Question regarding Z axis: I have an original Mendel with belt driven Z axis. I have considered printing the Z-motor mounts from the Prusa Mendel and use them, skipping the belt and go for driving the Z rods directly (and adding another Z motor of course).
My question is, what would I gain from using two Z-motors to drive the rods directly? I know that I get rid of any problems with slack in the Z belt, but is it worth it? I use 3mm filament and a wades geared extruder with an e3d hotend. Love that thing :).
Basically just cuts down on complication. It was used as a step to reduce parts count. Easier just to get another $14 stepper than bearings, properly sized belt, pulleys, etc.
@Normand_Chamberland , if you don’t know what to print, you can always find something to tweak.
Thanks, I think you saved me some time. I always did wonder, why people stopped using belts on Z axis and I just asumed it was an improvement, not to save parts. If I don’t gain anything in quality from going from belt to direct drive, I’ll just find something else to tweak
Two motors are actually less reliable than having just one. They can (and will) skip independently of each other and skew your X-axis in the process. When running off of one driver, two motors are almost impossible to fine-tune and you end up missing out on a lot of torque and therefore speed.
However, the belt drive on the Sells Mendel was built way too complicated and finicky, just like everything else about that printer, and using two motors to get rid of that belt seemed like a logical choice at that time. If i’d design a printer from the ground up today, i’d probably use a belt and properly restrained leadscrews.
There is no mechanism in place that would keep them synced, so since each motor is inherently different from the other and the guides along Z are also unlikely to have the exact same resistance, one motor will skip sooner than the other when it’s running on the limits of its torque. Which is more often than one would think - driving the axis too fast, the filament spool running too rough or simply running into an obstacle (be it the bed surface or your arm) while driving the axis will all have the motors skipping. And once they skip, they’re out of sync.
Ok, I can see that. Has anyone tried to use encoders to keep track of the position of the x/y/z axis? If it can be done fast and precise enough, it should be more reliable than sending pulses to a motor and cross your fingers and hope that it will actually move.
Encoders are a valid concept to limit step loss and are widely used in industrial and high-end solutions. For RepRap purposes, encoders and matching drivers are still way too expensive (possibly more expensive than the whole rest of the printer), so sane people haven’t bothered with them yet since steppers alone work well enough. But if i remember correctly, @ThantiK experimented with servo drives, which are a similar concept.
As @Thomas_Sanladerer has mentioned, adding encoders is an expensive but effective option. I design motion control at work and we still omit encoders when using steppers quite often as the added cost of the encoder and the upgrade motor controller just isn’t worth it. As long as the stepper is rated with proper torque, we can simply home the stage and not worry. If we need it very accurate, we go servo, and the encoder is coupled to the output motion of the stage, not the motor. Also keep in mind that marlin and repeteir have no encoder support AFAIK.
@Normand_Chamberland as we need to transfer axial forces (gravity) as well as forces along XY (from the belt drive), there are very few other options than an angular ball bearing with a properly machined leadscrew. The Sells’ solution of using unmachined threaded rod was only flawed by the fact that an M8 thread, by design, never has an outer diameter of 8mm. Combined with the way the rod is mounted in the (8mm ID) bearing, this resulted in more or less severe Z-wobble.
Nevertheless, I’ve seen a lot of people spending more time modding their printer than actually printing.