Can someone help? motor/encoder in fixture


Thread Starter

Vince Turnbull


I am new to encoders and their use. I designed a fixture for testing a device which uses a motor/encoder as a means of applying torque to a shaft. The collet at the output radially deflects a device which I must measure to 1/2 degree. The motor assy was dictated by the electrical engineer on the project and I now come to find that it has backlash and windup that are post encoder. I am looking for an easy way to either change the motor to one that positions accurately or preferably fit a encoder of some sort into the fixture with minimal rework. In an ideal world, I could bolt on a non-contact reader that would sense a piece of graduated tape on a disk attached to the output shaft and forget the motor position sensing. Does anyone have any an idea of an easy way out.
It sounds like you need to couple an external encoder as directly as possible to your shaft. Can you attach a small encoder to the end of your shaft? Dynapar, and probably others, makes a hollow shaft encoder that could be attached directly to the shaft.

Bill Sturm
This is a standard product from quite a variety of sources.

We have had excellent results from:

* Northstar linear encoders (these are actually ELGO units from Germany that have been resold in the US by Northstar - formerly Lakeshore Encoders)
* Sony Magnescale also is a good product
* SIKO also makes a product for this.

Outputs can be quadrature or SIN/COS depending on your needs. Based on experience and price, we would probably look at the Northstar units
first, talk to Brian Eaglin ext 280 EMIX Product Line. PH # 614 891 2243

This should get you right on the money.

Ken Brown
Applied Motion Systems, Inc.

Michael Griffin

I assume you have a gear box in the system? If so, it is no surprise that you see some problems under high torque. You have not mentioned where the measurement is fed into, or if the encoder is necessary for process feed back as well as for measuring. If it is used for servo feed back you should be careful before you move it as backlash and windup may cause control
You may find it advantageous to use two encoders, one for control and the other for precise measuring.

However, if you have the room on the shaft near the collet you might be able to fit a hollow shaft encoder into the system. Hohner for example ( makes such devices.
You could also use a toothed pulley (next to the collet) and belt to drive a conventional encoder. Since the belt drive only has to drive the pulley, there should be minimal windup. With this method you could use different pulley sizes at the collet and encoder to give you more resolution (e.g. more encoder turns per main shaft turn).

Michael Griffin
London, Ont. Canada
[email protected]

You did not mention if your goal is run open or closed loop around position or torque.

If you are simple going to a position or torque and turning the motor off (open loop), then the second encoder is probably a good answer.

I am never a strong advocate of adding a second position loop as solution, unless absolutely necessary.

Don't rule out going to a larger motor and/or stiffer gearbox. It may cost you more up front, but save time in the longer run. I believe
Bayside, Thompson-Micron & CGI are all companies that offer gearboxes with 15 arc minutes backlash.

Another possibility is to change the way you reference, using existing equipment. If you can take the reference position with the gearbox 'preloaded' in the direction of interest you can eliminate the backlash/windup as a factor. Use the known position at specified
torque as the reference position. This is not always the best solution, but certainly an option.

Dave Kane
[email protected]

Larry Essary

Alpha Gear Inc. makes gear reducers that have less than .5 arc minutes of what we call "torsional backlash". We include lost motion in with our backlash measurements so the figure we call out is absolute. Some companies measure backlash and leave out the gear tooth deflection which causes lost motion. We also measure at 4 positions and call out the high not an average! Chances are you will not only solve your problem of measurement but you will gain a much stiffer and responsive system! Please see