Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: davegsc on August 06, 2008, 02:18:58 PM
-
Hi,
I'm hoping that someone more experienced than I am can suggest where to look to solve this problem.
I get a consistent creep of 0.016 on my +Y axis each time the following routine runs. That is, each time I go back to X0Y0, it is 0.016" from X0Y0, in the +Y direction. No problem with the X axis.
I've searched the archives, and I've adjusted the # of pulses on the Y axis, tried variations in Velocity and Accelaeration - haven't had any improvement so far.
I'm running Mach3, the Mini_Combo (steppers) from CandCNC.com, Gecko 203V's.
M3
G90
M98 P1000 L100
M30
O1000
G1 Z-0.125F20
G0 Z0.25
G1 X4Y4F120
G0 Z-0.125F20
G1 X8F120
G1 Y8
G1 X4
G1 Y4
G0 Z0.25
G0 X0Y0
M99
Any ideas?
Thanks!
Dave
Saltspring
-
Presume you are meaning it creeps each time you run the code and it is cumulative?
Try swapping drives around and see if it follows the drive.
Hood
-
Thanks for the quick reply, Hood.
Yes, the error is accumulative - creeps each time the program is run. .
I switched drives as you suggested, and the error remains on Y. So that points to..., the stepper motor? Or the software configuration?
Thanks.
Dave
Saltspring Island
-
Hi, Dave
Some drives are sensitive to Step Dir pulses, In ports & pins, Motor Outputs, All the Step & Dir Low Actives need to be in the same State (X or Checked).
Reversing an Axis direction should be done in Homes & Limits (Reversed coll- um).
Have you tried 5 for Step/Direction Pulses in Motor Tuning or Sherline 1/2 step mode in Ports & Pins.
Things to Check, Chip
-
If chips sugestions dont work I think I would swap the pins within Mach for X and Y and see if the problem follow, you could also physically swap the wires from the breakout to the drives.
Hood
-
Thanks for the suggestions, Chip.
I tried them each one of them in turn, then all together..., restarting the program and reloading the G-code each time - still no difference. No change at all in fact, except when I set my pins for X & Y - yours pins are set to 3,2, with Y, Z, and A being at 4,5..., 6,7..., 8,9. The the Y doesn't move at all. Any reason the X pins are reversed in the screen capture?
Another thing that might mean something to someone smarter than me, is that my pulses per unit for X are different than they are for Y. I'm guessing this has to do with my belt drives being different lengths, but I'm open to suggestions.
I appreciate the suggestions - please keep them coming! At this point, this machine purrs along beautifully, (much nicer than when I had it on ACM screws ). But isn't much use to me if it can't find it's way home after even small cuts!
Thanks again.
Dave
Saltspring
-
Hi Hood,
I swapped the pins for X and Y as you suggested - the error is still on the Y axis (goes 0.016" +Y from X0Y.) Not sure what that signifies. X axis is unaffected.
I don'think it matters, but I have the same motors and same speed reduction on both X and Y, but the number of pulses for X is 4750, Y is 4100. It should still go back to the original X0Y0, but it seems odd that they wouldn't be the same.
Thanks for your help!
Dave
Saltspring
-
Your pulses per depends entierley on each individual axis, motor, pitch of screw, micro stepping etc etc. If all your axis have the same then they will be the same, if not then they will be different.
What is the pitch of the ballscrews, the gearing, the microstepping (probably 10 with Geckos but not sure on 203v) on each axis.
Hood
-
They should be the same if everything hardware wise is the same, tell me the things I asked above and I will tell you what they should be.
Hood
-
It's actually a belt drive CNC 45" X 18". It was originally driven by ACME screws but I was getting too much whip, which cause vibration. The belts run it very smoothly.
The speed reduction is 17:1 using cogged timing pulleys. The belt drive version did run accurately with Omikron drives, but kept cooking them - that's why I went to the Geckos. Same motors that I used with the Ohmikrons.
I believe the Gecko 203V's are 10 microsteps, but I don't actually know - it isn't listed on their website like it is with the 201 and 202.
Thanks!
Dave
-
What is the pitch of the belt/pulleys? Or if you can tell me how far it travels with 1 turn of the motor?
Hood
-
Pitch is 0.375(L)
Thanks
Dave
Saltspring
-
Correction on the reduction - I have three stages:
10 teeth to 24 teeth
10 teeth to 36 teeth
10 teeth to 24 teeth
Then the long belts that drive the gantry
Sorry about that.
Dave
Saltspring
-
Dave,
Maybe a little more detail on what you are calling creep, if you start at Y0 and go somewhere Y+ then come back to Y0, and it is not in the exact same start point, that could be just simple backlash.
If you go from Y0 to Y+ then back to Y0, and do that 20 times and it comes out .016 X 20 off from Y0 would mean something different. If exactly .016 X 20 off you have excellent repeatability and it is probably something to do with pulse triggering.
If it is more random it could be binding causing you to miss steps only in one direction (going Y-) or noise on your system causing false stepping in the Y+ direction.
Arnie
-
Hi Arnie,
It's your second paragraph - it is very repeatable. If I run the test program 20 time, Y+ is out by 0.16 X 20. If I run the program 100 times, it's out 0.16 X 100. In each case, , the DRO is reporting Y0. I'm not saying that there is NO backlash, but it's not noticable.
Thanks
Dave
Saltspring
-
Dave
afraid I dont really follow on the reduction, is this all on one axis or is it three different axis?
Do you have a pic of your setup that you could attach to clear things in my mind
Hood
-
Hi Hood,
Sorry - my explanation is lacking badly. I've attached a photo of the speed reducer. Please excuse how it all looks - I had no faith in the beginning that I could do this, so I built it out of MDF and ACME screws..., to prove out the concept. It worked, but showed up the weak points, so now I'm converting it to 1/4" aluminum and belts. I'm an ex-millwright, and metal makes a whole lot more sense to me.
Yes, there are three axles for each speed reducer, the final one being the shaft that drives the X axis belt on either side of the table. Each axle runs on sealed bearings, First one is tensioned by sliding the motor mount, second one is tension with the idler wheel, third one is tensioned by sliding the Reductioon assembly. The long drive belts are tensioned with sliders too (see lower left). The Y axis is reduced in exactly the same manner.
Hope this helps.
Thanks again for your time.
Dave
Saltspring Island.
-
That is brilliant :)
Its late here and my brain isnt working too well at this time of night so working out how far it will move with one rev is beyond me. I will look again in the morning and things will be clearer.
Presuming the white belt is clamped to the table bearings?
Hood
-
Well, thank you for that kind word but around here it's known more as the "Rube Goldberg" approach than it is "brilliant". As in..., "if you can go expensive, go rustic!" :-) However, if you liked that one, here are a few more views.
I've already converted the Z Axis to aluminum, and it still uses an ACME screw. I think I'll leave that as it is..
The next picture is of the Y-Axis Reduce (short axis). It's the same as the X-Axis Reducer (long Axis), except that the final driver pulley is behind the reducer frame.
Next is the where the long drive belt mounts to the gantry. The reason that it is so far away from the table, is because I decided to widen the gantry when I make it out of aluminum, to give me 26 inches of cutting width. At that time, the belts will be attached directly to the gantry, without the outboard mount.
The table is stationary, because I wanted to be able to cwork on twelve foot boards. According to my plan (I've never done it), I should be able to make four feet of carvings, move the board along the X Axis, re-register it, then carve the next section. Theoretically, there is no limit to the length of board I can carve, other than the length of the wood I can find.
As I said, it's pretty crude at this stage - I'd never seen a CNC machine before I built this one, so there are many things that I'll do differently on the next iteration.
Thanks again fro your help, Hood - I look forward to seeing what you come up with tomorrow.
Thanks to everyone else, as well.
Dave
Saltspring Island
-
Hi, Dave
Looks to me you have 5 stages 10 to 36, (36 to 10), 10 to 24, (24 to 10), 10 to 24, The first 4 cancle each other out.
Stepper 1 turn = 2000 * 3.6 / 3.6 * 2.4 / 2.4 * 2.4 = 4800 Steps Per. Inch, So you only have a 2.4 ratio at the end.
If both axises have the same ratio then 4800 steps per is the number.
You really need to change the ratio around to get the step count up
You Will\Can get DRO not quite showing the full move on the display with under 10,000 Steps Per Value.
Is your error on the DRO or Actual measured loss at the cutter.?
Your pins should remain as you have them , Only addressing the Low Actives and Pulse width settings in my post.
Some drives are sensitive to "Step Dir pulses", "All" the Step & Dir "Low Actives" need to be in the same State ether (X or Checked).
If you need to "Revers" an Axis direction, It should be done in "Homes & Limits" (Reversed coll- um).
Missed the drive type "Gecko's", Pulse width Of 2 or 3 should be Fine.(This adjustment just gives your drives more time to see Mach's Pulses).
Thanks, Chip
-
Hi Chip,
I may be misunderstanding you, but all my ratios are small pulley to large pulley, so there wouldn't be any cancelling out. I have: 10 teeth to 36 teeth (3.6 : 1), 10 teeth to 24 teeth (2.4 : 1), and 10 teeth to 24 teeth (2.4 : 1), so I have 3.6* 2.4*2.4 = 20.736 : 1, correct? What would I change around?
The error is at the cutter tip. The DRO reads just as it should (X0.00 Y0.00).
I have the pins set exactly as you described in your previous post. Tom Caudle frmo Cand CNC.com says that the Step needs to be positive (Red "X"), in order to get proper accuracy, but if I do that, the motors jsut groan and dont move any discernible amount. Putting a Green checkmark on them makes the motors sing along very smoothly. Just with this extremely frustrating creeping on Y. (It starts at X0.00 Y0.00, goes out and does some movements, and when it come back, the DRO reads X0.00 Y0.00, but the cutter bit is offset by 0.016".) It's entirely repeatable - if I run the program 100 times, the offset is 0.016" X 100, everytime.
Let me know if I'm misunderstanding your post.
Thanks
Dave
Saltspring Island
-
Chip: BTW, I'm using Gecko 303V's. I set the Pulse Width at each option from 1 to 5.
Thanks
Dave
-
2000 steps for motor per rev
20.736 revs of motor to 1 output rev
41472 steps for 1 output rev
1 output rev moves 24 x 0.375 inch = 9inch
41472 steps = 9 inch so divide by 9 to get per inch =4608 steps per inch.
Think thats right ::)
Hood
-
Even is the traverse is set at the right (or wrong) steps per unit, this cannot be the reason for the progressive error because the steps per could be anything - but it goes one way, and comes back so they cancel each other out.
Check you are on Absolute Stop, and bring down your speed and acceleration substantially on your motor tuning, and try again and see if the error varies.
This could be backlash, but I cannot see how if you do not touch anything. On the first move you would be out (up and back - only one change of direction) - but the next move you would have two changes of direction so this would subsequently cancel out any further backlash.
You are loosing steps somewhere and the only reason I can think of is - your gearing, although substantial, is still quite high. and the motor is struggling as it sets off, to accelerate the whole contraption up to speed. One full motor step is nearly two thousanth of an inch - so you don't need to loose many of them each way to loose your distance.
-
Dave,
The repeatablity that you describe leads me to suspect that it is a triggering error of some sort. When your drive changes direction it will change the dir pin output. It is almost like when you change directions it may be missing a step pulse, maybe triggering on the trailing edge instead of the leading edge? Just a lot of guess work on my part. If your steps are long enough you might be able to determine this by stepping one pulse in Y+, with a dial indicator to monitor the movement, then step back 1 pulse.
One other thing to consider, One of my 4 cnc machines is a recently aquired engraving table, it has the same problem you describe but it is on all axis. I get a cumulative error after multiple moves. On my machine the problem is the computer is not fast enough, I think it is a 500mhz machine. I can improve the error by turning off the toolpath display update, but haven't totally eliminated it.
Hope this is helpful
Arnie
-
Thanks Hood! That's going to help on a smaller CNC conversion I'm planning for my lathe. I appreciate that!
Thanks!
Dave
Saltspring Island
-
Just to repeat - are you on Constant Velocity or Absolute Stop.
Try running on Absolute stop.
-
Hi,
Thanks for that - I am using Absolute Stop. It's slower, but I found it to be far more accurate on the corners.
Thnaks
Dave
Saltspring Island
-
Hi, Dave
Ouch !!, When your Brain, Eye's, Keyboard and Calculator aren't in gear = Bad Numbers, Bad Day in "Bed-Rock".
You stated earlier that setup's per for X = 4750 and Y = 4100.
If there geared the same then the steps per should be the Same in the End, As "Jim" stated, Your X, Y axis may have Backlash,"Belt Sage" Issues.
None of that is causing your Lost Step's though.
OK, "Direction" and "Step" "Low Actives" Are all set to the "Same State", In Config, Port's & Pin's, Motor Outputs.
I think with Gecko's it's "Checked" not sure.(Just to be different)
Also this "Checked" State may just be needed to have the BOB work It's OPTO's Properly.
Now your BOB may have a way to set the Output's State to your Gecko's
Do you have a link and part # for your BOB.?
I've also seen on some "Brake-Out-Board's",With Bad OPTO Isolator's (Slow response Time issues). Did you try the Sherline 1/2 step Mode.?
It may be time to Wire your Gecko's Direct to the printer port, Try a different computer, Get a bottle of Whiskey.......
Chip
-
Hi Chip,
The Gecko's only run when the "Step Low Active" is set to Green checkmark. Both tom Caulde (the manufacturer - very knowledgable), amd Dan Mausch (another very knowledgable CNC guru from way back), both say that they need to have the Red "X". When I do that though, I get a low groaning from the motors, with very, very little movement. Put them back, and the run silky smooth. I have no idea why it doesn't work as it was designed.
The BOB is CandCNC's Mini-IO combo board. They're at: http://www.candcnc.com/ The website has changed to some new products, but the The BOB is under "Quick Pick and Buy". I also have their MTA Mass Termination Card, two EZPlug Gecko dual Stepper Cards, their Mini-IO Remote Relay, and Mini-IO_Port2. The manuals for all of these can be read from the menu on the right of the website.
I tihnk the opto-isolators must be OK - the board runs fine. In fact everything does..., : motors are silky smooth, there is no excessive heat, X and Z axes do exactly what they're told, etc. The only problem is that the router tip increments on the +Y direction for every round of my test program. (traces a 4" X 4" square hole, in Absolute mode.
You might be right - it may be time to go directly from the BoB to the Geckos.
Thanks!
Dave
-
That is strange, my Geckos have run no problem either Hi or Lo. They are G201 and G202 but I wouldnt have thought there to be any difference with the G203's. Maybe its your breakout that is supposed to be Active Hi?
Did you set the Steps Per to 4608? Not that it will affect the cumulative error but just curious whether your axis move the correct distance.
Hood
Hood
-
Hi, Dave
I was saying run the Gecko's direct from the computer P-port, To eliminate the BOB interaction for a test.
Another thought, Are the Y axis wires routed in the same manner as your other axises and away from high/current power wires.
Are all your steppers the same and current set resistors the same, Yellow Led's on the Geckos lighting up the same.
Have you adjusted the midband adjustment's also, per the Gecko's manual, Checked all the connections..
Hood, I've always run my Gecko's direct to P-Port in testing, Don't recall any issues ether, Hi or Low.
Thanks, Chip
-
Chip
I know for a fact that on the mill one of mine is High and the other two are low and I dont get problems, so looks like it may be the breakout that is fussy about it in this case. Why though it doesnt work on the setting that its meant to be is a mystery, maybe there is a difference with the G203's?
Hood
-
Hood
The manual stated something about not running the stepper fast enough could have issues, But x and y should be the same hear, Maybe it's the mid band adjustment ?
I need to go back and re-read that.
Chip
-
Could be, its certainly a strange one :(
Hood