I'm looking for ideas on cabling a sensor that is undergoing cyclic motion in a way that will avoid fatigue failures in the cable.

The sensor is an accelerometer mounted on a shaft that rotates through ~375 degrees and back at about 15-60 RPM. The shaft is around 100mm diameter. It might go through ~250,000 cycles in the sensor life. It's okay for the cable to fail if it goes through more than 375 degrees. What's the best type of cable to use so that the repetitive winding and unwinding won't cause failures?

Bonus points if the solution is (a) cheap, and (b) avoids lots of slack cable flapping around.

Alternatively, anyone with good ideas on how to accurately measure the motion of a church bellis welcome to put them forward. By "accurate" I mean better resolution than 0.5 degrees in the motion of the bell, bonus points for measuring angular acceleration directly.

  • \$\begingroup\$ Why is your shaft going through that much rotation, and what do you need the precise rotational information for? What is moving the bells? If they are driven smoothly by a motor, you might be able to infer the bell's position by the relationship between the current applied to the bell and the rate at which it changes speed (the latter being inferred via IR-compensated voltage feedback). \$\endgroup\$ – supercat Jun 12 '12 at 16:05
  • \$\begingroup\$ It is going through that motion because someone is standing 40 feet below it tugging on the rope. I want to measure the precise rotation so that I can model the motion of the clapper with it and accurately predict the time the clapper hits the bell for use with a ringing simulator. \$\endgroup\$ – Tom Jun 12 '12 at 22:15
  • \$\begingroup\$ If you have a rope, how about having a thin electrical cable fastened to it? Somewhere along its length, the cable could hang down in a "U" shape. I would guess the distance traveled by the rope is probably 5' or less, so one should be able to make a "U" which wouldn't get even close to either travel limit. \$\endgroup\$ – supercat Jun 12 '12 at 23:08
  • \$\begingroup\$ Actually the rope moves as much as 20', and there is not 5' of clearance between the shaft and the floor. \$\endgroup\$ – Tom Jun 14 '12 at 9:27

Use an FFC cable.

FFC cable

Arrange it in a spiral, with your accelerometer in the centre. As ThePhoton mentioned, you must use strain relief at each end of the cable, otherwise you're liable to get fatiguing there.

Rotating FFC

Alternatively, use a longer FFC, and arrange it like this if you want lots more rotation:

Revolving chain

YouTube video of revolving cable chain.

This application uses cable chain, which is better, and can take whatever cables you want. It's more expensive, but still affordable. 2.4ft of chain is less than 20 GBP from RS. You can arrange the chain in the spiral form as I have shown in my purple rendering.

If you're really strapped for cash, you may be able to get away with simply wrapping your cable around the axle a couple of times in a helix. This spreads out the bending motion of the cable along the spiral length, so that no one part of the cable is bending much. We do this in our robots where there's absolutely no room for any solutions involving cable chains. And, while those robots don't usually do 250,000 cycles, we have never seen a cable fail in this situation.

Helix Cable

  • \$\begingroup\$ I highly recommend a solution like revolving chain, or similar cable track. \$\endgroup\$ – tallganglyguy Jun 12 '12 at 15:41
  • \$\begingroup\$ I know the middle drawing is really only a sketch, but if you built something that matched the drawing, I'd be concerned about fatigue failures at the joint between the rotating part and the cable. Some kind of strain relief is needed to avoid stress concentration at the sharp angle between those two parts. \$\endgroup\$ – The Photon Jun 12 '12 at 15:54
  • \$\begingroup\$ @ThePhoton - Yes, that's absolutely right. He would certainly need to add strain relief there. \$\endgroup\$ – Rocketmagnet Jun 12 '12 at 15:57
  • \$\begingroup\$ Thanks. I wasn't sure how this cable would go with the fatigue cycles. Any opinion to offer on whether ordinary IDC cable would be suitable, or whether I'd need to more rigid stuff found in laptops / printers etc? \$\endgroup\$ – Tom Jun 12 '12 at 16:25
  • \$\begingroup\$ Strain relief on this goes without saying, I think. \$\endgroup\$ – Tom Jun 12 '12 at 16:25

As the bearing ends are closed, why don't you wrap some linear encoder strip around the perimeter of the shaft, then use a fixed encoder to read it. you can even increase the radius of the shaft by using a flat pulley and mounting the strip to the perimeter of that. None of the electronics has to move this way, and its simpler to decode the position. Avago have some smt reflective quadrature encoders that are like 3-4$ that should be more than adequate. Linear magnetic encoders are also possible.

  • \$\begingroup\$ I was going to suggest that! \$\endgroup\$ – Rocketmagnet Jun 12 '12 at 21:16
  • \$\begingroup\$ Excellent suggestion! At 150LPI around a 14" circumference shaft, the resolution should be plenty. Only two issues: 1, no-one seems to sell codestrip (at least in the UK0. 2, I'm aiming for some thing a hobbyist with minimal soldering skills can reproduce (mine are nothing to write home about) - smt is a down side here. Anyone know of a similar device in DIP or other easy package? And where to buy cudestrip? \$\endgroup\$ – Tom Jun 12 '12 at 23:05

Use a cable conduit.

enter image description here

It will prevent the copper of your wire to make sharp bends, which would cause them to break. Cable conduits have a better longevity than copper.

Pay special attention to the connections of the conduits. The cable should go straight into the conduit.


Answering your 'alternatively' question.

There are plenty of non-contact, high resolution rotary sensors around. How about these magnetic, Hall effect based sensors from Renishaw?

Renishaw rotary sensor

These consist of two parts: one is a simple magnet, which goes on the rotating part, at the end of the axle. The other is the stationary part, which has all the wires.

They're a little bit expensive, but keep running just about forever. Alternatively, you can buy just the electronics from inside, or just teh chip itself:

Renishaw rotary sensor module Renishaw rotary sensor chip

These are pretty nice, and have a 13-bit resolution across 360º, giving a resolution of 0.044º.

  • \$\begingroup\$ I've looked at these and I was going to use one for a long time - particularly the Cherry AN820032 or similar. It pretty much fits the budget (about £25 for the sensor) and does what I'm after. But when I got into the belfry for a look around, the old timber gudgeon pins have been replaced with modern bearings with closed ends, so I don't have access to the shaft end. \$\endgroup\$ – Tom Jun 12 '12 at 16:27
  • \$\begingroup\$ @Tom - In that case, you won't be able to put the accelerometer on the shaft end either, and the accelerometer reading will be affected by the angular velocity, unless you use a pair of dual-axis accelerometers mounted on opposite sides of the shaft, and average their readings. \$\endgroup\$ – Rocketmagnet Jun 12 '12 at 17:09
  • \$\begingroup\$ I planned to mount a linear accelerometer on the shaft so that one axis is tangential to the rotation and infer rotation from this, knowing the radius of the accelerometer reference point. \$\endgroup\$ – Tom Jun 12 '12 at 22:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.