Backlash - weird issue

More
19 Jul 2017 12:12 #96052 by GarthN
Hi

At first sight I seem to have classic backlash from testing a very shallow cut using a sharp v-bit. I cut lines (using the gcode below) 10 mm apart in one direction and then in reverse. The lines are all 10mm apart - but the second row is staggered from the first by over 1 mm.

But - if I move the axis 1mm in either direction, with a dial gauge on the router bracket it shows backlash of around 1/20mm (which is quite acceptable to me). If I hold the belt finger tight and move 1mm either direction, it is moving....one would say (if that was the only test) that there is no backlash.

I really don't get it. I can't detect play by trying to move the v-bit by hand that would cause an issue with the cut being only .5mm deep - its all at least fairly rigid

Is it perhaps something to do with acceleration??

Regards
Garth

%
(start at 0.000 0.000 10.000)
(prep 1)
G90
g1f300 x0.000 y0.000 z-.500
g91
(cut 1)
o100 repeat [10]
x10.000
y-10.000
x10.000
y10.000
o100 endrepeat
(prep 2)
g90
x200.000
y-20.000
g91
(cut 2)
o200 repeat [10]
x-10.000
y-10.000
x-10.000
y10.000
o200 endrepeat
(end)
M2
%

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

More
19 Jul 2017 12:28 #96054 by GarthN
Replied by GarthN on topic Backlash - weird issue
Its not the code above that is at fault - CAM generated files show the same problem eg if a rectangle is cut. One side has a misalignment where the reversal occurred. Really does look like backlash. But I can't find the reason...argh.

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

More
19 Jul 2017 13:19 #96065 by Todd Zuercher
You have to be careful lash can sneak up on you from all kinds of different directions and add up to be very substantial.

Some of those cuts above are loading the bit to opposite sides, and will show any lash in the machine. And not just lash in the driven mechanism but also lash in way bearings, and router/spindle bearings. And don't forget to include checking for Z axis slop.

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

More
19 Jul 2017 15:54 #96084 by GarthN
Replied by GarthN on topic Backlash - weird issue
I understand that actual cutting will do that - this "cut" is a .5mm deep 60 degree v groove using a carbide true pointed bit in a piece of mdf at high RPM. The lines are hardly visible (I'm trying to make them narrow). The router is a new Kress - I've
only made a few test pieces with it so far. The gantry is 1.2m wide, angle iron - not light....I need to recheck that there is no bearing slop in y or z - it is possible that its that.

Just thinking that maybe I should have the dial gauge measuring while I do a straight in and out cut - I could pull a piece of mdf across slowly to get a new cut to see the effect on the gauge. Moving y would mean the dial gauge would have to slide across the bracket

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

More
19 Jul 2017 20:52 #96100 by GarthN
Replied by GarthN on topic Backlash - weird issue
Further testing reveals - cutting many times over the same path does nothing to the cut width (it should possibly widen if backlash or flex is present). All bearings etc are readjusted and no play is detectable. Cutting the same pattern, but pressing progressively harder by hand against the spindle bracket makes no difference to the cut width (which it would if backlash OR flex was present). The cut width looks quite precise.

Hmm....perhaps it is a gcode issue. I will try a dwell before cutting the second pattern.

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

More
20 Jul 2017 09:49 #96130 by verticalperformance
Maybe an obvious question, but you haven't got backlash compensation turned on by accident in the .ini file?

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

More
20 Jul 2017 09:55 #96132 by GarthN
Replied by GarthN on topic Backlash - weird issue
Not so obvious :) I did have, but also tried with it removed. I am going to try a cut with G61.1 turned on.

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

More
20 Jul 2017 12:48 #96137 by Todd Zuercher
Try this test, setup your dial indicator as if you were measuring backlash on the axis where you are having your accuracy problems, then make a 100mm move away from it, then make 10, 10mm moves back to where you started. How far off is it from where it should be? Next try 10 steps out and one step back Were the results the same? Now repeat the test in the other direction. Are the results the same? With the backlash compensation off the results should be an error equal to your backlash. With it on (and properly adjusted) the result should be no error.

If the results are significantly different in one direction than the other or from the typical 1 or 2 step backlash testing, there is a very good chance you may be loosing steps on acceleration. If that is the case try reducing your acceleration settings.

Another possible (though unlikely) problem that could be difficult to pin point is if there is slop between the pitch of the belt and the sprocket. Belt tension would make detection of this form of backlash unreliable and inconsistent.

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

Time to create page: 0.125 seconds
Powered by Kunena Forum