To smoothstep or not to smoothstep.

That is the question.

On long jobs (9+ hours) I am losing steps. I am running my machine slowly (50 in/min) with low acceleration. If I turn my speeds up, I am guaranteed to miss steps almost immediately, despite not feeling any binding on any axis. My machine is 4 axis, 3.5 amp motors, 12.5 amp power supply 48v etc. etc. I can jog the machine very quickly (1000 in/min on rack axis, 600 on screw axis) and it sounds alright. When I cut at 50 in/min, it sounds like garbage no matter what I adjust.

I've done driver tests and these are pics in the middle of them:

http://i.imgur.com/olpMYoc.png

http://i.imgur.com/cybQi0c.png

What does an ideal drivertest look like? Are my times too long?

The noisiness gets worse if I try running anything else. The computer is a quad core that only has windows 7 and mach 3 on it.

Would a smoothstepper help me?

Thanks for any input.

I installed my SmoothStepper over a year ago, and would NEVER run off of a raw parallel port again. Solved ALL the issues (intermittent) that I was having, lets me run at a higher kernel speed as well. HIGHLY recommended. Mine is the USB version, got it before the Ethernet version came out.

I just updated to the latest version of mach3 and it seems like my machine might work better (it actually accelerates gradually when in cv mode versus instantaneously like before) .

I suspect this will be a big improvement. I'm stuck waiting to test it until new motor brushes get here though.

Quote:

---

What does an ideal drivertest look like?

---

A single flat horizontal line. Yours is not ideal, and may be the source of your problem. I think there was a change in the driver in newer version of Mach3 that work better with multiple core CPU's.

Well, I think my problem is a combination of factors.

1st. I'm not sure the driver lpt driver I had was the latest one. I actually reinstalled windows and mach3 just to make sure everything was fine there. I turned off all power saving features, blocked all startup programs etc. The driver test still looks the same (not great), but the machine seems to behave a bit better.

2nd. changing acceleration and velocities did not eliminate the problem. It just changed the amount of time that needs to pass before it became obvious.

3rd. For the trouble axis, I have leadscrew with a delrin antibacklash nut. At one end of the leadscrew, there is a gear that connects to a gear on my stepper via a belt. I suspect that this introduces more backlash into this setup. I moved the stepper to increase tension on this belt. I still had the same problem, but the distance that the machine is off seemed to decrease for any given toolpath run.

This leads me to think that the main cause of the problem now is a small amount of backlash in this axis that gets ampified by long toolpaths. I put together a makeshift tensioner with a spare bearing, and it seems to have improved even more. I'm going to pick up a dial indicator tomorrow and measure the backlash on each axis and have mach3 compensate. Hopefully this resolves it. The machine still isn't smooth sounding at speeds around 30-70 ipm. It seems great above that. I might still look into a smoothstepper if they support backlash compensation or more simply, get a PCI parallel card.

Thanks for the help.

Quote:

---

This leads me to think that the main cause of the problem now is a small amount of backlash in this axis that gets amplified by long toolpaths.

---

Backlash is not cumulative. If you have .01 of backlash, that's all it should ever be off.

What stepper drives are you using? It sounds like a resonance issue? Or just poor output from the parallel port.

g540.

Do you think a pci parallel port might help?

It might, but my guess would be no. I'd say either try a different PC, or get an external motion controller.

Well,

I tried a different PC today. This PC had a very good driver test result. It did not, however, help with the resonance issue. It sounded just about the same at the same speeds. For the hell of it, I installed linuxcnc to see if there would be any change. It was a minor improvement at best (I could be imagining it)

I setup the original system, and reran a toolpath.
It does great until around the 2 hour mark, then gradually loses its location (off by a hundreth or two) at first. Throughout this, my motors do get very hot (though the axis with the biggest problem doesn't seem to be the hottest) Any chance high temps could be responsible for my woes? If so, does anyone know if a smoothstepper helps in this regard? My motors are 3.5 amp and I have 3.5kokm resistors across each connector. Would it behoove me to use less resistance to decrease the current going to the motors?

The motors do start making a squeaking sound after they seem to get hot.

Thanks for the help Gerry. It is really nice not being in an information vacuum with this.

Is the G540 getting hot? That can cause what you're seeing. It should have a fan blowing on it.

I haven't checked on the g540 during a run, but I don't think it is getting hot. It is in a case with the powersupply. It has a heatsink on it and a large fan moving air through the case:
http://i.imgur.com/W1IYxPh.jpg

(There are now holes in the case opposite the fan)

I'll check the temperature tomorrow after a couple hours of running.

My other thought was that it was getting EM/RF interference from the 48 volt switching power supply since they are so close. I might take the gecko out of the case and see if it makes any difference.

Lastly, these are the motors I bought:
NEMA 23 Stepper Motor 3 5A 300oz in 2 4MH Inductance | eBay

They have 6 wires, I use 4 and wired it according to the diagram they have. Am I right to still use 3.5kohm resistors? This may be a better question for a different subforum.

I realized 3 of my steppers were miswired. Half of one coil of each stepper was wired rather than the full coil. This caused an imbalance. I have since rewired them, and the machine is incredibly smooth. I can't believe it cut so well before this. I also had a wrong setting in mach3. For my x axis, the one that had the biggest problem, I had 'dir low active' checked to reverse the direction. I realized this was the wrong way to reverse the axis and have since corrected it.

My problem now is that the 3 motors that were fixed, as well as the good one, get incredibly hot sitting at idle, even with the 3.5kohm set resistors. I'm talking boiling water hot (a drop of water on one of them steamed off after running a 20 minute job). It makes sense, since there is more resistance in each motor since both coils per stepper are having current going the whole way, rather than just half. I'm going to use different resistors to get the heat under control and hopefully all my issues thus far will be solved.

And the gecko is actually about room temperature no matter how long it runs.

Lowered the current to 3.1 amps. The z still gets very hot when constantly in motion. I may come up with a heatsink, though I'm not sure if it is exceeding the 230 degree limit or not. I ran a test job today without cutting. 250 ipm 3d job that took 7 hours. My problem axis was dead on at the end. :)

I think the main problem was having 'dir low active' checked. Hopefully this helps someone else eventually. The resonance was likely a separate and frustrating issue.

Quote:

---

I think the main problem was having 'dir low active' checked.

---

No, that would just cause the direction to be reversed. It won't cause lost position.


Where are your resistors? Having them at the motor end of the cables can cause excess heating.

Other than that, you can try running at a lower voltage.