[abond13]

whats up guys, i recently got the G0704 conversion done and have some problems that ive been trying to figure out, i been searching for days and nothing.
My set up
keling drivers kl-8060 d
nema 23 kl23h2100-50-4b x and y axis set to 4.28A and 1/8 micro step
x axis steps per 8131.806686 (acceleration 50 velocity 50)
y axis steps per 8153.114349 (acceleration 50 velocity 50) (i see this number of steps is different than the rest)???
nema 34 kl34h295-43-8b z axis is set to 6A and 1/8 micro step
z axis steps per is 8131.298253 (acceleration 50 velocity 50)

arizona video c7 ball screw and mount kit (very nice work he dos i might add)

problems, when cutting a circular pocket using a wizard, the pocket is egg shaped in the y direction.
the hole i milled was a .310 dia. it comes out good in the x direction but the y is .330 using a 3/16" EM .1875dia
also cut a rectangular pocket 2inch in the x direction and 1inch in the y direction, x and y zeroed in the center of the part .200 deep
the y did well in the positive direction, but in the negative direction it would travel less and less negative as it made passes
leaving steps that a tiny person could walk down into the bottom of the pocket.



im wondering if steps per is wrong as i let mach calculate it. i would love to calculate myself and get the real number,maybe find where the real problem is.
also wondering if it could be in the y axis cuppler or gib. backlash is not as bad as the error im getting in the cutting tests. (no backlash comp on any axis)
could gib adjustment cause this, i have them set, but maybe they need adjustment. ?

hoping this is enouph info you need to help, let me know if you need more.
im looking for best accuracy out of this set up,at least the best a stock g0704(no scraping ways etc.) with steppers can do,thats the goal.
thanks again guys hope you can help.

[abond13]

ok so after reading thru my post. my best guess is gonna be the problem lies in the y screw or nut or bearings or mount or gibs.
but the Y axis dos travel and repeat 1.000 travels with dial indicator,so i figure i should see the problem doing that but i dont. im hoping to get some experienced opinions
before i tear the machine down. maybe its in my settings. thanks

[Kingjamez]

The problem is bigger than any error in your steps per inch calculation. The difference is 0.2% between X and Y where your error in milled dimension is 6.5%. What do the "corners" of your circles look like? I'd first look for backlash in your system. With good ball screws, the major source of backlash becomes your thrust bearings and how tightly they are squeezed together. After backlash, I'd look at steppers loosing steps. Look to make sure that the Y axis isn't binding, and that your stepper driver current is set appropriately, and the same as the X axis.

-Jim

[abond13]

well ran the Y axis back and forth 1 inch 50 times with dial indicator on it set to zero and what ever the error is it accumulates,i guess you would say,so zero moved from zero to +.005
and if i kept going i assume it would go further. is this a backlash issue?

[Kingjamez]

In my opinion, that is a red herring. Moving back and forth 50 times causing a 5 thou error won't explain your circle being 20 thou off. I'd bet that your Y axis steps per revolution is different because of whatever is causing your problem. It is exceedingly odd that your Y axis is larger than it should be. Are you using Mach? Is backlash compensation on? You can check for backlash by commanding a movement in one direction, then immediately moving the other direction by the same amount, the tool should return to its origin exactly. Or course to do this test, backlash compensation must be turned off.

Are your Y axis gibs tight?

-Jim

[abond13]

The problem is bigger than any error in your steps per inch calculation. The difference is 0.2% between X and Y where your error in milled dimension is 6.5%. What do the "corners" of your circles look like? I'd first look for backlash in your system. With good ball screws, the major source of backlash becomes your thrust bearings and how tightly they are squeezed together. After backlash, I'd look at steppers loosing steps. Look to make sure that the Y axis isn't binding, and that your stepper driver current is set appropriately, and the same as the X axis.

-Jim

first off thanks for the reply.i very much appreciate the help. the circles have steps in the negitive y direction. also the hole is tapered. large at the top and as the y steps in gets small. so with the top of the hole being oversized the y is starting out of position to begin with,they are a smooth oval shape aside from the step going down in them.(hope that answers your question about circle "corners". as backlash gos i wanna say its about .002 but it seems hard to tell really with the way its acting.

steppers loosing steps-how do i check for this?
i ran the y axis by hand to check for binding,it did not seem to be.
and how do i check for stepper current?
its real weird, the y+ side of the circle is line free looks good,the y-side of the circle,has steps going down into it. its really noticeable on the rectangular pocket.
thanks again

[onocyclone]

Try setting your steps per in x any y to 8131 and turn backlash comp off. that way both axis move at the same rate hence leaving a perfect circle. In an ideal world that is.

[abond13]

In my opinion, that is a red herring. Moving back and forth 50 times causing a 5 thou error won't explain your circle being 20 thou off. I'd bet that your Y axis steps per revolution is different because of whatever is causing your problem. It is exceedingly odd that your Y axis is larger than it should be. Are you using Mach? Is backlash compensation on? You can check for backlash by commanding a movement in one direction, then immediately moving the other direction by the same amount, the tool should return to its origin exactly. Or course to do this test, backlash compensation must be turned off.

Are your Y axis gibs tight?

-Jim

thanks for the reply, yes mach3,yes just checked backlash and surprisingly about .001 in 1inch. backlash comp i have never turned on. yes gibs are tight enouph to move the table freely back and forth
with no binding.thanks again

[Kingjamez]

You're going to need to post a photo.

-Jim

[abond13]

You're going to need to post a photo.

-Jim

ok, well while i was taking photos i got to thinking about the z axis gibs so i grabbed the head and could rock it up a little. so could it be the pressure of the tool on the negative side of the hole pushing
the tool to be move positive. so essentially the head pitching back and forth during cutting.?? i will adjust. thats my fault i adjusted the x and y gibs because i had them completely apart and the z i just
removed the head and dident pull the dovetail slide. i will adjust and get back with you. thanks again.

[abond13]

You're going to need to post a photo.

-Jim

ok here are some pics hope they come out.
now ignore the holes in the pics of the square pocket,
that is what the -y side of the square pocket looks like the other side looks good

[abond13]

ok i adjusted z axis gibs, no change. and onocyclone i did what you said and no change. still oval holes with steps. now i did observe that after the bottom of the
hole was cleared out and the hole dia. was finished. the cutter would cut on the positive side of the hole as it made its rounds,even though it should have not cut.

[Kingjamez]

Man, your finish is horrible. What tool are you using and what are your feeds / speeds ? Are you sure that everything (vise, tool, workpiece, quill, ballnuts, thrust bearings, etc) is rock solid and secure?

-Jim

[abond13]

Man, your finish is horrible. What tool are you using and what are your feeds / speeds ? Are you sure that everything (vise, tool, workpiece, quill, ballnuts, thrust bearings, etc) is rock solid and secure?

-Jim

im using a cheap,hss endmill, vise is good tool is good workpiece good, quill good. ball nuts good, now the thrust bearings im wondering about. how would you recommend adjusting them i have 2 bellville washers stacked like this () on the y axis. i thought there was to much load so i tore it down and adjusted them. same thing. i just swapped motors x and y to see if it was a motor issue,but
no same thing. man im tired,i been at it all day. i can visually see the endmill + of y 0 when it returns to x0y0 to plunge down in the hole to take another pass. thanks for all your help im
hoping to figure this out. but this thing has me beat today. thanks again keep the ideas coming and let me know how you would adjust the load on the washers for the bearings.
im thinking thats what it is but im not sure. thanks feed rate was f10. i know thats slow.

[Kingjamez]

I initially thought the bellevilles were flexing. However at only 10 IPM the forces are very low. However, I can't think of anything else to try. With that terrible terrible finish something must be loose.

Stop working in aluminum for your test. Move to plastic or wood and at start making squares (inside and outside). That will help narrow down the issue.
-Jim

[datac]

If I am correct, you earlier stated that you are seeing accumulated error ? If you lost .005 in 50 moves...... this would almost always indicate a problem with pulse width/timing/edge settings. You could prove it by running 100 moves which should exactly double your error.

I'd re-examine your direction change driver requirements.....

[abond13]

I initially thought the bellevilles were flexing. However at only 10 IPM the forces are very low. However, I can't think of anything else to try. With that terrible terrible finish something must be loose.

Stop working in aluminum for your test. Move to plastic or wood and at start making squares (inside and outside). That will help narrow down the issue.
-Jim

I did pull the bellvilles,and no change. i will try a different material to cut.

[abond13]

If I am correct, you earlier stated that you are seeing accumulated error ? If you lost .005 in 50 moves...... this would almost always indicate a problem with pulse width/timing/edge settings. You could prove it by running 100 moves which should exactly double your error.

I'd re-examine your direction change driver requirements.....

Thanks for the reply. i will run some more tests. how dos one go about re-examining my "direction change driver requirements" also, "pulse,width,timing,edge, settings" sorry noob here.
in my mind after no change with bearing adjustments, i had thoughts of bad parallel port pins or something electronic or settings. cause i can watch the cutting move more and more positive of
center with every cut.This would explain why the first job i was running was not working out. It was a rectangular plate with holes down the center of it "a jig", i set up to mill the ends to length and also put a .125 rad.on the outside corners and the bottom (negative y side) left corner, always cut more material than the top positive. Also the hole locations on the Y was always out in the same direction.
Then i run a 90deg chamfer tool around the top edge of the hole thing, there you could really see it. its like someone shifted the Y offset positive with every pass of the tool, thanks for your help
off i go to mess with it.

[abond13]

ok i tore it down,noticed there was slight binding and then a little click in the y ball nut. only in the negative travel, its perfectly smooth in the positive travel. it will bind a little then a little
bitty "click" IM STUPID i Noticed this when i first received it, but grease seemed to smooth it out so i installed it,like a dumb*ss. so i think im gonna remove the balls and investigate. first question how many balls are in a c7 nut (i will google it when im done here.) and im gonna search to see the best way to go about it. any tips on getting the balls back in there from you guys would be great. and i believe the nuts come repacked with slightly larger balls for less back lash. so i guess that would make sense, it would bind in the neg. and stop short
then travel from that position to further positive then come back neg. even less and then further positive and keep going. accumulating more and more.maybe there is something in there,
but i would think it would do it in both directions. i dont know but i will have to see whats in there. be back in a while with my findings. thanks fellas

[abond13]

alright got the balls out of the nut there are 16 per curcit as hoss said in his video and these balls measure .125. whats the purpose of the plastic red things labels 1 2 3? just wondering.