Suitable cables for CNC

[moffy]

A friend has asked for help constructing a DIY CNC machine. I haven't seen it yet but I was sent all the documentation on the GCode controller, the Motor Drivers and the motors, 2 phase steppers 60V/8A with an encoder. So I have drawn up a rough schematic, but am thinking about the cabling. For the actual CNC machine I am thinking of twisted pairs for all of the signals including the stepper motor drive phases for flexibility in the cable drag chain and noise immunity. Would any sort of shielded cable be better for say the limit switches or the stepper motor encoder signals? The shields I would consider could break/fatigue and the extra insulation would make the cable less flexible. Low level signals are 5V and 10s of ma, highest level would be the 60V/8A steppers, plus whatever spindle motor is chosen, as yet unknown. Your experience or opinion is appreciated. Allow for a machine say 2m by 2m for discussion purposes.

[Doctorandus_P]

For cable chain guides, you need extra flexible cables which have many thin strands of copper, Or else you will fatigue the copper wires over time and they will break.

I do like the idea of using twisted pairs for stepper motor windings and end switches, but as far as I know this is not very common. And unless you want to do the twisting yourself with individual strands, you will be limited by whatever cables you are able to buy.

Mechanical switches do have a minimum switching current to keep the contacts clean. Inductive end switches do not have this problem, but still I would prefer to put some current though the wires and connect it as a current loop, instead of just sensing voltage.

And for the controller part. Any cables connected to a PCB are an opportunity for EMI to both enter and exit the PCB, and needs proper filtering.

[moffy]

@Doctorandus_P, good advice, I was planning on twisting suitable flexible cable myself, it's pretty straight forward.

[Doctorandus_P]

Normal cables have two layers of isolation, which reduces the chances of damage though mechanical wear and chafing. The easy way out is to have a look at which cables you can buy, and if you can't buy cables with twisted pairs that are capable of handling stepper motor currents, then you can assume nobody does this and it's apparently not needed.

But also, I've read plenty of problems with CNC machines such as plasma cutters and EDM which generate so much noise that everything becomes unreliable.

One common remedy is to use different cable guides for power cables and signal cables. Physical distance between the cables reduces coupling and thus interference. Especially on longer cable runs (such as though factory buildings) coupling can be so bad, that switching loads on mains cabling, can blow out filament lamps on 24V signal cables. Normally these cables are strictly separated with a metal shield in between, but that was in a place where they cheaped out on the cable guides.

[pardo-bsso]

Stranded ethernet cable also works if you are in a hurry.

[reboots]

Digi-Key and other distributors sell name-brand "energy chain" cabling made specifically for this purpose, but it's fairly expensive. There are Chinese knockoff products for quite a bit cheaper, although I have not used them myself.

https://www.amazon.com/dp/B0CJDXWRXX

Canare Star Quad microphone cable might be an economical alternative for switches and other small signal pairs. It's designed for flexibility, longevity, and excellent noise rejection. Note that while there are four conductors, they are designed to be used as a single signal pair.

https://www.canare.com/analogaudiocable
https://www.markertek.com/product/l-4e6s-bk/canare-l-4e6s-star-quad-microphone-cable-by-the-foot-black

[Doctorandus_P]

Stranded ethernet cable also works if you are in a hurry.

Even when stranded, CAT5 etc, does not have many strands, and I would not use it for energy chains.

Using the silicone insulated wire also used for bench test cables may be a good choice. They often have very many thin strands, which makes them both supple and fatigue resistant, but I'm not sure about the silicone insulation. silicone is not very strong, but it is quite tough. The part I'm unsure about is how that translate to abrasion resistance due to repeated movement and friction.

[Randy222]

Internal wire chafing is probably something to look at for cable that has to continually bend into and out of shape.

[moffy]

@Reboots, some excellent suggestions, I might purchase some of the Amazon version just to have a look.
@Doctorandus_P, I was thinking along the same lines about the silicone cable, but of course abrasion resistance isn't the best.

[moffy]

If it could be afforded, something like this PTFE insulated 19/0.12 would be a good choice for the signal cables, twisted: https://au.rs-online.com/web/p/hook-up-wire/8776203

[jbb]

You’re probably already thinking about it, but don’t forget about the terminations. You’ll need to design how the cables bend: nice and even on the drag chain, and then fully immobilised before you terminate the far end.

Given it’s for a CNC, might the cables be exposed to coolant, cutting fluid, oil etc? If so, chemical resistance will be a big factor. I’d probably start off looking at silicone test lead cable too.

[moffy]

You’re probably already thinking about it, but don’t forget about the terminations. You’ll need to design how the cables bend: nice and even on the drag chain, and then fully immobilised before you terminate the far end.

Given it’s for a CNC, might the cables be exposed to coolant, cutting fluid, oil etc? If so, chemical resistance will be a big factor. I’d probably start off looking at silicone test lead cable too.

Good point, coolant could well be used, both silicone and PTFE have good chemical resistance. Shrink tubing for the terminations?

[jbb]

I was really meaning that the cables should be securely fastened down on exit from the drag chain, and the final run of cable into the sensor / motor / PCB / whatever should not move at all.

[tooki]

For what it’s worth, the best manufacturer I’ve found for super-flexible wire (LiFY) and cable is Kabeltronik from Germany. They use finer stranding than anyone else I’ve found. (E.g. for their 0.25mm² they use 128 strands while everyone else uses around 60.)

[jpanhalt]

Daburn also produces an extremely flexible wire with a modified PVC insulation (not silicone).   It's generally made of awg 44 bare copper  (i.e., 105/44 for awg 24).  It's also sold a multiconductor cable.

Link to the single conductor version:
https://www.daburn.com/2671UltraFlexibleSub-MiniatureWire-U/LSTYLE15681692.aspx

I have never compared it to Kabeltronik wire.  It is available from DigiKey in the US, but it is quite expensive compared to other versions.

[tooki]

Daburn also produces an extremely flexible wire with a modified PVC insulation (not silicone).   It's generally made of awg 44 bare copper  (i.e., 105/44 for awg 24).  It's also sold a multiconductor cable.

Link to the single conductor version:
https://www.daburn.com/2671UltraFlexibleSub-MiniatureWire-U/LSTYLE15681692.aspx

I have never compared it to Kabeltronik wire.  It is available from DigiKey in the US, but it is quite expensive compared to other versions.

I’ve been meaning to try that. It’s got thinner insulation than the Kabeltronik.

This is the single conductor:  https://www.kabeltronik.de/en/kabeltronik/sale/art/hook-up-wire-stranded-extremely-flexible-lify/

And the multiconductor cable, which is also gloriously flexible: https://www.kabeltronik.de/en/electronics-industry/control-cables-unshielded/art/control-cable-extremely-flexible-lifyy/

Their PVC sensor cable is only slightly less flexible than the one above: https://www.kabeltronik.de/en/electronics-industry/sensor-cables/art/sensor-cable-sensocord/

[pardo-bsso]

Even when stranded, CAT5 etc, does not have many strands, and I would not use it for energy chains.

Good point.
I used a not so cheap patchcord but then the bending radius was quite big.

[Infraviolet]

Just a thought, as sooner or later I'm starting a CNC build of my own...
What might be wrong with the following idea?

Instead of having cables routed via cable chains and having them need to bend  and unbend relatively tightly (down to the radius of curvature of the chain), what about having all cables leaving the machine going straight upward, then arriving overhead at a gantry (higher up and much lighter weight than any gantry on the CNC machine if it is a gantry type design), with some slack along the way. From the gantry the cables would be rigidly fixed as they came back down at the side/back/(other fixed location)  and went to the control box(s). As the cables would all be going upward they wouldn't risk dangling in to any of the moving parts or work area, and whilst they'd be moving around as axes of the CNC moved, it would be just a gentle movement (like the sort of movement a pendulum makes as it swings) in the cables' slack rather than a tight bending. To visualise it think how intravneous drip piplines are hung, only instead of a bag at the top you're cables come back down via the rigid strut that is holding the high "gantry" up. The only disadvantage I can immediately see would be that it would need to be rather higher than the machine, so not an option if the machine is already relatively tal, or mounted on a high table, and the ceiling is low.

[Roehrenonkel]

Hi moffy,
 
i can recommend Ölflex from Lapp
https://products.lappgroup.com/online-catalogue/power-and-control-cables/various-applications/pvc-outer-sheath-and-numbered-cores/oelflex-classic-110.html
or H05 or H07 (harmonised cables) and high-quality "stage-proof" microphone-cables for sensors/switches.

Good luck

Edit: H07RN-F to be specific.

[radar_macgyver]

TPE and PUR jacketed cables are usually specified for high-flex applications. In the US, a good supplier for prototype and low-volume applications is IGUS. They offer various grades of cables, depending on the amount of flex encountered, chemical tolerance level (water vs oil vs corrosives) and whether the cables are tolerant of twist (in addition to flex). Their website lets you buy in small lengths.

[moffy]

Just a thought, as sooner or later I'm starting a CNC build of my own...
What might be wrong with the following idea?

Instead of having cables routed via cable chains and having them need to bend  and unbend relatively tightly (down to the radius of curvature of the chain), what about having all cables leaving the machine going straight upward, then arriving overhead at a gantry (higher up and much lighter weight than any gantry on the CNC machine if it is a gantry type design), with some slack along the way. From the gantry the cables would be rigidly fixed as they came back down at the side/back/(other fixed location)  and went to the control box(s). As the cables would all be going upward they wouldn't risk dangling in to any of the moving parts or work area, and whilst they'd be moving around as axes of the CNC moved, it would be just a gentle movement (like the sort of movement a pendulum makes as it swings) in the cables' slack rather than a tight bending. To visualise it think how intravneous drip piplines are hung, only instead of a bag at the top you're cables come back down via the rigid strut that is holding the high "gantry" up. The only disadvantage I can immediately see would be that it would need to be rather higher than the machine, so not an option if the machine is already relatively tal, or mounted on a high table, and the ceiling is low.

Overhead suspension should work, but the cable runs would be longer and untidy, very visible.

[moffy]

I purchased 20m of 0.2mm2(19/0.12) PTFE twisted pair cable, might well do the trick for the signal cables: https://www.ebay.com.au/itm/126539649000

[moffy]

One common remedy is to use different cable guides for power cables and signal cables. Physical distance between the cables reduces coupling and thus interference. Especially on longer cable runs (such as though factory buildings) coupling can be so bad, that switching loads on mains cabling, can blow out filament lamps on 24V signal cables. Normally these cables are strictly separated with a metal shield in between, but that was in a place where they cheaped out on the cable guides.

Just realised how that can be done by having two cable drag chains, one on each side of the CNC and run power down one side and signals the other, keep them separate and distant. The spindle motor kinda worried me, it's a 3 phase 3.8kW induction motor with variable speed drive, bound to be noisy, but then so are the steppers.

[SeanB]

You want trailing cable for the power side, which is used for hoists and such. They actually have versions with integrated optical fibres, and Cat5 conductors in them as well, along with steel flex cores that keep them from moving.  But are kind of expensive, but are good for millions of cycles.

[donlisms]

I do not have any CNC cabling experience, but in audio, the most flexible cables I’ve ever used are Mogami, including star quad, and 2- or 4- or 12- or more-pair snakes - still amazingly limp.  I don’t know how the spiral copper shield would hold up; probably okay if it’s grounding outside the signal flow. It really helps with the flexing. But might tend to bunch up over time.  Dunno.