Z Axis - precision and accuracy

With M8 1.25mm pitch threaded rods the steps/mm is 2560.

3200 is the number of steps for a full turn of the rod (regardless of its size or pitch). (ie 200 * 16 = 3200 since the motor is 1.8 degree and we are using 16 microstepping).

To raise the layer height .125mm the rod needs to rotate 1/10 of a turn (36 degrees) which = 320 microsteps. There should be no rounding error issues with that number of microsteps.

However the data is showing the rod actually turns between 35 to 37 degrees (or 311 to 329 microsteps approximately) in a repeating and regular pattern.

There is no reason why the motor can not move reliably and precisely 320 microsteps every time (assuming it has enough power of course). IMO there is no reason to disable microstepping (although perversely that might actually mask the underlying problem).

The pattern of microstep variation is so precise and repeating that IMO everything points to a firmware (or maybe A4988 driver) issue causing this behaviour.

I have some gear to count the number of microstep pulses being sent to the A4988. This should help isolate the issue further.

 

Yes. I am in contact with Jerry and getting support.

Re the Z Ribbing

I connected my pulse counter to the Z Step input on the A4988 driver to monitor the number of steps Marlin was asking the stepper motor to move. I was hoping I would see the number of steps vary to explain the variation in the way the rod turns, but there was none. "Hoping" because the easiest issue to fix would have been a firmware bug in Marlin.

The video shows the cumulative step count for an M8 1.25mm pitch rod. I used a test gcode file which raised the Z height 20 times in 0.125mm steps. The counter shows the correct number of microsteps 320 is being sent by Marlin each time.

https://vimeo.com/97378302

Just to be sure I watched the counter during a real test print (in case there was some other interaction going on) but the microstep count was correct.

So I am left with some mechanical (stiction, friction, motor, coupler, screw or nut) issue, Marlin speed/acceleration/jerk configuration issue or possibly an A4988 issue causing the variation in screw turn.

I am still thinking through what to do next.

 

No. But he has pointed out that the Z Stepper will be always powered on during Auto Levelling. So if there really are issues with the Z Steppers always being powered, they will definitely show up on Robos with Auto Bed levelling.

Your entire machine freezing is very unlikely due to the Z Steppers being on. Maybe I could understand the Z axis freezing but not the entire machine. Did you look at the Repetier log?

 

As you know auto leveling relies on the Z axis being "adjusted" during the print to compensate for the varying bed height.

So if you have a slightly uneven bed, the Z stepper will be working to make the adjustments for the entire print (as well as layer raises) whether you have told it to disable or not.

However if your bed is level or close to level, then of course the Z motor is just raising the height every few seconds/minutes of print or making the occasional level adjustment . And this is where there can be an issue if the Z motor power option is set to disable. Either the Z stepper will make an inaccurate layer raise and/or there will be an inaccurate bed height adjustment. To avoid this type of inaccuracy, the Z Motor must be always powered on during the print.

If by "for good" you also mean whenever the printer is powered on - then I have noticed that all of the motors (x, y and z) seem to time out shortly after a print or when the machine is idle for a while. I will check how long it takes.

 

I have two good quality and higher torque Nema 17's which I bought for another project. So Yes I have planned to try that next few days.

The protractor is at the top end of the threaded rod. I did look at whether the couplers could be winding up and releasing slightly. But it takes a lot of torque to do that and it's hard to see how that could be happening in such a regular pattern. However I haven't ruled that out.

There are also some other possible variations which could be caused by the couplers (eg slippage between the coupler and rod or the coupler and motor shaft). I will be checking this out.

Wiring the steppers in parallel will reduce the available power to each. There is an option in Marlin/Ramps to use separate A4988 drivers for each stepper, but that means a second extruder is not possible. But I haven't ruled this out as a solution as it might avoid having to replace motors for most users (if this turns out to be the problem).

Yes. Turning off microstepping might reduce the inaccuracy and worth a test. But as a permanent solution it comes at the cost of lack of flexibility in setting layer heights. Most other 3d printers happily use microstepping. I think there has to be a better solution than turning it off.

 

Mike

I checked this once but will do it again. I've had some experiences in the past where you get fooled into thinking something is obvious/proven only to realise latter it was not.

Re the earlier comments about microstepping it is fairly easy to test by changing Ramps jumpers and configuration.h. It's on my list of tests but I have some others to try with the couplers first.

So far I must have spent over 40 hours on this with lots of help from others via one on one conversations. Fixing the Z_DISABLE inaccuracy is a step forward, but there is a solution to the ribbing as well, just a matter of more time.