Torque Sensor

Similarly, you could just measure the position (instead of load) on a spring tensioner.

Easiest solution is to use two or four load cells and with some basic physics use the force applied to the load cells to calculate the applied torque to whatever it is that the system is connected too. The other solution is just to have the wire run over a pulley that is pushing down/up on a load cell, this removes the need for torque equations and just relates the tension to force applied vertically downward/upwards onto the load cell allowing for a constant tension on the wire during winding.

Kas

If you're prepared to spend a few thousands you can buy a torque meter or even a full dynamometer from Magtrol or Torquemeters.

I designed the electronics for a torque meter on an aircraft engine that used the wind-up of the shaft as the strain element. The shaft had 2 sets of timing markers that were detected by proximity sensors. The time between rising edges varies with the wind-up of the shaft and hence the torque. Obviously this was measuring high torques, no expense spared, but the whole concept could be scaled down to be small and cheap if you can use, say, a nylon shaft with a couple of timing discs.

Aldi sells springs!? Were they spring onions? Or spring greens?

The torque calibration at motor stall is easy - you need a pulley on the motor shaft and a piece of cotton (for low torques) and a weight. All you need to do is increase the motor current until it can just lift the weight. Once things are moving it gets much more difficult to estimate the torque actually applied to the load because the mass and inertia of things starts to matter.

MK

Hi Kas,

Another way of doing this (I'm no expert, so this is just an idea in case it meets your needs) might be to connect up a motor to the thing you're measuring, and control the motor supply until you reach the point that the thing starts turning. (I tried this once, see here: BBB - Super-accurate small motor control with a BeagleBone Black ). That would be good for (say) comparisons of torque, or a rough idea of torque, but of course is not calibrated against anything. But, it is repeatable, so could in theory be calibrated first. The torque should be proportional to current, so could be measured with a multimeter. It would be interesting to try this out on several motors (now I think of it, would also make a good physics experiment for schoolkids), to establish a simple reference we can use as a torque measurement device : ) if it was repeatable across different motors (but it might not be - I don't know how precisely motors would match with each other) and if someone had something to calibrate it against.

I measured a spring too (one of a box of 200 tension springs I bought from Aldi years ago !).

I also detected only a very small change in resistance (once all the coils were separated)  - probably of the order of 0.1 - 0.2% at big extensions.

No surprise there - metal foil strain gauges typically have a gauge factor of about 2 which means that for 1000ustrain (increase in length of 0.1%) you get a change in R of 0.2%. When a tension spring is stretched the change in strain the metal wire is nothing like as big as the increase in length of the spring so it all fits together.

Real load cells usually use four gauges in a bridge arranged so that their temperatures will track very closely - the 4 spring bridge would be much worse for this.

AFAIK none of the major players makes MEMS load cells - the problem is that using large chunks of bulk silicon would be very expensive - a few people have tried making small stick on silicon strain gauges but although you get bigger gauge factors and bigger electrical signals the temperature drift and mis-match to a metal substrate mean that over all they work no better than metal foil.

Some wide range postal scales (not the ebay for a tenner kind !) use a resonant quartz sensor under tension and can achieve very good dynamic range so that they can do letters with an error of less than 0.5g while still being capable of weighing 20 or 30 kg.

If the OP only wants to measure/control torque he might do as well by controlling the current in a real DC servo motor (or even a brush-less one but it's much harder to do).

MK

One of the tests appeared to demonstrate the resistance of a spring varying under extension and then there were 4 springs in what looked like a bridge arrangement. I just did a quick manual test with a small spring and measured 236mR at rest and 238mR at extension. I'm not sure this could be consistent under all conditions of temperature and loading etc. etc. Not for precision performance perhaps.

In another scene of the video there were MEMS devices being used. Analog Devoces make MEMS parts MEMS | Analog Devices. These are precision parts that you can integrate into a mechanism to measure torque.

Can you get a translation of the commentary?

As I posted above - you can buy 100g loadcells from Alliexpress for £4.19 each, free shipping.

That way you get as many as you want with a drawing etc. They won't be very good for that money but probably better than you get in a cheapo postal scale.

This link is to a slightly different one:

1pcs strain gauge pressure sensor load cell electronic scale 100g 200g 300g 500g 1kg 2kg 3kg 5kg Range precision weighin…

MK