Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: watsonstudios on December 17, 2011, 12:07:16 AM
-
I just upgraded my router to a 5x10' R&P drive and now I seem to be having accuracy issues. I've been playing with the settings in Mach, slowing velocity & acceleration and also bumping up step & direction pulse width but I can't seem to get my accuracy back. I think I'm missing steps but I have no idea where. I'm going to rule out the possibility of losing steps going from the computer to the breakout board to the drivers because it worked fine when I was running my smaller screw drive system.
I'm basically cutting 4 parallel lines that are 4 inches apart. I'm getting inconsistency whether I cut them in X or in Y. I've attached a quick diagram of what is happening.
In the attached diagram, The blue lines are what it's supposed to look like and the red lines are what I'm getting. You can see how it's almost random where the errors lie. 2 or 3 lines can be right on and then one may be slightly off. Any ideas about what could be going wrong? Everything worked great on my smaller, screw drive router and accuracy was fine.
Also what is happening is when I cut out a profile, the position where the bit is on the exit cut is slightly off from the position of the entry or start of the cut. This is telling me that steps are being lost in the process of cutting the profile and not ending up where it started.
Now could there be an issue of noise or interference in the cables? I did just replace my Y & Z motor (http://[img])[/img]cables with larger 18ga 4 conductor shielded wire but have not done so yet on my X & A motors, which still have thinner 24ga unshielded wires. Could this make a difference? Also, could it make a difference if the cable lengths are different? I'm looking for all possibilities here.
Any thoughts would be greatly appreciated. Thanks,
Jon
-
Would recommend that you test and confirm your steps per unit. Don't have the router on when you do the testing.
Five and 10 feet are long distances and the distance testing is only as good as the measuring device you are using.
Test each axis and one axis at a time. Set the steps based on the longest distance, then check for linearity of the axis,
ie; use 5' / set the steps per / then check at say every 6" of movement and see if you are over or under and note what your findings.
Now move the same distances but with the router on as that would show if noise is involved. Do the above at a reduced velocity and at
your max.
RICH
-
Hey Rich, I don't think it is my steps per unit. I've already played with that and determined that the deviation is more random and not compounded over distance. I actually cut a 24" x 48" profile and the measurements were right on. I put my vel to 100 and accel to 2 and cut 13 lines, two inches apart along Y. Measured between them and there was a .02 variation at most. I put the ruler on the first line, the second & third were slightly off from the first but exact in relation to each other, the the next 2 or 3 were slightly off the the next was on and the next 2 were off, etc. very random in the variation of distance between all the lines. I will check with the router off but I'm keeping the router cable away from the motor cables and even have it in a separate power outlet. I may try to put it on a separate circuit just to be sure. The thing is, when I ran the exact same electronics & motors on my smaller 2" x 3" table, I didn't have any issues so everything is still the same except now I'm connected to an R&P and the axis' are triple in size. It sounds like noise to me because of the very subtle errors in accuracy (.02) I just don't know id its coming from the router. Any other thoughts? Thanks,
Jon
-
IF it seems to lose position, then gain it back, my guess would be something mechanical.
-
the deviation is more random and not compounded over distance
The axis can gain, lose, or deviate but it is just a matter of degree relative to the total ( a perfect linear plot of travel) and will be consistant.
If you have / know that info to begin with, then when the router is used, and if it is no longer consistant, it is being caused by noise or even something else. Just want to completely eliminate any mechnical issues to begin with.
RICH
-
Ok, I will stick a sharpie in the collet and draw some lines with the router off and see if there is any difference. What other mechanical issues could there be? Keep in mind that I have checked & tightened all the motor shaft pulleys, tightened the belts, even tried to move each axis by hand to see if there was any play or back lash issues in the rack & pinion and all of these checked out ok. What other mechanical things should I look at? My motors are not whining or stalling either.
Could the difference in wiring be a factor? Having the Y & Z motors using 18ga shielded wire and the X using 22 or 24ga un-shielded wire? I'm going to switch the X today over to the heavier shielded wire just in case.
Thanks,
Jon
-
What you need to understand with your conversion IS with the R&P it normally requires MORE motor torque than you needed with a Screw. This is because with the R&P you cannot get the reduction you had with a Screw. Yes you get more speed BUT that comes at a price of needing more torque to drive it.
The R&P i have seen either use belt reduction(which can limit top speed) or LARGE motors to drive them.
Just a thought, (;-) TP
-
Yes TP, I have the CNC Router Parts R&P system, which has a 3:1 belt reduction. I'm also using 495oz NEMA 23 motors which should be sufficient. Again, I haven't seen any stalling or heard anything funny yet with the motors. I'm getting pretty good results keeping my velocity at 100ipm and accel at 2. I ran my 2" line test again and it seems to be getting much better. I actually have my power supplies for the motors & drivers plugged into their own circuit, separate from the router, computer and dust system.
-
How do the steps/ per unit compare from the old setup to the new setup? Close? not so close?
(;-) TP
-
Steps per unit on the R&P is ~1388 and the steps for the old screw drive was 8000. My Z is still a screw drive and it's at 8000. They were all ACME 1/2" 2 start screws.
My drivers are currently set a 1/8 microstep. Do you think I can change them to higher or lower microstepping to get better results? Will that help?
Thanks,
Jon
-
Well you went from 8000 to 1388 so that means the R&P is geared a lot higher (faster) and it will require a lot more torque to drive it at the same acceleration rate as you had and to help keep it from loosing steps. You are already down very low on acceleration vs velocity
It will depend on how much reserve torque you have to work with from the original design criteria.
Changing microsteps I don't believe will help you out any.
Just a thought, (;-) TP
-
So let's say it has nothing to do with torque because I'm running it so slow. Should I even go lower with the acceleration? Maybe 1? I was also told to put it in "exact stop" mode in the Constant Velocity settings, was that correct? Does this setting make a difference?
If it's not the torque (because it seems to be running fine and not stalling at all) What else can I check? I need to fix this without having to buy new components. I have Keling KL 4030 drivers and they have always been reliable and accurate. I have 2 24v 4.5 amp power supplies, one for the Y&Z motors and one for the X&A motors. Do I need more amps? Better PSU's?
-
"If it's not the torque (because it seems to be running fine and not stalling at all"
Could you not be running near the torque limit of the motors, and this is causing you to loose steps and because it is not loosing alot of steps, stalling or grinding, you think it is n ot an issue with lack of torque?
-
How is the pinion engaged? Spring or air? On a heavy gantry with rapid changes of direction, a spring engaged pinion can bounce a little, which looks like lost steps.
-
It's spring engaged. It has a lot of tension and I've watched the gear and not even close to coming up off the rack. That's definitely not the issue here. Also, the changes in direction are not even close to being rapid or quick at 100ipm velocity and 2 acceleration.
-
You went from 8000 steps/inch to 1388 steps/inch, which means you will now need almost 6X the torque to get the same performance! It is incredibly unlikely you had enough reserve torque before to be able to handle this, even at low speed. Seems painfully obvious to me you need larger motors, which will likely also require better drivers.
Regards,
Ray L.
-
Yes TP, I have the CNC Router Parts R&P system, which has a 3:1 belt reduction. I'm also using 495oz NEMA 23 motors which should be sufficient.
How did you determine this was sufficient? How much weight are you trying to move? How smoothly does it move. What kind of drives, and what power supply voltage?
Most people use the 381oz motors with a G540, which will usually outperform the 495's by a good margin, due to their lower inductance.
Imo, Nema 23 motors are too small for a 5x10 machine, and I also don't think the 3:1 reduction is enough for a heavier machine.
For the 4x6 router I'm building, I've calculated that I'll need 570oz motors and 4:1 reduction to get the performance I want. (150-200 accel up to 500ipm)
-
You went from 8000 steps/inch to 1388 steps/inch, which means you will now need almost 6X the torque to get the same performance!
Ray, most people with ~4x4 machines find that they actually get 2-3x better performance from the same motors when switching from acme to R&P.
You are correct about needing 6 times more power, but I believe that the increased efficiency of r&p and the lower motor rpm can actually give you more power. But you still need to have enough power to do the job
-
Gerry, I'm using Keling KL 4030 drivers with 2 24v 4.5a power supplies. One for the X&A motors and one for the Y&Z motors. I looked and I have the drivers set to 3a each. My Y axis moves very smooth and probably weighs 15 pounds or so. Of course my gantry probably weighs 50 -75 pounds but still moves pretty smooth. Any advice would be appreciated. Do you think my problem is with the overall power & torque? I can get better results moving very slow, 50-75ipm and low acceleration. It will be a few weeks before I can upgrade the motors and/or drivers & power supplies. But I do need to make this work until then so I'm thinking I'm going to have to sacrifice speed for accuracy for the time being.
Do you think I'll get my accuracy back with better power & torque?
Thanks.
-
Do you think I'll get my accuracy back with better power & torque?
I honestly can't answer that, but I know those motors are very weak running on 24V. Ideally, they need about 100+Volts for maximum performance.
Your problem could actually be caused by 100 different things. The first thing I would probably try, would be a different PC.
-
It's the same PC that I was using when I had my 3' x 4' screw drive system and never had any problems, Accuracy was perfect. Basically, I'm using the same PC, motors, drivers, power supplies as I was on my smaller system so the problem should be mechanical I would think. I'm thinking it must be the motors, drivers, or psu's. I've already checked for slippage, backlash and freedom of movement and all those check out ok.
Again, even though I may be under powered, I have not seen or heard any stalling, even running at 400ipm - something is happening on a much smaller scale. I'm still getting measurements that are .01 to .02 off when cutting parallel lines but the errors are smaller at much slower speeds.
Any other thoughts? Thanks,
Jon
-
BUt yourare geared much different(requiring more torque) AND I bet the gantry is much heavier leaving more inertial weight to deal with(requiring more torque) also with the higher gearing your resolution per step has changed to the worse.
If you can change the gearing to about 5-6:1 instead of 3. That will give more torque BUT it also slows down max velocity.
My guess is you are loosing mini steps at each change of direction. Try setting the Accel to 1 but then you face the delima of slow accelleration and CV rounding. So you may have to run exact stop mode.
Just some thoughts, (;-) TP
-
Yes, I have already switched to exact stop mode and acceleration to 1, I have my velocity set at 75 and in Sherline mode. Accuracy is noticeably better at these slower speeds but not perfect yet.
-
With accel at 1, I doubt it's the motors. Most CNC Router parts r&p users are using 30-40 accel with 380oz motors.
Try lowering the velocity to 40-50 as a last test, as the voltage could still be the issue.
-
I will try an even lower velocity and raise the accel. If it is the voltage, would getting a new power supply help? Maybe 48v? My drivers only allow 3amps max, is that enough? Thanks.
-
Make sure you check the max rating of the drives.
(;-) TP
-
Your driving 2 motors at 3 amps each using a power supply which is 24V @4.5a. I would recommend that the power supply have the current capability for both motors which would be 6 + and an increase in voltage also. That will increase the motors preformance. Without a motor
curve it is only a guess on available torque and rpm unless someone has experience with that particular motor. You have not noted how the motors are wired and that can also make a difference.
FWIW,
RICH
-
Dont know if its relavant, but I have had issues with the 4030`s on highly detailed small parts. (Jewlery waxes)
They would just act stupid as in what you are describing. On simple profile shapes they would work fine, it was the small very small 3d movements that would give me grief. I replaced them with the 5056`s and all was fine.
-
Well I'm going to try out some Nema 23 380's and Gecko 540 with a 48v 12.5a power supply. We'll see if my problem goes away. We can't find anything mechanically wrong at this point....
-
For a 5x10 table I would consider Nema 34s as a better solution.
Just a thought, (;-) TP