Glass scales as encoders

[Chris D]

Originally Posted by Titaniumboy Chris D, Thanks for your reply. My sense of intuition keeps telling me that a machine should be able to do more accurate work with linear scale feedback than without. I accept your statement (and Bob's, etc.) that a lot of backlash is not acceptable. Sounds very interesting. Can you tell us any details about the alternative control system that you are working on?

Certainly a machine with a linear scale is more accurate than one without. The general rule used to be that a machine with a scale would be twice as accurate as one just using servo encoder feedback. In other words, if the quoted accuracy of a machine is +-.0002" without a scale, it would be +-.0001" with a scale. Again, this was just a general rule but it does make sense. A scale helps to insure that the axis is positioned correctly, but this is only relative to the spot on the axis that is attached to the scale. Other factors influence position at the point the tool is contacting the part.

As for my control there is not much to tell at this time. It started out as a quest to see if I could design/build a CNC control that did NOT utilize a PC. I have been working on it for about a year and a half and have made really good progress with it. While it was not intended to be a marketable product, current results are starting to impress me. But that is premature as there is still a LOT of work yet to be done. I only can work on it about 8 to 10 hours a week which isn't a lot for something as complex as a CNC control.

Just some general details (all subject to change as development continues)...

200,000 steps per second max frequency with a more practical range of under 100,000

Continous path control for contouring

User configurable parameters for tailering the control of various machines

SD cards for program storage

Design to work with Step/direction type servo/stepper drives

Can run open loop or closed loop

User interface is all textual using standard VGA montior, Pc style mouse and keyboard.

No PC required except to copy CNC program onto SD card or use the DNC interface to download the programs into the control.

As mentioned, all this is subject to change throughout the development cycle.

Chris

[noob4life]

A backlash compensation table is used for (at least older) CNC machines; this is because backlash will be different at different points of each axis, from wear on the lead screws, from wear and snugness of the gibs, and from wear on the lead screw nut in combination. Such a table can be generated from a program that moves the table in different positions and directions while taking precision measurements from an auxiliary instrument. The compensation table is interpreted on the machine side of the code, not from the CAM side. This is probably your cheaper option to solving your problem than using glass scales, if you have access to such equipment & knowhow.

Our machine shop used crusty old CNC mills from the 70's but with modern controllers which compensated for all the backlash and "slop" in the machines. Dialing in those comp tables made the owner a lot of money: those tough old machines could spit out parts like new.

With contour milling, it can get complex because the direction of your approach and cutting path will be a big, complex problem with backlash: whether your cutter pulls into or pushes from the surface, your cutting tool makes a huge difference and .005" backlash will ruin your piece if all of the "slop" is in the direction of the material you want to keep --your cutter may draw your material into itself with disastrous results. I doubt that even glass scales could prevent this from occurring unless your machine knows which side of the backlash to push against.

For hole spotting/drilling, good programmers often code the axes to approach the location from the same direction every time which overcomes backlash, guaranteed. Hole interpolation is tricky, and your gain & such must be well adjusted to achieve accuracy here; you're probably better off "roughing out" the hole and then boring it on a secondary operation.

If you don't "get it", then get a cheap Enco manual mill and loosen the gibs a bit on it to simulate backlash--then try to machine an internal slot or square to learn the hard way. (sometimes that's the only way)

Hope that helps, good luck!

[Frankea]

Your problem with backlash is becoming much larger. glass scales are relative to actual movement. Where rotery encoder feedback is dependent on the tolarence of screws, gears, belts and bearings. when this problem was observed years ago, companys went to extreams to take out all slop. this way when the unit moved to a position they could dither the motor to hold position. The problem is when a servo motor dithers it never realy stops but moves the hardware slightly off location in a back and forth motion.this motion is so slight on a rotery encoder that you can hardly notice any movement. When using a linear encoder and there is slop in the system it can setup a chugging ocilation. As when the motor tries to dither. it has to move every thing to make the slight change it needs to see in the linear encoder, and over shoots. This problem has been around a long time and many things have been tried, tighting up of the system, add hand wheels that work like flywheels, exct. Now here is where the problem gets larger. there are a lot of 25 year old machines that are hitting the market that are still good but have developed slop in the hardware. you want to use the linear for the tighter location tolerance but the slop is causing trouble. the other option is to use a rotery on the motor and lose tolerance. The only why that i can see this problem being solved is if the driver board manufactures get off there dead butts and stop making the same board with no advancement. the way that this problem can be solved is simple use two encoders a linear for all location moves and a low grade rotery on the motor to give dithering location. the rotery would not be used for any locations. once position is made on the linear a sample point is taken on the rotery to dither up and down from. this sequence would repeat at each move. if done in this manner the servo would not care about slop or errors in the screw. the backlash control would then work without sending the servo into convultions. at the same time most of these machines run in the 100 volt range and the board would need to handle the higher power requrments. At this time i do not know of any one that makes a servo board like this. If any one does let me know i have two Anilam machines that would be perfect to use them on

[Al_The_Man]

Originally Posted by Frankea At this time i do not know of any one that makes a servo board like this. If any one does let me know i have two Anilam machines that would be perfect to use them on

Did you watch the video link in post #15?
This explains the PID loop being tuned on dual loop using a Galil card, this is basically how it is handled by Mitsubishi and Fanuc etc.
Al.

[rokag3]

the way that this problem can be solved is simple use two encoders a linear for all location moves and a low grade rotery on the motor to give dithering location. the rotery would not be used for any locations. once position is made on the linear a sample point is taken on the rotery to dither up and down from. this sequence would repeat at each move. if done in this manner the servo would not care about slop or errors in the screw. the backlash control would then work without sending the servo into convultions

I agree this is a very clean approach try to contact Mariss(gecko) or UHUHP group

[samco]

backlash is bad. Lets admit it. Our large maching center (k&t from the 60's) has all the axis running on roller bearings (think of them as recirculating kneedle bearings) so if there is any backlash in the system - you can actually push it back and forth by hand (a few tons). The cutting tool will have the same effect. That is why on the good old machines - there are usually adjustments to take the backlash out.

anywoo.

as has been posted earlier in the thread - Emc2 has experimented with this. They summed 2 pid loops - P and D in one pid loop from the servo encoder and I in the scale pid loop. He also talks about what happened when they tried to just use the linear scale...
http://jmkasunich.dyndns.org/cgi-bin...-02-20-08.html

We were able to get the X axis working nicely, without hacking any of the EMC code. We used two PID loops, and summed their outputs together in HAL. Both loops get position commands from EMC. One of them gets its feedback from the encoder on the motor, and the other one gets its feedback from the linear scale. The Proportional and Derivative gains, and the FeedForward term, are set for the loop that uses motor feedback. That loop has no Integral gain. The other loop, using the scale for feedback, has Integral gain and nothing else, to correct for the steady state error between screw and scale. As always, it took some trial and error to get things tuned properly. The P gain was a lot lower than I expected, probably as a result of the huge mass of the table and the two angle plates. After everything was working with both feedback devices, we started experimenting with gains in an attempt to use only the scale. We didn't really think it would work, but if it did it would save Stuart the expense and hassle of mounting encoders on the other motors. (The Y axis motor is fifteen feet up, on top of the column.) We were right - it didn't work. When we reduced the effect of the motor encoder feedback, the axis went unstable. We got to hear what it sounds like when a multi-ton table starts to oscillate at 5-10 cycles per second, pushed by a 45 amp servo drive. Fortunately we didn't break anything.

hal example http://wiki.linuxcnc.org/cgi-bin/emc...es_On_One_Axis

I don't understand the step/dir route. (I am biased) when you can get inexpensive interface hardware that will allow you to buy cheap servodrives off of ebay that take +/-10v or pwm/dir input. Plus the feedback goes back to the control (again talking about emc2) Just like the big boys.

sam