Repeatability and Backlash Testing

[bbook1]

I need to perform an acceptance test on a machine that has an X,Y table. The manufacturer says that it is repeatable to +/-0.00015". Does anyone have a document that they can share that specifies and acceptance test to validate Repeatability of an X,Y table?

Asked the manufacturer how they test it, they were very vague. I asked how they validate the repeatability before they ship the machine and then at the customer when it is installed to varify that it is still calibrated and working as intended, they told me they had a test but it doesn't record any data. I would like some data to validate that it does what they advertise.

Thanks for any help you can offer.

 

[Skogsgurra]

Four microns. That is quite good. Not rocket science, but still good. And not trivial to test. What scales are used? And what movement? Screw, linear motor or what?

I have used LVDTs for that purpose. Make sure the LVDTs are firmly secured and that they measure clean and smooth surfaces on the table. An LVDT easily resolves down to .1 micron - a lot better than needed in this case.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[bbook1]

It is ballscrew driven. They told me that they will mount a camera to the Z-Axis of the machine and place a target on some mounting fixture on the table. I would like them to make some random moves away from the target, in both X and Y directions and plot the standard deviation of the return position in a scatter plot. I'm not sure how many moves and what distances I should require and what information I should get from them that gives me enough information to say that it does what they advertise.

 

[tygerdawg]

Repeatability: not too difficult to do this. "Accuracy" is quite another thing entirely.

"Repeatability" means being able to return to a taught location within a specified tolerance. Typically this is done under max dynamic conditions of speed + rated payload and test conditions.

"Accuracy" is when you command the device to move 10mm, does it actually move 10mm +/-(tolerance).

For repeatability tests, I've done something like this many times before:

1. You need to determine a suitable sample size of quantity of 'taught' locations. They might be 'taught', they might be programmed locations. The population of samples should cover the entire work volume of the device. I'm an advocate of the Shainin Method flavor of Six Sigma analysis, so I always used 31 samples. '31' because that was the number required for an isoplot to determine if your measuring device was repeatable or not. But you'd probably only understand that if you went through a Shainin course.

2. You should also determime if you want to measure the repeatability to a taught point at max published "high" speed or something else. This may / may not affect results.

3. I would typically set up a dial (or digital) indicator and have my robots move it to target 'taught' locations. The locations themselves may require some tooling elements. If not done already, then get the measuring tools calibrated and make them pass an isoplot test or Gage R&R test. Design & build your tooling heavy & rigorous, the magnitude of the measured value is tiny. Be careful of vibrations.

4. Run your experiment and record the positions. Run them in random order (use a random number generator or table) to determine the sequence. You may even consider running the experiment repeatedly at different (again, random) times during the day to capture temperature effects on the devices.

5. Plot it. The points should all fall within the stated repeatability tolerance. If your vendor is being vague, then it probably will not. Be prepared to argue about it. Better yet, design your test and then tell them (before purchase) that they'll have to meet this test as part of the acceptance requirements.

TygerDawg
Blue Technik LLC
Virtuoso Robotics Engineering

http://www.bluetechnik.com

 

[blacksea]

The correct way for testing such key parameters of XY stages (i.e. accuracy and repeatability) is using laser with optical prism - and not with stage position sensors. That allow testing both parameters in customer's POI (point of interest). And testing would be performed for each axis separately as well.

 

[sreid]

The traditional "Gold Standard" is the Renishaw "Ball Bar" Test.

http://www.renishaw.com/en/laser-calibration-and-machine-performance-monitoring--6330

Basically you command a circular move and the Ball Bar LVDT length data is recorded. The display is deviation fron a constant radius.

You can probably get a Demo.

Heidenhain makes an X-Y optical encoder that can be used for the same purpose [and any other X-Y position application].

 

[blacksea]

In any case you need special tool for performing moving of XY stage with collecting Data from dual axes measuring system (LVDT, laser, optical encoder). As a rule such complex test must perform stage manufacturer for design approval and not costumer.
BTW Heidenhain KGM dual grid encoder need complex setup and have limited dimensions.