I have built a fixed gantry mill and while milling it will suddenly start milling in the wrong place, like the coordinates have shifted, by anything up to 100mm. This happens in X or Y directions, intermittently. Everything seems to work fine apart from this strange fault. Machine details, as I know you need. The gantry is 250 x 150 x 5mm RHS steel, the rest of the frame is 100 x 50 x 5 RHS steel, all MIG welded. The linear rails are 20mm diameter with 20mm x 5 ball screws and nuts. So its pretty strong. It has a working area of 1000 x 1000 x 120 mm. Nema 23 485 oz motors, 9mm belts to the ballsrcews. I run at 24V through a Wantai 4 axis TB6560 driver board and a 11 Amp PSU. motors are all 8 stepping. I am using grbl-Mega-5X-edge on a Mega2560 board.(I intent to add a rotary axis very soon and use cn5X 4 axis G code sender). I have a VFD with a 1.5kW spindle, this is isolated from the machine as well as I can do). All the cables to motors and limit switches etc are shielded (1 end only), the PSU 0v is tied to mains earth I use Openbuilds Control (I like the interface) for sending G code. I use ArtCam for designn and generation of G code. In Openbuilds Control the 3D toolpath shows correctly so I assume the G code is correct. When the problem has happened, I have stopped the machine, reset the alarm and then homed the machine. The coordinated shown are the work coords (negative of course but correct. If I then send the machine back to the work coords it goes to the correct place. If I don't home and then go to work coords the machine goes to the wrong place (off by the amount it has jumped) All the motor spindles and ballscrew spindle ends I have ground flats on, so the pulleys are not slipping. Obviously it's not missing/adding steps by that huge amount. G0 speed is max at 2200mm/min I have used Universal G code sender and experienced the same problem, so I don't think it's Openbuilds Control. This has me puzzled, what is the problem, sending software, grbl, the windows10 laptop, WHAT? Any ideas people?, Any help very much appreciated.
Do a Search above, the TB6560s very often come up as Duds. Tb6600, TB67S109 or DQ542ma are all better choices. Some have luck with TB6560s but very much depend on who made them
Thanks for the reply Peter. I've heard the same rumours, however I find it hard to believe the driver could miss/add that number of steps. Work it out, to jump 25mm the driver would need to be in error by 240x25 steps, a rather large number. I suspect grbl mis interpretation of G code. I am however looking at new drivers, mach4 and accompanying motion controller.
Wheres the gcode from? If it's fusion, similar posted today: Openbuilds Control or Black Box or grbl or pilot error?
As much as that is true, the viewer parses it, but grbl has its own parser too. Not impossible for them to view it differently. But more of a fusion thing. Their default post generates gcode that does make Grbl move oddly
What is the ballscrew pitch, pulley config, final steps per mm count? (still suspect of the TB6560s because i have yet to see more than two people like them) want to see if you aren't still exceeding minimum step pulse in the rapids. We use high helix 8mm screws and on the forum here i recall Tb6560s topping out at 3500 mm/min with a finer pitch it would be even lower. In which case one can up the step pulse time in Grbl, or drop down to lower microstepping (or both)
Yes, like I said, I suspect grbl. This is a good strong machine and I think it's worth upgrading to mach4.
Either or. Grbl is fine if you feed it valid Grbl compatible gcode. Still not too sure that's the issue here though. But do post the file and i'll check / run it
Ballscrew pitch = 5mm motor pulley 2mm pitch 40 teeth ballscrew pulley 2mm pitch 30 teeth 8 stepping. So 5mm = 0.75 (30/40) motor revs 200x8 stepping =1600 steps/rev 1600*0.75= 1200 1200/5=240 steps/mm The max speed I can get is 2400mm/min before the motors complain. I have accelartion at 300mm/sec2 The feed rate for this cut was 900mm/min, spindle 12000rpm
Do another test with max feedrate dropped to 60% of the maximum you got (so instead of 2200, try 2400*0.6 = 1400-1500mm/min Just to help zoom in or eliminate one cause
I think we are left with EMI if the drivers are ok. shielding, star ground, separate signal and power wires. all the usual stuff....
I had some trouble when I was setting up the VFD, it is now mounted on the wall about a meter away from the electronics .
I also connected the PSU 0v to mains earth, mains earth to the machine frame. The spindle is earthed therefore. Unavoidable to have a seperate mains earth to the VFD but they are all connected together. All limit switches and motor wiring is shielded and shields are connected one end only at the control box to 0v. The Mega 2560 has its GND terminal connected to PSU 0v, but the USB GND is connected to the Mega GND and is obviously connected to the PC GND. The cable to the spindle is shielded and shield is connected to the VFD mains earth. Yeah, loops everywhere but what can you do !! Basically all the GNDs, 0v and earths are connected to mains earth and the chassis.
I doubt its EMI though (sorry David hehe) I have seen TB6560 headaches enoughl times and had them myself back when i got into the hobby. Try that test with 1500mm/min rapid today, at best it works and we found the issue, or at the worst it still happens but we eliminated at least one check. As it stands none of us trust them (; hehe
Hi Peter, I gave it a good test, without the VFD plugged in; all was good at 1800mm/min rapid. When I plugged in the VFD and started to cut some wood I got Alarms, don't know what they were, no way of tracking. I moved the mains cable to the VFD further away from the control cable (which are screened) and it was better but still alarming now and then. I ran the toolpath (attached) and Openbuilds hung up close to the end of the program. Had to use task manager to close it. Don't know why. Anyway getting there, I feel I must rethink the wiring. I've ordered new drivers, here Over Voltage Protection Stepper Motor Driver Digital Automatic For Nema 17 23 24 | eBay. Hoping these will be better.
google 'star grounding' and then do it, no exceptions (-: VFD's are electrically noisy so every care has to be taken.
