where the back was when it was finally installed in the equipment). ;-)
We had to re-wire a 'Dolls eye' switchboard from scratch. So that was designing the loom from the schematic, punching headless 'nails' into a large board, running all the individual wires and then lacing them up with the waxed cord. One of the fun bits was pulling the finished loom off the board and hoping you hadn't left any wires out or left any out of the lacing. ;-(
No selector arms or relays to 'stroke' and 'tweak' ... ;-(
Those were the days .
Cheers, T i m
p.s. I thought I was going to be impressed by all the wiring pictures mention on the first post but consider most of them too inflexible ITRW. If I was bringing a bunch of cables to say a 24 port managed switch I'd make them all the length required to reach the furthest post and then just dress the cables back from there. As long as they are 'tidy', very few would care if they weren't 'pretty', especially if you couldn't plug their particular cable into a temporary port or even switch *quickly* (assuming no patch panel etc).
Quad, the Hi-Fi equipment maker, still used cord lacing until at least the early '90s. On, for example, the 405.
Not sure it does anything cable ties can't do.
Based on what I used to see in the GEC production lines, it was quite often a combination of both lacing and heatshrink. During production the looms are pre-prepared using forming boards with pins marking all the end points of each wire. Once all the wires are in place, they are laced. Then they are heatshrunk (usually with adhesive lined heatshrink). Lastly the connectors go on, and more heatshrink or boots added on the connectors. Usually all nicely terminated with water poof brass circular multi-pole connectors on each end with screw thread locking nuts.
You can always use another short patch lead and an RJ45 coupler...
True, assuming you can get to the offending cable in the first place! (With some of the pictures I'd say a connection in the middle of the bunch was pretty inaccessible).
Having big bundles of say Cat5 tied together tightly / regularly makes them both pretty inflexible and can put strain on individual connections.
No, by all means have the main runs loosely tied together but I'm all for flexibility when it comes to the actual devices themselves ... after all, even the most scruffy of cabinets will generally *function* just as well as a tidy one (or even better if you take crosstalk into account) so it's only just about the 'looks'?
It's the same with wiring inside a PC. I would always generally route cables out of the way of the fans and keep them 'tidy' but I *never* bound IDE cables up into bundles because I believe it can have a detrimental effect technically ... and made the whole thing less flexible and easy to work on.
Cheers, T i m
I've always wondered about putting cable bundles in heat shrink because of how inflexible it then makes them? I've noticed on car and motorcycle wiring looms how the binding 'tape' is often not even slightly adhesive, allowing easy(er?) flexibility of the bundle as a whole.
Maybe not the thing you want in the wiring of a missile though? ;-)
On that, I was working with a guy who had worked at BA on the 'Rapier' missile I think it was. He said that because of the G forces involved, they didn't mount the electronics on a PCB but soldered them all together in a mesh with wires attached at the relevant points and then scrunched it all down into a small ball. Then when it was teased open (so that there were no shorts etc) and tested it was then dipped in resin to form a solid 'lump'.
Apparently the small explosive fired turbine / generator produced over
100A but would only last for 30 seconds before melting down. Not really an issue when the average flight time was 7 seconds. ;-)Cheers, T i m
I've still got a reel of waxed lacing twine, only partly used.
Some of these look like patch panel rear connections. You don't repair these if they break - you include enough spares for the life of the data centre wiring, and the ability to expand them if necessary. There's no slack loops on any of these -- they are each cut to exactly the right length to run between the relevent patch panel sockets.
You are't supposed to use tie-wraps (even releaseable ones) any more, as the sharp edges distort the cable sleeve and can affect performance in edge cases.
For avionics stuff, rigidity was often partly the goal - the cable form would go in its intended place and stay put almost by itself.
Inside individual LRUs its a different game again - I was thinking more the cable looms that join the boxes together and form part of the vehicle / aircraft etc.
Never seen that style of construction - but I suppose it would work!
I have looked inside things like FLIRs, which tend to be hermetically sealed double skinned magnesium boxes with single euro card PCBs. They would have heat ladder on the top of each PCB, then components tied, clamped or glued on before conformal coating. The board then engaged and locked into side rails in the box (forming a thermal path from the heat ladder). The walls of the box had forced air flow through them from a blower in one end of the box. It was pretty high density and very rigid. Having said that it was designed for normal jet flight levels of acceleration and vibration. A missile may need to survive higher levels.
A bit over designed really - lasts three times the working life of the unit ;-)
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.