High Resolution Encoder Help

[Donovan]

I need some help understanding encoder pulses I guess you would call it. My plan is to make a old Brown and Sharpe OD grinder into a cnc. So I am looking for a high resolution encoder. I have read that some of the new cnc grinders have 1,000,000 to 4,000,000 ppr encoders. How are they coming up with that? I can only find 4096 line ppr encoders. Does the computer multiply it, or how do they get 1,000,000ppr. I am looking for very fine movements. I found these encoders what do you think. http://www.fagor-automation.com/Line...ryEncoders.htm I also am looking at the www.machinemate.com control. One more thing the ball screw that I am looking at is 32mm with a lead of 4mm. I don't know if you need that info but there it is. Thanks

Donovan

[Al_The_Man]

My question is why do you want such hi resolution encoders, if you put a 4096 p/rev encoder on a 4mm ball screw and the encoder was used quadrature for 4096x4 = 16,384 counts/rev, on the end of that ball screw you would have a resolution of .9 to -10 The other thing to consider when going very high resolution is the capability of the control to read the count frequency at the desired feed rate.
If you had the encoder on the end of any gear/belt reduction the count frequency etc will be even higher.
Al

[Donovan]

I need my slide to (or the grinding wheel) repeat to the .000010" in size. So the better the resolution I am hoping the better the grinder wheel repeat. I would use a glass scale to position it but I have heard that the glass scale will run into conflict with the encoder and the servo will hum. When that happens my surface finishes go to hell. That is what I am told for a guy that has done about 100 cnc grinder retrofits.

[NBrigode]

They are coming up with the high ppr counts my gearing down the encoder to the shaft. In a quadrature encoder there are 4 peices of information in each count. So if you have a 10,000 cpr encoder you are getting 40,000 peices of information in each revolution. Currently I am involved in designing a telescope traking system. We are trying to acheive sub arc second accuracy. The best way that we can do this is by putting a very large gear reduction with a high count encoder on the servo control. I have spent some time researching encoders and I found Agilent AEDS-3200-Txx series encoders. They may be too small for your application, but I will upload the PDF I have from agilents website but they have a very detailed information on how they work and the setup.

Hope this helps
Nick

[lerman]

If you use a worm gear with 360 teeth, you have one degree per revolution of the worm. A second of arc is about a 4000th of a degree. So, 1000 cycles per rev (or 4000 counts) would give better than a second of arc resolution. 40,000 counts would give better than a tenth of a second of arc.

For the grinder problem (assuming linear motion), you could use a 40 pitch screw (like a micrometer). That would give .025 per revolution. A 1000 CPR encoder would give .000 025 per cycle -- or .000 006 25 per count.

Of course, that wouldn't be a ball screw, but if you need that precision, I assume you are moving fairly slowly, and wouldn't think that would be a problem.

Ken

[unterhaus]

They are using multipliers to get that high resolution. If you have a fairly large code disk, you can get a large number of physical lines on the disk. Then using the analog signal, you can get a fairly decent number of subdivisions of that. Look at Heidenhain or bdiied.com. The Japanese encoder makers go to very high resolutions. There was a guy on ebay selling cannon encoders that went up to 250000.

[Al_The_Man]

I need my slide to (or the grinding wheel) repeat to the .000010" in size. So the better the resolution I am hoping the better the grinder wheel repeat. I would use a glass scale to position it but I have heard that the glass scale will run into conflict with the encoder and the servo will hum.

I would think that your biggest problem when working at those repeatable tolerances is going to be machine growth (Expansion) as the temperature rises on the machine, You can have the smallest encoder resolution possible, but if the machine grows .01" over a production period there has to be some compensation.
Due to low Thermal Expansion Coefficient of glass, are why glass scales are used on the final motion, even then the accuracy is only good up to the reading head of the scale, any growth happening after the reading head introduces error.
With the accuracy you are aiming for you should be aiming for zero backlash on all ballscrews, gearing etc, There are also systems out there that utilize 2 encoder feed back on each axis, one on the motor and one such as a linear scale. I retrofitted some large boring mills a while back that were originally fitted with a LVDT for machine growth compensation, The difference between the start of the shift when the machine was cold, to the end of the shift, the LVDT had recorded a .05" growth. They were scrapped on the retro-fit and the operator would dial in a tool wear comp. as the shift progressed.
Al