Is there an optimal rate of twist for keeping the system quiet? I'm asking because I don't know.
I would imagine at some point there could be so many twists it would create some funky issues (that's a technical term😉).
This is what I was going to run from the back panel fuse to the power switch, 120VAC - USA style electrical.
What about when it is not AC, but signal or DC?
If anyone is curious how they were twisted, use a drill. Start with one person holding the wires at shoulder width, then walk towards them while slowly letting the drill do the work. Then put it in a vise and tighten more. It wanted to be "springy", so I made a couple of passes with a heat gun on low then let it cool overnight in a cold garage with the drill still on one end and the other still in a vise. The next day, no more spring. I learned the trick about the drill a loooong time ago from old timers at big blue, the wires were actually for DC that had to run the height of the server rack. It was in the spec's for a multi-million dollar set of cabinets, so some genius EE probably had a clue as to why. Unfortunately I never had a chance to ask about the twist rate. We went with about 1 twist per cm with what was probably 12 gauge solid wire. In the years I had that job I did about 100 cabinets, there were only two with this twisted pair spec written into the instructions. DC rarely had to be taken the length of the cabinet, usually the device that needed an external DC power supply was placed next to it and a pre-made cable was used.
I had that job long enough to know I don't have the skills for doing the power wire panels with solid cable. Some people are artists at it, not me
I would imagine at some point there could be so many twists it would create some funky issues (that's a technical term😉).
This is what I was going to run from the back panel fuse to the power switch, 120VAC - USA style electrical.
What about when it is not AC, but signal or DC?
If anyone is curious how they were twisted, use a drill. Start with one person holding the wires at shoulder width, then walk towards them while slowly letting the drill do the work. Then put it in a vise and tighten more. It wanted to be "springy", so I made a couple of passes with a heat gun on low then let it cool overnight in a cold garage with the drill still on one end and the other still in a vise. The next day, no more spring. I learned the trick about the drill a loooong time ago from old timers at big blue, the wires were actually for DC that had to run the height of the server rack. It was in the spec's for a multi-million dollar set of cabinets, so some genius EE probably had a clue as to why. Unfortunately I never had a chance to ask about the twist rate. We went with about 1 twist per cm with what was probably 12 gauge solid wire. In the years I had that job I did about 100 cabinets, there were only two with this twisted pair spec written into the instructions. DC rarely had to be taken the length of the cabinet, usually the device that needed an external DC power supply was placed next to it and a pre-made cable was used.
I had that job long enough to know I don't have the skills for doing the power wire panels with solid cable. Some people are artists at it, not me
Attachments
It's not a good idea to use a drill as it deforms the natural twist of the wire inside. If you twist it naturally by hand it will fall into its own spacing which will be fine.
On one hand, we know that the more twists per unit length the higher the interference rejection frequency will be. But for this to work the circuit (not just the cable) needs to be symmetrical at the interference frequency. AC power systems are anything but symmetrical even at rather low interference frequencies. Plus AC systems roll-off a lot as the frequency raises. So one twist every 3 inches should be OK. But from a mechanical viewpoint, it won't keep the wire uniformly together. So let's go with about one twist every inch.
I twist the wires fairly tight by hand, and use single strand wire, so the pair stays together and so it stays away from any sensitive high impedance areas/wires. With tube circuits I keep the twisted pair right against the chassis, and keep other wires and parts up off the chassis.
I would go for a minimum of 2 or 3 twists per inch. The actual twist length depends on the distance to the circuits you don't want to couple to: the twist length should be significantly shorter than the distance to the circuit.
The drill probably does mess up the copper strands, maybe even pull the sheath off of solid wire? I used stranded wire for the example. The "springy" feel to it was probably the sheath, it's not like I heated it up so much as to anneal the wire. I think the heat loosened up the tension in the covering. A shame because it does look pretty and makes for a rigid wire.
Why would the drill be worse than other methods, just the amount of torque which it can put on the wire?
It looks to be almost 3 twists per inch, or would that be counted as closer to 6 in the photo?
Interesting "twist length depends on the distance to the circuits you don't want to couple to: the twist length should be significantly shorter than the distance to the circuit". Trying to picture it.
Why would the drill be worse than other methods, just the amount of torque which it can put on the wire?
It looks to be almost 3 twists per inch, or would that be counted as closer to 6 in the photo?
Interesting "twist length depends on the distance to the circuits you don't want to couple to: the twist length should be significantly shorter than the distance to the circuit". Trying to picture it.
The problem with the drill is that the ends of the wires are fixed, when they need to rotate. If you do it by hand you can feel and see what the issue is.
Takes a lot longer by hand though!
Takes a lot longer by hand though!
I agree.
Look at the very complex machines that manufacturers use for multi-stranding cables and ropes and wires and probably CAT5 (although I have not seen any of this last category).
They do that for a very important reason. It is to do with strain in the wire.
If you have short pairs of wires then it is possible to twist the PAIR without TWISTING the individual wires/insulation.
try it.
Compare to drill twisting.
The effect on the wire and on the insulation is quite different.
I have twisted pairs and star quads upto 3m long. Fiddly and time consuming.
The method would work for longer sets.
I tie the ends to a door handle.
Wrap the wires around each other taking care to NOT twist either wire.
Keep some tension on the fixed end tied to the door handle and gradually work your way across the room or along the hall, as the twisting progresses.
Last edited:
the twist rate depends on the distance from or to other circuits.
If you have a large distance between the wire pair and another circuit then a very few twists per foot is good enough.
Basically the loop of one half twist generates a field that decreases with distance.
The next half twist generates an opposite phase field. At a great distance approaching infinity the two opposite phases cancel.
Using the same argument for a hundred (or a thousand) twists, each alternate half twist creates a field at a distance that is the sum of all the alternate half twist fields.
All the other half twist fields sum to pretty nearly cancel the opposite phases.
Due to the distance small differences between each half twist and small differences in the distance average out and still we have effective cancelation.
So at a big distance any twisting that is significantly smaller than distance/factor will on average create fields that cancel.
At a short distance the averaging is not as effective.
This can be improved by using more twists per foot. We are using that "factor" from above.
The evenness of the twisting affects the cancelation. The changing distance, due to kinks in the wire, or curves in the routing will affect the averaging.
cat5 uses one twist every ~15mm to ~20mm. The four pairs have a very deliberate different twist rate to help avoid crosstalk.
Cat5 is often used at very close distances to other circuits and to other very heavy current cables/cords. Yet they are substantially immune to interference. Some of this immunity is down to digital signalling.
Audio cannot use the digital immunity to interference, so slightly tighter twisting may offer improved immunity when distances are small.
What is slightly tighter than one twist per 15mm?
two twists/15mm? That would be a doubling of twist rate, ending up at ~3twists per inch
By the time we get to 3 twists/15mm (~5twists/inch) we have tripled the twist rate. That to me seems too many twists for our normal wire diameters.
Last edited:
- Status
- Not open for further replies.