Hello Guest it is August 18, 2019, 09:42:54 AM

Author Topic: Encoder Problem  (Read 6802 times)

0 Members and 1 Guest are viewing this topic.

Encoder Problem
« on: March 09, 2012, 09:16:22 AM »
Hello,
I have an encoder attached to a stepper my plan is to use it to determine if the stepper loses steps. But I am having a problem. The encoder and stepper DROs match when moving in one direction but when the machine returns back to where it started the DROs do not match. For example Starting from 0 and moving to 1” the DRO will match after the move but when the machine is returned to 0 the DROs will no longer match. This is also true if the move is in the negative direction 0 to -1" and back to 0.

The calibration  per unit is about 12000 for the encoder. Could the encoder be to high of a resolution (but then why would this happen in only the return)?

Any thoughts on the problem.

Thanks,
Paul


Offline BR549

*
  •  6,886 6,886
    • View Profile
Re: Encoder Problem
« Reply #1 on: March 09, 2012, 11:21:09 AM »
Backlash

(;-) TP

Offline ger21

*
  • *
  •  6,288 6,288
    • View Profile
    • The CNC Woodworker
Re: Encoder Problem
« Reply #2 on: March 09, 2012, 12:01:18 PM »
+1
Gerry

2010 Screenset
http://www.thecncwoodworker.com/2010.html

JointCAM Dovetail and Box Joint software
http://www.g-forcecnc.com/jointcam.html
Re: Encoder Problem
« Reply #3 on: March 09, 2012, 12:55:41 PM »
The encoder is connected directly to the back of the motor with a helical beam coupler.
This  setup up is on the Z axis where the spindle motor and assembly (greater than 300 pounds) on the ball screw. When checking with a indicator the Z axis is returning to the same location.

The problem is interesting in that it is not accurate on the second move and is not dependent when moving in the positive or negative direction.
Re: Encoder Problem
« Reply #4 on: March 09, 2012, 01:10:57 PM »
Is that 12000 pulses/1"? also what is the resolution of the motor in revolutions/inch?
If it were backlash I would expect one direction error only?
Nosmo.
Re: Encoder Problem
« Reply #5 on: March 09, 2012, 02:23:59 PM »
The encoder resolution is 12000/inch the motor is about 5200/inch. The reason I am thinking it is not related to backlash is because the the encoder is directly coupled to the motor and I am comparing the Z-axis dro value to the encoder dro value. Also, When I test moving in the positive and negative direction for the first movement the error is in the second direction.

Will be doing more testing this weekend.

Offline BR549

*
  •  6,886 6,886
    • View Profile
Re: Encoder Problem
« Reply #6 on: March 09, 2012, 02:39:27 PM »
How much error are we talking ?

(;-) TP
Re: Encoder Problem
« Reply #7 on: March 09, 2012, 03:09:23 PM »
What I was getting at is how many motor revolutions to the inch?
Ok I see it would be around 2.3?
N.

Offline BR549

*
  •  6,886 6,886
    • View Profile
Re: Encoder Problem
« Reply #8 on: March 09, 2012, 04:00:13 PM »
I think what he is alluding to IS the encoder is attached to the stepper motor shaft. AND the encoder should read what the DRO says IF the two are properly calibrated and IF the steps per rev indicate close enough to be repeatable.

(;-) TP

Offline RICH

*
  • *
  •  7,351 7,351
    • View Profile
Re: Encoder Problem
« Reply #9 on: March 09, 2012, 05:37:08 PM »
I have found through testing that the DRO is more reliable than most indicating devises. The DRO is relative to distance. Actualy if you want to be anal about it one should count the pulses sent by the computer to the drive, from the drive to the motor, and if you have a BOB the in and out of that device should also be counted. In our testing they were all compared via a calibrated multi channel counter accurate to 20 parts per million. So all the measuring done was apples to apples.

A stepper is a eltro-mechanical device and as you know they are all not alike in how they react to a pulse when you get down to the single pulse senerio. So you could for instance find that it it will take more pulses to move it a small distance and it will behave dfferently going the other way. Just test jogging at 0.0001" increments. At 5200 the resolution is
0.00019".

If one adds distance to pulses then the variables associated with the distance must also be callibrated to a known standard. Note that one must be able to decipher what is
contributing to any distance inaccuracy measured and that could be due to a number of things. ie; Backlash can be due to a number of different causes and it can be difficult to narrow it down.

In the end, my point is, one must know exactly what they are measuring and if comparing know exactly what they are comparing and to what degree of accuracy they can compare.
 
FWIW,
RICH