The machine does not work aligned with emc

More
12 Nov 2013 00:38 #40742 by pazh
Using mesa cards 5I23 and 7I33, on a D510 motherboard with emc2.4 installed on the hard drive, I started to test the machine to mill a cylinder.

Everything works all right for about 10000 line when the spindle starts moving in the wrong direction, like it lost where the center of the cylinder should be. It still moves in circles but continuously changing the center, but in emc everything seemed Ok like it didn't get any feedback! After that I check it again and it still works wrong.

What can be the problem?
here under part 5, Mesa Configuration, it says "The 7i33 requires PDM and a PDM base frequency of 6 mHz", which I didn't pay attention at first. What will happen if 7I33 frequency wasn't set on 6 mHz?
Could It be the 7I33 card is burned? If yes, then why isn't emc detecting it absence?
And what exactly can I do to prevent such losses?

Please Log in or Create an account to join the conversation.

More
12 Nov 2013 00:48 - 12 Nov 2013 00:49 #40744 by cncbasher
if it is working fine for the first 10,000 lines , then It is part of your gcode that is wrong , and certainly not linuxcnc or your interface
how are you producing the gcode ? cam program etc ?
Last edit: 12 Nov 2013 00:49 by cncbasher.

Please Log in or Create an account to join the conversation.

More
12 Nov 2013 01:33 #40748 by pazh
thanks for replay, but no, I rechecked with the same gcode and it didn't work. btw, I use master cam.

Please Log in or Create an account to join the conversation.

More
12 Nov 2013 02:46 - 12 Nov 2013 02:47 #40750 by PCW
Can you be a bit more specific than:

It didn't work


Did it fail at the same spot? If so this suggests a CAM error

If motion is not accurate on a servo system (assuming the system is well tuned),
a possibility is problems with encoders. This can be checked by moving to a reference
point (perhaps indicated with a dial gauge) after you notice a motion error and checking
if the position has drifted.
Last edit: 12 Nov 2013 02:47 by PCW.

Please Log in or Create an account to join the conversation.

More
12 Nov 2013 04:06 #40752 by pazh
yeah, sorry.
The first time I did the test the machine worked about 10000 line of gcode then started circling for it self while in emc every thing seemed fine. After that, at next tests the spindle position was always wrong and not in the place where it should be, not even for 10 line. Of course I did the referencing, no difference, which means the machine was good at first and then something happened.

It is more than not accurate motion. It's like spindle can not remember where it was, you know?
If anything is wrong with the encoders or inside the computer, shouldn't emc give an alarm?

Please Log in or Create an account to join the conversation.

More
12 Nov 2013 04:23 #40756 by PCW
Are the errors confined to one axis?

A encoder error (missing or added counts) can show up as position drift in one or many axis.
Encoder errors may or may not be detectable by linuxcnc depending on the type of error.

I would verify that you can move each axis a long distance back and forth and
get back to the same mechanical position (that is the DRO matches your reference position)

Encoder errors can be due to a lot of things but here are some common culprits:

1. Electrical noise (especially if TTL type encoders are used).
This may be fixable by proper encoder wire shielding, or elimination
of noise at the source (VFD/axis drive line filters etc)

2. 7I33 not jumpered to match encoder type

3. Bad encoder/bad encoder wiring

4. Low 5V to encoder (check at encoder)

Another more obvious possibility (especially if 10000 lines of GCode were OK and it went bad)
are mechanical problems (loose couplings etc)
The following user(s) said Thank You: pazh

Please Log in or Create an account to join the conversation.

More
12 Nov 2013 22:08 #40771 by pazh
thanks, I'll check them

Please Log in or Create an account to join the conversation.

Moderators: piasdom
Time to create page: 0.075 seconds
Powered by Kunena Forum