Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: dray on May 28, 2012, 12:09:36 PM
-
When cutting 3d my z axis works it's way up out of the wood. I have tried lots of changes but nothing has helped so far. Help please!
-
Cut your Z axis acceleration in half.
-
I have done that and did not help. I even hung a 3 lb. weight on the z axis and no help. Tried step pulse width no help. If I run an up and down loop on the z of .1 inch it will climb .04 to .05 inches in 500 cycles of the loop.
-
Make sure your steps per unit are correct.
How is the Z axis driven ...... any backlash?
RICH
-
Hi Rich
yes the steps are correct for a given distance. New acme lead screw and delrin nut. It would seem to me that back lash would not cause an up word drift. There has been 2 times when it cut 3d reliefs just fine. I've checked the y axis is level with the table. The z lead screw is a 3/8x16 thread and the pulses per inch are set to 25600 pulses. Kernel is set at 25000. The steppers are 200 steps per rev and 435 in/oz The machine works fine on other types of cuts. The problem is only apparent when cutting 3d surfaces with lots of up and down movements in the z axis. I can plane a surface, vcarve and do pockets with out any issues. Also I am using 1/8 steps. Tried 1/4 steps no help. I'm beginning to think steppers do not do 3d well. But the 2 that cut nice says that may not be so. When this happens the lines of code are calling for the bit to be in the material but it is cutting air. Thanks for any and all suggestions.
dray
-
Hi Dray,
It sounds to me as though you may be loosing steps when the cutter enters the work, just one or two every now and again would soon mount up with 3D work.
If you have reduced your Acceleration try reducing your Velocity as well. What are your current settings for the Z axis ??
Another possible cause of this type of problem is when the cutter is not tightly clamped in the collet and moves during the course of the work (although it is usually the other way around and ends up cutting deeper).
Tweakie.
-
I think you will find the Z lead screww is TOO FINE and you are having to run the stepper FASTER than its torque curve will allow it to keep up and it looses steps.
Steppers run high torque at LOW rpm and require course feedscrews to keep the stepper inside its torque curve.
Try what tweakie suggested and cut the VEL setting in half as well. IF that does not do it cut eht settings in half again.
Also you are running close to max on the kernal speed might want to consider also going up to 35K.
(;-) TP
-
Thanks Guys
When it is running the loop test it is not cutting wood, just going up and down. I have tried all kinds of vel. & accel. settings. No change. I will raise the kernel freq. and see if that does it. If it has an effect I'll post the results Thanks
-
I echo what others have replied about the accel and velocity. Post back with the results but do the testing in logical way.
I would recomend that you set your velocity to 20 IPM with an acceleration value of 2.
- ? When you jog does the Z motor run / sound smooth?
- ? How much are you gaining after the loop test?
Swap the x or y axis drive and do the same test with the same settings. Report back on those results.
POST the following:
- Pic's of your Z axis
- The xml you are using
- Pic of the driver test graph
Now give us the following info:
- Windows system -?
- Description of your pc...on-board graphics, PP,
- Mach version-?
- Screen set your using-?
- Type of drives-?
- amp, voltage,micro step or any other settings
- How is the motor wired?
- Name / description of any BOB's
- Power supply you are using / specs
Yes , I am asking for a lot, but then would like to know what your system is composed of and would rather review what you have than say a whole bunch of things that can cuase what you are experiencing.
RICH
-
Hi Rich
hope these work this time
-
Your steps per unit for the Z are calculated correctly.
Now test the Z axis movement accurately using a dial indicator. Do this at vel setting of 10 and accel of 2 and over a larger movement
of 1". The movement should be in the Z downward direction. Bring the Z all the way up and then move it down say .5". Now move the Z in the downward direction via MDI. Record the distance for the move as shown on the indicator. Do another downward movement of 1"
and again record the indicator ( you may need to reset the indicator). If the two moves were 1" each your steps per are ok. If not
adjust the steps per until the moves are correct as indicated by the indicator.
You may want to also try the reverse of the above and see if screw has a different pitch ( small maybe but it can add up) when the Z goes up.
With the steps correct, you should now check for backlash. So just raise the Z all the way up and down again but for 1.5".
Set the indicator and jog the Z up in .001" movements. The number of jogs required to just move up .001" up is your backlash.
How much is there? You can use backlash comp to deal with, not ideal, but may serve you well.
----------------------------------------------------------------------
Do you have specific info on your Z motor?
Are your ther steppers the same?
Your power supply is 31V @ 2.42 A. That is is not adequate for your three steppers. If all the steppers are the same then the min required
amps maybe around 3 x 2.8 amps per stepper x .7 = 5.88 amps.
Don't know the basis of your motor torque rating but will say that the microstepping, reduced amperage available will reduce the the available torque and speed of the steppers.
---------------------------
Can't see how your motor is attached to the screw and how the screw is supported in the pic you posted.
Can you post another pic showing that?
Post back,
RICH
-
Hi Rich
Thanks for your responses. The z motor is direct coupled to the lead screw with out a bearing. Before changing the lead screw it had a bearing on the motor end and had the same problem. The z motor is a 290 in/oz motor which replaced a 90 in/oz motor which had the same problem. It has not improved with a stronger motor but I have 1 more 435 in/oz motor which I will install on the z tomorrow. I will check backlash as you said but have varies values with no effect. The ps is only running z during the loop test, I would like a better one but that may have to wait. Thanks again.
dray
-
DRAY,
You should properly support the lead screw with a bearing. The stepper motor has axial play and you'll have backlash along with putting
unnecessary loads into the motor. The bearing should be preloaded to minimise backlash and also needs to aligned properly.
Consider that when microstepping you reduce the motor torque probably in half, then the torque required to raise / lower the Z axis, headroom for cutting, reduced power requirements ......thus skipping is just waiting for the right condition.
Your going to have varied values when you measure because the screw is floating.
Now on to the coupling between the motor and screw.........What kind are you using?
I didn't see any recomended step or direction pulse durations for your drive and you may want to ask Mechatronics what they recommend.
Just a quick test you can do:
Mount a ring ( can be made out of paper) about 3 to 6 inches on the motor shaft along with a stationary index. With the motor energized make a fine mark. Do an up and down movement for say 1". The index should be in alignment with the mark after the moves. Doing this will isolate the motor from the screw and thus eliminate the mechanical influence of the axis. You can also calculate how many steps for a given movement and thus measure steps within reason.
So the poor mans step counter works like this:
- 6" dia x 3.14= 18.85" circumference
- 18.85" /200 ( steps per rev of motor ) x 8 ( micro steps per )= .012" of circumferential movement per sent step
So a fine mark could be around .005" and you can see or even measure steps to see if there is an electronic issue.
(BTW, one of the reasons I like motors with dual shafts)
RICH
-
Consider that when microstepping you reduce the motor torque probably in half
Have you got a citation for this RICH?
I've read this: http://www.micromo.com/microstepping-myths-and-realities.aspx (http://www.micromo.com/microstepping-myths-and-realities.aspx) but on the other hand I've read this: http://www.cnczone.com/forums/127081-post7.html (http://www.cnczone.com/forums/127081-post7.html)
Personally I like Mariss's but then he's not always right - or is he? (I got burned over his advice re: caps on switching supplies...)
Ian
-
Hi Dray,
Simple way to isolate positioning errors is to mark the motor shaft and the faceplate of the motor. With the marks lined up, note the position of the DRO (or set a zero) now run it all you want -minutes, hours, days etc... when done, command a move back to the DRO position that you noted at the beginning.
If the marks line up - EVERYTHING on the control side is good. -step rates, power supplies etc... etc.. and your problem is something mechanical from the motor shaft out.
If the marks don't line up, it's something in the control side.
By doing this you don't have to worry about lash or anything else. Your steps per unit can be completely wrong -it doesn't matter. Running this test will immediately and unequivocally isolate the issue as either a mechanical issue on the other side of the motor shaft or something in the control/electrical side.
-
Have you got a citation for this RICH?
Ian,
Nope. Just a ball park number to make a point.
RICH
-
Hi Rich
I ran the test that Automation suggested. With the stepper removed from the z axis and a pointer on the shaft and a mark on the housing. The stepper is set to drive a 16 thread acme shaft. If I have it run a 50 cycle test to move up .0625" (1 full rev. of the motor) it always has a .004" error in the up direction. So I tried the test in a down direction and it still had an error of .004" in the UP direction!! This happens at 10"/min or 40"/min. Accel. doesn't seem to effect it either. If I double the cycles of course the error doubles. Any ideas?
dray
-
Hello,
I'm new to the site and to CNC. I am using Mach 3 demo just getting my machine to run so I can start learning. I have a question that kind of relates to this topic. I need to know how to changes the Z movement + -. Just got the machine running this afternoon for the first time and the Z axis moves down when I jog Z+. The roadrunner demo Gcode has the Z set up just the opposite so when it starts the code the bit comes up when it should be going down. Do I have to changes all the Gcode lines for Z or is there a + - reversal somewhere. I can't find it.
-
Config menu then Ports and Pins then Motor Setup, change the active state of the Z axis Dir setting.
Hood
-
Hi Rich
well I ran the test with a pointer on the shaft of all the steppers and they all show an error in the direction which would be up. I also tried a different power supply, no change. Finally deleted the xml file and reconfigured every thing. No help. Would reloading mach3 be of any help?
dray
-
dray,
I have it run a 50 cycle test to move up .0625"
So you moved down and then up some distance just to remove any backlash, and then, you did 50 consecutive movements of .0625" for a total up movement of 3.125" up?
And the result was that in the total movement you were off by 0.004"?
RICh
-
Hi Rich
I'm sorry for the confusion. The test I run is to rotate the stepper back and forth 1 revalution for 50 cycles and stops at 000. But all of the axis wind up having an error of .004" always in what would be the up direction if on the z axis. The DRO reads 000 but pointer is 4 pulses from start position. Vel and accel do not change the error. If I double the length of the test the error doubles. At first I thought it was only on the z axis, but now realize the same happens on all axis. The z error really shows up in 3d cuts because of the frequent direction changes. The x and y axis are driving the machine during the test. Any help will really be appreciated
dray
-
dray,
Only one axis running, namely the Z for the testing. Later you can swap the axis by using the X and or Y drive in place of the Z.
The X & Y should not be run while doing the test because of the inadequate power supply you have. We are trying to isolate the problem from other influences. You don't do a loop back and forth / up and down, but should first see how the number of pulses / indication in one direction is as compared to the number of pulses in the opposite direction.
Pulses .....one pulse sent to the drive is equal to the resolution in terms of linear movement. So 1/25600=.000040" of "possible" linear movement. Each move per pulse can and will vary some from each other and it can take more than one pulse to change the rotor position, etc. The rotor of the motor requires 200x 8= 2000 pulses to turn it one revolution. So 1/2000 x 360 deg=0.18 degrees =10.8 min which
is the approx equivilant of .0094" of circumferential movement on a 3" disc. So if the scribe line / mark was .005" wide ( a razor blade cut) you
would see the rotor rotate by one mark.
RICH
-
i have set up a grid on my table to drill holes every 2 inches in the y and x axis when i run the mach 3 program the x cuts every 2 inches but the y only cuts every 1.5 the table display shows both axis cutting at 2 inches both my y and x axis motors are set the same in the motor tuning in the config page anyone have any ideas
-
Rich
I only test 1 axis at a time. I am sending the gcodes i am using. Hope this helps explain the tests
dray
-
KGB59 it is NOT good form to piggy back YOUR problem into someone else's post.
Please start your own post, (;-) TP
-
Hi Dray,
Have you tried changing you Step Active state (High or Low) in Ports & Pins : Motor Outputs. ??
Tweakie.
-
dray,
I only test 1 axis at a time.
Please disconnect the cables from the x &y or diasble the drives when running the test as you only want power going to the Z axis.
You are cycling back and forth, dray. That is not what i want you to do. Only move the axis in one direction and before you do that, you make a move in the z to position the rotor for the direction you are going to move,then zero the Z and have the marks alligned.
RICH
-
Rich
I will try that later today. Not sure what that will prove as what I am doing is the same as the type movements the z axis does in 3d. The error I'm seeing has nothing to do with backlash or straight line movements. The test I run only involves the stepper and it's repeatabily back to 00.
dray
-
Rich
After I have diabled x and y, moved z in the up direction, set the 00 and aligned the marks ,am I to cycle up and down or just return to 00? If Icycle up and down I have an error of not being a 00 but the DRO shows 000.
dray
-
Tweakie.cnc
Yes I've been trying anything that I could change but none of it effects the error that happens. Automation suggested a test using a pointer on the shaft of the stepper and a mark on the case. If I run a loop of say 50 cycles of back and forth for 1 revolution on any 1 axis at a time they all show an error in the up direction if that particular stepper was used on the z axis. This is run on only one axis at a time. Any suggestions? I have another bob with drivers on order just so I can eliminate that component. The error of .003"-.004" is very repeatable. It just seems to let the 000 position move ever so slightly with each rev.
dray
-
dray,
- I wanted to eliminate the power supply as the culprit by only only using it's power to run a single axis. So no power is being used by
the other drives.
- To eliminate the mechanical influences we are looking at pulses. Since you can't electronicaly count the pulses we are doing it
indirectly via a marker. It is not absolute as if you were using a counter. How well you can interpret the number of pulses is dependant on
the visual indication. I already explained the intent and you can measure form a simple disc.
- As said only move in one direction. NOT NOT back and forth or up and down. I am isolating the influence of different rotor positioning
and want the movement as simple as possible. Changing direction can require additional pulses or even skipping which you cannot account for.
Once you have a base to relate to, by that I mean, the measured results of just moving in one direction and a comparison to moving in the the other direction. Then you can change some things, only one at time,....... to see if something makes a difference.
- swap drives, pulse duration, active settings, etc.
All of this imust done in a very disciplined way.
RICH
-
Rich
Finally got some testing done hopefully in the right manner. I want to thank you for your patience. I am attaching a text file which explains the test as it was easier to write up in notepad.
dray
-
Lets do some math:
- For a 20" diameter disc, the circumference is 62.832"
- You commanded a move of 1.5" in Mach, and since your steps per inch is 25600, then idealy 1.5x25600=38400 pulses need to be sent to the stepper for that linear movement.
The DRO in Mach should have iindicated 1.5000" change in position. The mechanical indication of that movement is relative to the pointer position as compared to the
the mark on the disc. Note that we are "relating" electrical to mechanical movement and idealy you would measure the pulses sent out the pp, the pulses recieved by the
drive, the pulses sent from the drive to the stepper, and finaly use a physical indication of the rotor positioin . I have done the this using a custom input pulse counter accurate
to parts per million which provides direct comparison. You don't have that kind of equipment, nor I, but the disc measurements although not exact, are very indicative
of the TOTAL pulses sent.
- One revolution of the disc requires 200 pulses and since you are microstepping, then, 200 x 8= 1600 total steps to rotate the disc 1 revolution.
- If 38400 pulses are recieved by the stepper, then 38400 PULSES / 1600 PULSES PER rotation of the stepper rotor = 24 rotations of the disc.
- The disc actualy rotated 24 times plus some additional amount ( .060" on the disc) which is equal to .3437 degrees too much ( approx).
- One degree on the disc, 1600 pulses per 360 deg per roatation= 1600 / 360 = 4.444 pulses / degree.
The rotor turned too much, thus 4.444 pulses per degree x .3437 degrees = 1.5274 pulses to manny. Well kind of ....readon....
Now to relate some of the above info:
- Not all drives are equal and thus there is non linearity associated with them. Not all steppers are equal on thier ability to position accurately in the microstepping.
If you read the posted links by Ian in reply #13, you would understand that when a stepper recieves a pulse to make a small move it may take more than one since
since it must overcome the detent torque and internal friction. Then depending on the magnetic field and position of the rotor it can after a few pulses then overmove.
A good example of this can be done using your disc to see how manny jogs it takes to move just .0001". May take more than calculated or rotor may move more than deisred.
In fact each position will vary over the complete revolution, won't be exactly linear and you can't tell if the drive or motor has smore influence on this.
This is just when moving in one direction, now apply the same thinking to a change in direction, and then consider that linearity will be different when going the other direction.
Do you now understand why i only wanted to do one direction moves and i also wanted you to position the rotor so that it would not need to reserse direction?
So you get some data for only a single direction move, then you get get some data on changing direction, then data on the other direction. You will then have an
understanding of how things relate and then a base do do changes to see if some settings will improve things.
Now remember that the motor needs to ramp up to speed and then also ramp down to stop. An the power supply can have an impact on that and also positioning.
Do now understand why I wanted you to do this at a slow feed rate and acceleration and only one drive at time?
BTW, pulses are on and off comming out of the PP, whole pulses, so there is no such thing as sending a 1.5274 pulse.
- The smallest move you can do, if all was ideal, is 1/25600=0.000039" for a linear move. So say, two pulses too manny were sent for example,
then your axis would be out of position by 0.0008". Is that repeatable for a one direction move?
I have lots of patience and also had problems with my mill when i started doing 3d cutting.
Your the one doing the testing ..........., ;)
RICH
-
Rich
Thanks and I will try to get some more data. But right now am on a 5 day camping trip so it will have to wait till I return.
dray
-
Hi Rich
Finally got started improving the 20 inch circle segment and a more accurate .1 inch scale on the machine frame. will send a photo of setup. seems like it always over shoots in the up direction and when going back down falls short of the previous point. With a 1 inch travel seems to average .1 inch error. I put a scope on the LPT cable on pins 6 & 7 and there is only 3.5 volts available and are very jittery when running some gcode. I think I saw a way to use a 5 volt pull up on these pin just have to find it again.
dray
-
Hi Rich
I am attaching 3 file. 2 of which are photos. thanks
dray
-
Hi Rich
We finally got it solved. That 20" dia. circle really shows up the errors The following settings have stopped the errors. Only the Z axis had power to the stepper during the testing. I am sending pictures of the pointer and scale used to determine the errors.
kernel 25000 hz
pulses/inch=25601 for Z axis
vel.= 20"/min., accel.= 2, step and dir.= 1 Us
set computer to "standard pc"
turned on the Sherline 1/2 pulse
I ran a 1" test up and back to zero 20 times with out any visible error.
We than cut a 5" diameter buffalo nickle very nicely in MDF.
Thanks to everyone for your help and suggestions, and especially you Rich, Mach 3 is an exellent program and at a very reasonable price.
dray
-
Happy you got it working right.
Have fun,
RICH