Torque feedback from steppers

Interesting...
Sorry to ask this but...why do you want to measure torque?
Presumably the steppers are producing 'enough'.

Have you looked at the current wave form on a 'scope?... my guess is that there will be a high initial current spike corresponding to torque which will rapidly decay as the motor starts to take it's step.... average current may not show it.. Perhaps a peak hold circuit to measure this?

We drive small PM DC motors in a pulsed mode on our peristaltics and I've noticed the huge currrent spikes...makes it a bugger to fuse them effectively (Slo Blo surface mount fuses are a stupid price)

Del

Hello Del,

You are trying to get the most 1st replies don't you? Many threads have you as 1st replier so to speak.

Steppers respond to a pulse by moving the magnets, on the rotor, to the new position between the energized coils, the stator. The movement is obtained by sequencing the pulses in rapid succession. At standstill the coils are still energized with maximum current, which is inherent to the system and does not tend to vary with load.

This is how steppers are different from DC motors which just turn with constant voltage, go faster with higher voltage and go slower with load but are compensated by higher current. In this system you can monitor the current and combine readings with fixed parameters so you know what power the motor is giving or what torque it is performing.

With steppers you always have max current over the coils at constant voltage. The only difference you can get is with pulse modes in single, dual or quad type generators. There is no or little feedback here. A scope will just give you the castellated figure of signal max or none, almost digital.

In the case of motor not delivering enough torque to move the load, it just looses the pulse count and the positioning can not be guaranteed. The motor should therefor be carefully chosen for the application.

The reason I want to be able to measure torque (actually force) is to enable me to see the difference between one build and the next on some test bench. I would like to see the mechanical inertia difference between products, depending on components and build quality. Unfortunately we use steppers so I don't think I can use the motors for their torque value but maybe I am wrong.

My other option is to use a force sensor and record the force needed to move the linear slide along and plot force over distance. Any variation should relate to difference in build and component quality.

Hoping people here might have some experience with similar in order to help me along.

Thanks for your reply, much appreciated.

Ah I geddit..
Cheers I've not played with steppers...your lucid description was helpful.

Where you could fit a torque sensor depends on the mechanics really.

It sounds like you are trying to measure the torque required to move the mechanism rather than per-se the torque provided by the stepper?

If you want to measure the max torque from the stepper then that's easy enough (measure the stall torque..or hang a spring balance on the slide and see what it reads at stall)

If you want to measure the torque to move the slide then maybe the simple spring balance pulled manually, slowly increase the pull until the slide starts to move. Ok it's not torque... but isn't it Resistance to motion you are trying to measure.

If you are checking for tight spots then good old feel is probably your best bet...unless you have a BIG budget and want it automated.

Sorry in advance if I am miowwing up the wrong tree.

Del

Hi again,

No you are not meowing up the wrong tree at all.

The problem is that we use precision machined parts that combined with high grade ball slides and very accurate build, need to produce movement accuracies of submicron range. Sometimes for no real reason, the product gives us grief and we are not always able to find out why even though we test them under 40 times microscopes with a precision etched reticule plate. It seems that the build and or the machining process sometimes determines the way the product performs even though we take the same care all the time.

I just think we maybe could predict better if the product is going to be fine by measuring the forces needed along the whole movement before final assembly with the motors. That is basically what I am trying to get to with this exercise, better pre testing.

The feel factor we already apply. The hand drawn spring balance meter will not be accurate enough, don't forget the submicron business which is very small indeed.

I am looking at these little force transducers that convert force into mV over a range. Not sure if we can get them small enough but something to try.

Ah..I see you are at a finer scale than I imagined.

But...bear with me here, I still feel a manual approach maybe simplest and cheapest..
Imagine if you will pushing your slider along it's track with a feather, wouldn't any slight change show up admirably in an increased flexing of the feather. (Think of a cross between a feather and one of those cheap toque wrenches where the shaft actually flexes maybe?)

Look at this 'how to crack a safe in 5mins' blog entry if you doubt the subtlety of our senses.

Even if you get loads of sensors and stuff you still may find the data hard to interpret, or you may find the measurement system causing problems itself.

I am a lo tech Cat. Simplicity rules.

Del (PS I may be able to assist in the feather acquisition process )

Stepper motors are very similar to synchronous motors in construction, although they generally have higher stator winding resistance. Would it be useful to use an in-line torque sensor and set up a test jig which varies winding current or voltage?

Alternatively, if you fit a stepper motor with an encoder with the same number (or multiple thereof) of lines as the motor has steps to monitor when the motor starts losing steps. And then accelerate a mass (flywheel) and measure the time it takes to get up to speed without losing steps will yield torque (e. g. Torque = Mass moment of inertia X angular acceleration).

Hi Case491

Galil had something about it but I am losing the paper war and cannot find the printout. Please browse the Galil site yourselves, I am sure you will find something relevant there.

We changed over to Galil controllers and I must tell you it is magic. Big speed increases with the same power supply without loosing steps.

Hi case491

The following site has some hardware/software for measuring and plotting stepper motor function and torque.

http://www.euclidres.com/apps/stepper_motor/stepper.html

Check the home page too. Its probably not cheap but looks pretty intense and thorough.

Just in case you haven't already been there.

Very interesting software, thanks.

We have been researching laser interferometers for submicron and nano meter movements to see what the dampening and resonance times were for these systems. It seems we could perhaps do something similar with this analytical software.

Cheap or expensive makes no difference, if it turns out useless I rather have spent hundreds on software then thousands to tens of thousands on laser optical systems.

I will study these pages and see if we could use them for future analysis.

I hope the above info will be of some help to you.

I also like the idea by andy germany of using a servo motor just for testing purposes of similar size but of a predetermined lessor % (window) of torque capacity than the stepper you are going to use.

You can then plot the torque over distance and time to pinpoint any mechanical anomolies and if it completely overloads the servo at a particular point or points then you can decide if the system will work properly using the stepper motor. The go/no go thing I guess.

But then you can always mount Del's feather on a known linear resistance and watch it closely for changes in deflection. I'm afraid I lost all of my quatloos in a poker game with a bunch of drunken dunebians but if I had any, I would wager that something that simple would probably work as well.

This sounds too much like fun. I know you'll make the best choice for your circumstances.

Good luck!

Jeff

We are discussing the software with our test engineers and with our R&D and software development department over the next coming days.

It looks promising and many thanks to you for that link. Do you work with the stuff or did you find it by accident? We just did some laser testing a few days ago and the equipment will cost in excess of £6000 and can go up to £15000 for all the bells and whistles. This software could probably do it sufficiently enough for us to eliminate that cost. The only proiblem would be that we would have to fit encoders to ALL our products for the test phase of the production. This would mean some change to the design but not too much to be a real problem I guess.

Del's feather is perhaps too close to his fetish to be of any use, he would probably giggle uncontrollably for the feather to be reliable

Hi case

You are most welcome. I hope it will do what you need. Please keep us informed as to results.

I'm sort-of semi-retired and haven't worked with any steppers or servos in a few years. We usually used software supplied by the motor manufacturers and mostly for troubleshooting. We really didn't require anywhere near the accuracy that your system requires.

I remembered the name from a search for an overall software package several years ago that we could use on all of our systems instead of individual programs. But it was over-kill for us and the bean counters decided we didn't need it. I Hate when that happens.

Take care.

Jeff

P.S. Too bad you're not trying the feather. I've never seen a cat laugh although I have one that smiles alot.

A few thoughts:-

1) you may need some sort of micro torque unit measuring unit that you can place between the stepper motors and the unit to be tested (assuming that you need this only in the testing, not all the time), thus you can actually see the amount of torque needed to turn though a specific range of turns. Assuming that the number of turns is more than one, you may need some sort of RF readout to obviate the need for cables......

I have never seen such a device, but maybe something already exists....or there might be a need for an externally shown spring scale that can be read by a video camera(s) in various positions.

In such a case, maybe only turning by hand will deliver enough infos to decide on whether a unit is good or bad......as once the scale rises above a certain torque, you will know whether the steppers could supply that reliably or not.....with your present system, it would appear that a unit that works each time will be accpted as OK, even though it is possible that it needs 99% of the stepper motor torque each time, which is really on the knife edge of working!!!! those units also need to be marked as defective really.....you need only units that will work with say 60% of the highest stepping current!!!

I also feel that servo motors, only for testing at least, may be a better bet, though I expect one could make some testing software that varied the input to the stepper motors to see at which point in the current to the motor(s), that the required movement correctly started and finished....comparison with say 5 bad units and 5 good units should give you enough info for a go or no-go test parameters.....

2) Although I have never seen one personally on a stepper motor, you could add a precision positional optical feed back system to know exactly at which point and how far any motor (not just stepper motors) (and how much motor current) has turned.....that might give you eventually a better handle on where the mechanical stiffness is coming from too on known "defective" units......

Interesting problem.....

Perhaps a flex mounting to the frame of the stepper fitted with strain gages.

Unless I'm sadly mistaken every sort of torque or force sensor ulimately relies on something flexing...be it a strain gauge, a peizo element or a micro machined gizmo...

That being true, it begs the question why is a millivolt output on a scope/screen/logger any better than simply eyeballing somthing that is flexing in the first place?
Surely it's a question of 2 things... resolution and displaying the data.

I'll still bet a thousand quatlous on my feather based purely mechanical system .

(Is anyone going to play or shall I just skulk of in search of Tuna elsewhere?)

Del

Dear Del,

I apologise for the omission of my gratitude of the super graceful feather solution you once proposed, entirely my fault so please do not sulk.

May fresh tuna rain down on your house for weeks to come, I promise to try the feather, once I have established the fore mentioned software does not suffice.

Maybe in the meantime, you would be so kind to submit to me a table containing the bend stress factors of different feathers in arc seconds, relating to slide forces in micro newtons over a distance of 120mm.

Very kind cat you are indeed

I karnt do that cuz I dont undystand them number thingies.. I'm a qualitative Cat not a quantitative Cat...
I shall ponder the questoing of feather kalibration .

GA.

I see exactly where you are coming from, the same place as I was.....it cannot be that difficult!