I have a 1000x1000 workbee with the following specs. Nema23 motors Arduino uno genuine TB6600 drivers protoneer shield 36v power supply 4.2w laser I have been chasing a problem for a while where my workbee appears to loose steps , I have just fitted a laser to it and thought it would be a good time to test for repeatability of stepping. The procedure I used was to generate some gcode to laser a circle and intersecting lines and somr gcode pulled from a box I have been attempting to cut that consists of 7 operations . I inserted the laser gcode before each operation with a X0Y0Z0 before it starts lasering the circle then aircuts the box operation code and repeats for each of the seven operations. When finished it comes back to home but it is off by a few mm inconsistantly and you can see the circle has moved each time it is lasered. I have repeated this test numerious times with a genuine uno and a fake one ( the fake one always is worse ) but there is no consistancy to the lost steps steps I dont believe it is in the gcode. I have moved all the signal wires to the drivers away from all power wires , the router is off so no power interference. I have checked all axis for calibration and they are good Any suggestions what is causing the issues ?
Are your steps properly calibrated? Instead of a nominal $100=40, you have $100=39.947 or whatever. If it's a belt unit, are your belts tight enough? If screw, are your anti-backlash nuts preloaded sufficiently? It doesn't look like gcode or signal interference to me, it looks like an issue with the physical machine or calibration.
Screw driven workbee calibration checked for steppers @199.50. I have just stripped all axis and checked wheel adjustment and lead screw adjustment . If i put any more preload on the screws are very hard to turn by hand If the stepper calibration was out i would expect the circle to just be out of shape consistantly but repeatable. Just trying a test at very slow speed of 300mm/min to see if its a stepper torque issue
Generally speaking, you shouldn't be able to turn screws by hand. I would expect that to be more of an issue with a router than a laser though. Step/mm calibration would normally produce the results you describe, but I also didn't know what you were doing between runs. You'd be surprised! It's not a stepper torque issue with NEMA 23s and a laser head, unless you have the microstepping at absolute max, maybe, but probably not even then. Have you tried homing between runs? If that's the problem, that's important, if it solves the problem that's also important.
Two thoughts (apologies if you have already considered them) - coupler slip on leadscrew or motor. What is your feedrate? are you driving your steppers close to the limit of what they can achieve? Alex.
I am doing tests at 300mm/min and 1000mm/m which I think the steppers could handle ok. Been playing with different stepper amp rates as well as I think the drivers could be the issue. In the gcode before every laser iteration it runs back to Y0 X0 Z0 just to make sure - machine homed before inital 1 hr G code run. I have been over all couplers and checked tightness , one thing I did do which has helped is to loosen of the Y gantry plate , preload the X leadscew bearing locknut and then retighten the gantry . this has put some preload on the leadscrew bearings which has halved the error. I am going to get a piece of steel today so I can get better holddown of my dial indicator to see if I can get the backlash better controlled , I had dropped the tension between the two nut blocks as I was having issues with stepper stalling at high feed rates over 1700mm/m with the lead screw tighter , if they should be firmer when turing by hand I will up the tension again. Honestly if I was just cutting wood I would suspect I would not even have noticed the errors as the machine would be rehomed after every job. I was making a heart box and noticed the two halfs didnt mate up perfectly when fitted together and thought it was my cad work. I have attached my gcode file for my test , laser only fires when doing the pattern , during the box operations simulation it is off. Zero Z is at laser focus point.
I would have thought if it was looseness somewhere causing play/backlash the errors would be repeatable. It seems more likely that there is binding somewhere or, as you already suspect, a driver problem. Good luck, Alex.
Just did some backlash checks with a heavier piece of steel for a base. Y1 = 0.010mm Y2 = 0.025mm X = 0.065mm Z = 0.010mm So it looks like I still have a issue on the X axis and to a smaller degree on the Y2 Here are my grbl settings and the lastest tests , last one has just about got it dialed in and would be about the backlash error I am seeing I think. I moved to a larger dia circle to try and see the error better. You can see my aircut tool path on the inerations where I slowed the laser down it marked the mdf. The last one is where it is after trying to preload the x leadscrew bearings.
Adjusted the x axis by putting more tension on the lead screw bearings/lock collars . Interestingly I was checking backlash by jogging 10mm then returning . When I moved 100mm and came back my y2 axis backlash is erratic other axis are fine , I think a lock collar is slipping and allowing the lead screw to move . Tried to put more pretension on the lock collars but it doesn't seem to hold . Eventually gave up will try again when I get some new lock collars and a clear head .
Well I delved into it again with a clear head. one Y axis tested perfectly and repeatable every time and the other was all over the place , it looked like backlash but after several attempts to remove it I though maybe its the drivers or wiring. So on the Y axis that was repeatable and accurate for backlash I first swapped drives over " BINGO " the error shifted to this good axis and the bad axis was now good. tried every combination of swapping wires to make sure it wasnt a wiring issue and it always stayed with the drive. tested my other 4 drives . So I have one good repeatable TB6600 , one really bad TB6600 that is all over the place and 3 TB6600 that are repeatable 90% of the time with a occasional error. LOOKS LIKE WHAT EVERYONE TALKS ABOUT IS CORRECT AND THE TB6600's ARE JUST PROBLEMS , to be fair I think the ones I got from the australian openbuilds makerstore are chinese copies as the screenprinting on the front is wrong for the 2.8A setting. here are some you tube video of what I was experiencing.
TB6600s can be unreliable and sometimes not even real TB6600s. People have found different chips when they have opened them up. I have some from a long time ago but I went with DQ542MAs for my CNC router instead and could not be happier. I thought about using the old TV6600s for modifying a K40 laser I just bought, but now you have me very hesitant. I think I will use my new Black Box.
Well changed my whole setup to a UCCNC AXBB-E motion controller and a Gecko540 , difference is night and day , drives are so much quieter and smoother , can rapid now at acceptable speeds and have had no lost steps at all. Repeated my previous testing as well as doing repeated backlash tests . happy to report everything is near perfect . At some point I will go back and test the arduino on the G540 when I move the AXBB-E over to the new build.
