Loosing steps running long program

I know this is a problem hard to give direct advices for, but suggestions where to start trouble shooting would be greatly appreciated.

The mill is a converted rung-fu round column and my configuration is here linuxcnc.org/index.php/english/forum/16-...er-and-db25-1205-bob

I have run a number of small stuff quite successfully. Both aluminum and steel.
Now I tried 3d.
Using the cambam sample "skull" and only the roughing part. Result in an about 8500 lines program. Cutting in aluminum.
After about 7500 lines of successful processing it suddenly looses steps ruin the work.
I tried to tweak some setting (rerun jitter test and tuning acceleration) and rerun the program and it looses steps at about the same place in the program.
I'm taking light cuts both horizontals and vertical, machine do not protest in anyway. Suddenly it just lost steps in both y and x.

Where should I start looking?

Anders

Heat can cause that.

Is the contoler and motors getting warm?

Norbert

You might want to determine where in the program the problem occurs, is it roughly the same place or exactly the same place and what is the code there.
You might be on the right course by reducing acceleration. I would try a DRAMATIC reduction in both acceleration and velocity. Keeping in mind a very short move may never reach full velocity where a long move may.
Try some very long high speed moves back and forth on the axis and then check they have not lost any steps...
A short program such as...Using max numbers for your machine run several times.
I think there are backlash and exercise example programs available.

Check your power supply is properly sized.

Rick G

Ok, spent another day in the workshop observing this and I have a theory of what is happening but not sure how to attack it.

I found it to hap pen not just with long running programs but in almost any program what look to be randomly.

I checked the heat thing.
Motors are worm but not hot around 30C. And electronics are at around 20C ( my workshop is at about 7C during winter)

But now observing when it happens I think it is when the bit is taking a "bite", like when it go back down after lifting over a holding tab. It is like the table "shakes" when the bit takes the first bite and it looks like the sideways force created by the bit on the workpiece overcomes the motors motion. And this happens as we'll at high federate as during low feedrate. And with both a 6mm bit as with a 12mm bit.

I guess the ultimate solution would be ballsscrews but I can't afford that.

So where should I start?

-better power supply? My power supply is a bit weak I think. Rated 36v while the drivers can take 50v.

- take lighter cuts? In general I take 1mm cuts. But I have at the same time taken some 5mm cuts without it to happen.

- try to lower plunger speed to something very slow? Not sure that will help as it feels like this shaking happens also with low plunger speed.

Or are there parameters in linuxcnc I could play with?

I should also say I tried both 2 and 4 micro steps and I have the problem with both. So playing with micro stepping has been tested so to say.

Any suggestions would be highly appreciated as now all the fun with this is gone as almost anything I try to do gets ruined.

Edit: should have mentioned. I do have a backlash on the machine that I use linuxcnc to compensate for. But I my theory is that it is this backlash that allows the machine to shake and generates this problem. Will try to see if I can tighten up the backlash without investing in ballsscrews and see if that helps

Anders

sounds more a mechanical problem than anything else
so rather than try and mask problems using backlash settings etc , try and remove and undue movement and tighten things up
so removing slop but keeping things smooth .

in the meantime taking shallower cuts should help and slowing down ,
what material are you cutting ? , using a smaller cutter dia will also help depending on the top speed of your spindle
so increasing your surface speed .

but overall look for solving them mechanicly where possible

I took the machine apart and tried to tighten up what I could. So now down a to 0.05-0.1 mm in backlash.
Still rattling a bit when the the machine starts to take a cut.
Also I see a difference when the force is more against the x axle as I was not able to take out as much backlash there as on the y axel and I wanted to tighten the gib on the x axel more but that caused the motor to stall. More rattle when force is against the x axel.

And tried to slow everything down (rpm and feed) + a bit lighter cuts.
So now I able to make a few parts successfully !

But I thing the core of my problem is my screws, so I'll start saving to get a pair of ball screws.
And a better power supply (50v as the drivers are rated).

Thanks for the suggestions I got!

Anders

Glad you are making progress. It is of course important to have everything as play free as possible.
In the meantime also look at your cuts, are they conventional or climb cuts?
If there is slop conventional may work better for you. (Cut the same direction you would when manually using a hand held router).
A climb cut may exert more force on the machine by trying to pull the machine into the cut.

Rick G

are the gibs well oiled ?
it could be also a factor that your motors are a little weak , or your stepper controller is under powered for the job in hand

glad your making progress

OK, a great step forward.
Changed computer and turned off SMI on this computer.
Also went down to 1/2 microstep and turned up the amp from the drivers to 3A RMS (had them at 3A Peak).

Two things happened:
1) I could run much faster without stalling (changed jogging from 470mm/min to 650mm/min and milled aluminum at 500mm/min)
2) No lost step. A 22000 line G-code program run perfectly.

I'm a happy camper!
And a 50V power supply is in the mail. Looking forward to what that one will do. Run 36V now.
Also about to order ballscrews.

Anders

Sounds like you are on your way!

Rick G