Blacklash with glass scale Heidenhain

I have a plan to buy a new VMC with glass scale Heidenhain. But I wonder "Does it has blacklash when using glass scale Heidenhain". Someone says yes, others say no.

anyone can anwers this question.

Thanks a lot.

Backlash is a mechanical issue. Yes.
A glass scale can give an accurate position. No, it can't stop backlash.

You can have accurate position, with backlash, but as you change direction, the other axis must wait (pause) while the bad axis 'catches up'

So the answer is yes and no.:withstupi

Because of direct mesurement, most of the backlash related problems can be avoided when using Glass scales.
Also the positioning accuracy of the machine will be improved when using linear scales.

So always good for using linear scales

Sivadas

Well, if I might add, a glass scale has no backlash but it has a very small linearity error. In most cases the error is so small that it does not matter. The glass scale positioning feedback can help to compensate for most existing mechanical backlash to some degree. It is always best to have none or extremely little mechanical backlash. Even with glass scale positioning feedback and some mechanical backlash on the axis, the controller can not do a perfect interpolation without a 'hickup' during an axis direction reversal. Best seen at the 90 degree points when interpolating a circle. Of course, this also depends on your accuracy expectation. If 0.01mm is good enough for you than you might not ever run into such a problem. But when you get down to 0.001mm things are quite touchy.

You may want to investigate where the backlash occurs or what contributes to it.

Lead screw, definitely
Ball screw, perhaps
Motor coupling, likely
twisting shafts, perhaps

Then improve things to get rid or at least reduce some backlash were possible.

Then you can enjoy the full advantages of precise glass scale feedback systems on a CNC controlled machine.

just my 2 cent worth

We are using linear scales on our mill
http://wiki.linuxcnc.org/cgi-bin/emc...akeAndRussells
and soon to have a detailed write up at http://www.cnczone.com/forums/showthread.php?t=109407

Anyway, we have done our best to eliminate the mechanical backlash in the system but theres still enough there that it poses a bit of a problem. The issue is when you go through a reversal and you wind through the backlash in the system the motor builds up speed such that when it comes out of the backlash its moving fast and overshoots.

We are hoping to add encoders to the system as well for velocity feedback into the system so we wind up with the best of both worlds.

Linear glass scales are generally more precise. However...

Uni-directional positioning, distortion elimination through correct method, and single point boring tools are recommended to ensure the most exact positional relationship between features. A CMM check will confirm this...

DP

Under the right conditions you can use the glass scale for CNC. I have a Galil DMC1750 controller card in a PC that I am using to drive servo motors installed on a Bridgeport knee mill. The DMC1750 controller has at least two modes of operation, it can send step and direction commands to the motor driver cards, or it can output a +/- 15 volt DC signal to the motor driver (also called amplifier) card. The step and direction command is normally used with stepper motors, and does not have any provision for feedback of position. Since I have servo motors instead of stepper motors, and the motors have an encoder on the shaft, initially I was using a Gecko motor driver. The Gecko driver receives step and direction commands just like a servo motor, and supplies a DC voltage to the motor, either positive or negative depending on desired direction. The motor moves and the motor encoder feeds back a shaft position to the Gecko driver. When the Gecko driver senses the move is complete it rapidly switches polarity of voltage on the motor, essentially causing it to lock in position. I wanted to make use of the glass scale encoders already installed for my DRO, so I purchased new motor drivers made by Copley (300 series). These amplifiers have an option of receiving an analog DC signal in the range of +/- 15 volts. They drive the motor CW for one polarity and CCW for the opposite polarity, and vary the motor speed proportionally to the magnitude of the voltage. The Galil DMC1750 controller can be configured to produce this analog output voltage instead of step and direction signals, and has provision for an encoder input to sense the position. This encoder may be located anywhere desired, on the motor shaft, or even on the table being moved, with some limitations. Since I was still using my old Acme lead screws with about 0.015" of backlash, when I connected the glass scale encoders to the DMC1750 it worked, but the motors were impossible to "tune" due to the backlash. The motors oscillated back and forth trying to compensate for the backlash. So for the moment I am using the motor encoders and just living with the backlash. However I have ordered a ball screw conversion kit that should arrive in a few days, and hope by eliminating the backlash that I can use the glass scale encoders to provide feedback. One huge advantage of this method is that the ball screw does not need to be super accurate (ground, $$$) provided the whole system is tight enough to prevent oscillation. Any accuracy issues in the ball screw will be compensated for by sensing the actual table position with the glass scale. So the short answer is yes, provided you have a tight machine with no backlash and a controller that accepts encoder feedback, and that allows you to tune the PID loop. For what it is worth, I added an opto-isolated comparator to the DRO to be able to use the same encoders to drive the DRO and to send the encoder signal on to the DMC1750. We will see if it works.

bbutcher,
No matter how close your screws fit there will be backlash and the real problem is a response lag time which makes the loop unstable.
(lag time plus fast controller = oscillation, the controller commands moves faster than the system can respond)
Your control is capable of running a "dual loop".
It pulls velocity information from the motor encoders and closes the position loop around the linear scales.
This works nice but is a bit touchy to get tuned right.
Check out "dual loop" on the Galil site.
This is much like the way a Fanuc handles it but in this case the velocity loop is closed in the amplifier section and the main control closes the position loop.
Bob

linuxcnc can do 'dual loops' too...

the very first time the group tested it..

Just a Blog

sam

OK, good advice. I was a bit bothered about whether the tuning would have to slow the system down too much. I did read a bit about "continuous dual loop" and as I understand it uses the motor encoder to drive the KD (derivative term) of the PID loop and the load encoder (glass scale in this case) to drive the KP (proportional term) and the KI (integral term) of the PID loop. I will try this technique since it should produce a more stable system.

I agree 100% with KCT and CarbideBob. I CNCed a Hurco SM1 Milling Machine:
X-Axis = Allen Bradley Y-2012-2 (0.51 KW, 4500 RPM, 1.4 Nm), Neugart 25:1 Gear Reducer, 25mm diameter by 25mm lead Bosch Ball Screw
Y-Axis = Tamagawa TBi II (0.75 KW, 3000 RPM, 2.39 Nm), Parker 12:1 Gear Reducer, 25mm diameter by 25mm lead Bosch Ball Screw
Z-Axis = Allen Bradley Y-2012-2 (0.51 KW, 4500 RPM, 1.4 Nm), Alpha 50:1 Straight to drive shaft. All drives are Allen Bradley 1398-DDM-009 and Galil DMC-2143 Motion Controller.

Everything is real tight (I can't feel backlash) but I still get 4 flat corners when cutting small circles. I guess it when the motor pauses / reverses on top of metal torsion and undetectable backlash.

I installed a Jenix 1μm scale at the Y-Axis and connect it to Galil DMC-2143 auxiliary feedback. I had to decrease the acceleration a little (stopping too hard). I didn't try to cut a circle yet because the same wasn't achieved with the X-Axis.

Y-Axis @ 8000 cts/motor rev, 3840 cts/mm which is 3.84 times greater than 1μm resolution of the Jenix scale - The scale should have less linear resolution than the motor encoder - I don't how much... But it worked good (0.01mm Accuracy at 200mm travel), but I am very curious to cut a small circles and whatch the results.

X-Axis @ 8000 cts/motor rev, 8000 cts/mm which is 40 times greater than 5μm resolution of the Acu-Rite scale. It did not work: While jogging the table jerk and faulted the drive. I just think that 40 times resolution of the scale is way too much.

I think if I get a 1μm scale and change the gear box to a 10:1 x 1000W motor = 3.2 times greater than scale resolution... Please see attached table and video:


https://www.youtube.com/watch?v=EeMsZmbrl98


Any help / suggestions would be appreciated!

Thank you!