Hobby CNC wobble

[zedman]

Dear All,

I made my first Hobby CNC, it's made of MDF and uses steppers from Epson Printers (EM-402 like) on Parallel port breakout board from eBay with A4988 stepper drivers.
I am using trapezoidal threaded rods, it's running smooth, almost no backlash (0,01-0,03 mms).
All of the motors wired with CAT5 cable, shield is connected to the 12V ground on the PCB side.
Motors are running around 500mA (reated 1A max), the are pretty cold they only get just a tiny bit warm after an hour of work.
Stepper drivers won't get warm either.
The movement is OK in every direction, torque is enough to move the spindle in the material. I am trying to engrave an anodized aluminium plate at the depth of 0,05mm.
Using Mach3 to drive the system.

The problem is that sometimes the Y direction is likley unable to step and plays a little wobble and this way skips steps and it looses position.
Look: http://www.neptunex.hu/public/cnc/wobble.mov it should just move left smoothly.

When I move with the arrow keys it works well. There are times when it engraves the whole program, and sometimes it fails to get to the first position.
I have cheked all connections and solderings they seem OK to me.

I tried to set the speed lower, played with motor tuning (they are set to 400 with 200 acceleration (100 step==1mm), and when engraving I set FRO to just around 50-100), event set the pulse length to 15 uS in Mach, no luck.
There are Ferrits on every motor's cable near the motor.
Even tried to put the Y on other outputs of the breakout board, no luck.
Set all of the input pins to port 0 in Mach in case of getting some interference, no luck.

It ruined a couple of plates already, I am getting insane, what the hell can cause this?

thanks in advance,
Zedman

Similar Threads:

• How much of a ballscrew 'wobble' is too much?
• X3 table has a lot of wobble
• Spindle WOBBLE?
• Quill wobble
• 7X wobble

[ger21]

Try running the motors at 24V instead of 12V.

[awerby]

When you use drawer slides instead of precision linear bearings, you can't really expect the same functionality when trying to execute a project that requires smooth linear motion. There's a lot of slop in those things; of course they wobble,

[zedman]

Gerry,

I changed the power supply from a PC PSU's 12V to a new transformator, now it's with 24V and I built it into the box.
I played with it last night and it looked promising, no false movement, went smoothly.
In the morning I tried to engrave a simple program, it went well till the last step and when it started to move back to the zero position it failed again and of course it did not get back to 0,0.
I attached a pic what it had done.


Andrew,

I know drawer slides are not perfect. I know.

But I don't think this problem is related to the slides. It moves smoothly if I remove the power from the motors and try to move it with my finger.
And it does move smooothly when engraving as well, but as you can see on the video it seems to me sometimes it just reverses direction a few times or fails to step and tries it again and again since it tries to move in one direction.

As It was working for a whole program in the morning and failed only after that I tend to think it's an electronical problem which may related to warmness.
But it's just a tought yet.

Zedman

[A_Camera]

Szia Zedman!

Considering the materials you have used and the way you built that CNC I think your results are very good. The use of drawer slides has been mentioned but after I looked at your video I see an even more serious issue. It looks like your lead screw does not have any support and that your coupling and motor axis is doing all the work, it is pushing and pulling the lead screw. That solution is not very good (to say the least), you simply MUST support and fix the lead screws properly if you want any precision. You will eventually also destroy your steppers unless you fix the lead screws properly and remove the load from the motor axis. You simply can't have the lead screws hanging on the motor axis, sooner or later either the motor axis will be bent, or the ball bearings inside the motor will be destroyed and it is not a question of if that ever will happen but a question of when. I guaranty it will happen very soon unless you do do something about that at immediately and stop using the CNC until it is fixed. You should also severely increase the motor voltage and current. If the motors are rated 1A why run at half of that? Also been mentioned that the voltage should be increased to at least 24V. Both of those have impact on precision, speed and torque.

The precision you are mentioning is very high. engraving at 0.05mm demands extremely high accuracy in the whole machine and every bit of problem counts. You can't have 0.03mm backlash if you want 0.05mm precision, and I bet you that you have more than 0.03mm backlash on the Z axis, especially if you consider missing/extra steps, which I bet you that you have under load. Have you measured table flatness, back lash on all axes, axis angles, missing and/or extra steps and so on? What kind of lead nuts are you using? Also, the fact the you can easily slide the sliders by hand isn't really important because it is a no load situation. Things change when you start loading, i.e. start using the machine for real work. It might be that the the drawer slides get stuck a little now and then when under load. It is also possible that your motor axis is not properly in line and centralized with axis lead screw so the flexible coupling needs to be bent a little. Never the less, like I said, in my opinion the result you show here is OK considering the design and the materials you have used, although I would give up that accuracy requirement because I don't think you can ever get to that without some serious upgrade. Sorry.

One more thing:

You are saying about the motor that "are set to 400 with 200 acceleration." Does this mean you have a speed of 400mm/min and an acceleration of 200mm? In this case you should lower that acceleration to 40mm or even below. Too high acceleration values will definitely cause jumps and skipped steps, just like too high speed would. Can you really run that machine at 400mm/min?

[zedman]

Szia,

I myself also got surprised how accurate it is.

The other side of the leadscrew also ends with a coupler and the coupler is in a ball bearing which is sunk into the mdf.
That support were you miss or am I missing the point?

I attached a pdf about the motor, it's not exactly the same by part number, but it looks the same size and parameters.
Do you think I should use it on 1A? Even if I feel enough torque at 500mA?
Now it's on 24V.

Yes I did measure the table flatness with a Dial Indicator, it's balanced to 0,02 mms for an area of 50x50 mms that's enough for my watch dial project.
Axes are also measured with a DTI and showed 0,01-0,03 mms.

Notice that this wobble thing happens when moving to a position so no load on it at that moment.

I am using leadnuts I made, see pics. Ther are made of plastic and I made the thread into it by a tap I made of the same trapezoidal threaded rod I use.
(Put the plastic nut into the fridge for a couple of minutes and made the thread while it was cold, this way I got a tiny clearence)

Now motors are set to 300 mms / min and acceleration set to 200. BUT I allways set FRO to around or under 25.
Btw newer tried to engrave at that speed. It would kill the 0,1 mm bit asap.

I am satisfied with the accurancy of it just asking how to eliminate the wobbling problem because that completely messes up the engraving by millimeters.

thanks,
Zedman

[A_Camera]

Szia,

I myself also got surprised how accurate it is.

The other side of the leadscrew also ends with a coupler and the coupler is in a ball bearing which is sunk into the mdf.
That support were you miss or am I missing the point?

The coupler should be before the supportive ball bearing and the lead screw should be fixed to the ball bearing. Same on the other side, which is normally called the "floating side" because the lead screw on that side is not fixed to the ball bearing, it is allowed to move to relax the lead screw.

I attached a pdf about the motor, it's not exactly the same by part number, but it looks the same size and parameters.
Do you think I should use it on 1A? Even if I feel enough torque at 500mA?
Now it's on 24V.

If the motor is rated to 1A then in my opinion there is no reason to drive it at half of that. Remember that you can never have too much torque, that is never a problem, but having too little torque is definitely a problem some times. Higher voltage means higher speed which definitely helps in stepping accuracy as well, so abandoning 12V was definitely a good idea and in fact, even 24V is at the low end, but should be usable.

Yes I did measure the table flatness with a Dial Indicator, it's balanced to 0,02 mms for an area of 50x50 mms that's enough for my watch dial project.

Sorry, I was not very clear about what I meant by flatness. I meant that if you zero the Z on the surface in one corner and move around on the table is the zeroed Z still zeroed all over? Given the +/- 0.03mm you measured...

Axes are also measured with a DTI and showed 0,01-0,03 mms.

Notice that this wobble thing happens when moving to a position so no load on it at that moment.

I am using leadnuts I made, see pics. Ther are made of plastic and I made the thread into it by a tap I made of the same trapezoidal threaded rod I use.
(Put the plastic nut into the fridge for a couple of minutes and made the thread while it was cold, this way I got a tiny clearence)

Now motors are set to 300 mms / min and acceleration set to 200. BUT I allways set FRO to around or under 25.

I think that a 200 acceleration is a problem. I am just an amateur but I learned that acceleration should be about 10% of the speed, so in this case you should set it to 30. It doesn't matter that you run at under 25 FRO (why so slow BTW?) because if you have a acceleration set to 200 Mach3 will in your example start pulsing immediately from zero speed to whatever is needed for the FRO 25. If you have a lower acceleration value you might be able to run the machine at faster FRO. Now it seems that you solve the acceleration issue with slow feed rate.

Btw newer tried to engrave at that speed. It would kill the 0,1 mm bit asap.

25mm/min is pretty slow and I think that if you have a spindle which rotates fast enough you should be able to increase that significantly but you must handle the acceleration value. If the bit breaks or not depends on the quality and the type of the bit, as well as the rpm of the spindle. Of course, a jerky movement might also cause breaks. Some experimentation is needed to get the right values. Start with something softer, like plexiglass or some other type of material.

I am satisfied with the accurancy of it just asking how to eliminate the wobbling problem because that completely messes up the engraving by millimeters.

Accuracy for me means not just axis movements, but also milling accuracy. In your example, the circles are not really round, so while it is good enough considering the materials you have used, it is far from professional quality. The wobbling is definitely something you have to get rid of, but it may mean that you would need to do some major modifications before you get the circles right.

thanks,
Zedman

You're welcome.

BTW, I have also made my own lead screws and nuts but mine does not look as good as yours. I simply made them out of delrin, like you, I also used a piece of threaded rod as tap. Anyway, I am currently in the process of upgrading my CNC with proper ball screws and nuts, making a completely new Z axis and so on...

[zedman]

Sorry, I was not very clear about what I meant by flatness. I meant that if you zero the Z on the surface in one corner and move around on the table is the zeroed Z still zeroed all over? Given the +/- 0.03mm you measured...

I put the DIT in place of the spindle and yes it was between a +/- 0.03 mm boundary (inside that 50x50mm area I need).

I now set the speed to 300 mm/min with acceleration rate of 30.

Accuracy for me means not just axis movements, but also milling accuracy. In your example, the circles are not really round, so while it is good enough considering the materials you have used, it is far from professional quality. The wobbling is definitely something you have to get rid of, but it may mean that you would need to do some major modifications before you get the circles right.

The large circle is not as ugly as on the picture, it got a bit disorted by the camera. But yes the 1,5mm center hole marker is ugly.

I am getting out of ideas.

I removed the leadscrew from the motor and was able to reproduce the wobbling without anything attached to the motor. I tought it's a motor issue, maybe some dirt got into it (btw it seems impossible) and randomly it swaps it between the magnet and the coils and locks it.
I managed to get another Epson Stylus C64 for getting the motor out of it. I put it in place of the old one, but it did not help. Still doing it.

Next idea: it sould be an electrical problem. It seemed 1 program goes ok, but the next fails it maybe related to heating up of some semiconductor. I resoldered all of the corresponding solders on the Breakout Board (they were a bit lame in spite of it could have been made by automatic placer / solder). Did not help.

Next idea: maybe the PC I am using has a parallel port problem (maybe it's 3.3V?, will measure later) but changed to my ancient Thinkpad notebook it's only 600Mhz at least 10 years old. No help. Drives it well, but still produces the wobbling.

Motors are running at 600 mA now, because everything got very hot @900 mA and I won't destroy the motors, I think it should do it @600 mA too.

Would be nice to attach the Y's lines to a Logic Analizer to see if there are signals at the very moment of wobbling, unfortunately I only have a non storing scope.

Took a look on linear bearings, maybe if I can get through this problem I'll rebuild the system with nice linear rods.



thanks,
Zedman

[zedman]

I already disabled all of unused port pins in Mach.
I don't know why I just scrolled my eyes over the screen and noticed this button: Jog Enable.
Ticked it off, this way disabled jogging. Never ever tought I'd ever jog while program is running.

Now it's not wobbling.

Blah.
Zedman

[zedman]

Is it well known I should disable Jogging manually while before running a program?

thanks,
Zedman

[awerby]

No, I've never had to do that. But if it solves all your problems, then go ahead...