Google+ post by D Rob on 2014-03-11 04:56:09 UTC

http://www.servocity.com/html/vertical_shaft_worm_drive_gear.html#.Ux6XE_ldWVE

There also worm drive stepper gearboxes on aliexpress

Ah nice, you’ll still get striping, but the wobble will be way less

Is it a servo, or a dumb motor?

Nice find. Replace the aluminum motor mount with a printed NEMA 17 mount. It looks like the worm gear mounts to a hub. Perhaps one could print a hub to adapt it to a 10mm shaft.

BTW, what is the ideal ratio we would want for the best compromise between layer height accuracy and Z axis speed?

@Dale_Dunn My thought is this z steps/mm should probably be set 1/2 of the steps/mm that give max capable resolution with current tech. So we can hopefully reach higher res, as tech improves, without having to change the drive only extruders and hotends and whatnot. What is the highest res print on a reprap ya’ll have seen to date? Once we know that we can do the maths to figure stepps/mm

I’ve heard of people experimenting with layer heights as low as 0.05, but I’m sure we want more resolution that that, just to be able to adjust first layer height. If I was trying for layers that low, I might want full steps somewhere around 0.005. That would be my guess, but I’m not sure what the more experienced users would say. Whether we can make an Ingentis platform flat enough for that is a separate issue.

Keep in mind as the layer height halves, the data pushed, and number of passes more than doubles.

This was also on their site…bet you could print an equivalent, and the gearing would be enough to prevent a fall http://www.servocity.com/html/gearmotor_gearbox.html#.Ux8nbtDnbqA

I don’t think a NEMA 17 stepper with the coils unpowered has enough torque to prevent a fall, even with a high reduction ratio. The worm gear is different because the asymmetric friction in the system makes it impossible to move the input shaft by turning the output shaft (for common worm helix angles).

yeah, but if it were connected to a leadscrew…

@Dale_Dunn @D_Rob 5 microns per step was my thoughts, but I don’t know how to work that out as a gear ratio, doesn’t it depend on the circumference of the rod that the line is wrapped around or the belt pulley used?

@Liam_Jackson , yes it will depend on what the belt or line is wrapped around.

@Mike_Miller A leadscrew is indeed a different matter. But they provide a high steps/mm already, unless you’re considering a multiple start ACME (trapezoidal) screw (which may not resist a fall). M6x1 provides 400 full steps per mm, 0.0025 per step. M8x1.25 gives 0.003125 mm per step. Both are a bit slow on a Mendel, but I don’t know if the Ingentis platform is too heavy to lift with fewer steps/mm. So, I’m very curious what Tim is using now. I don’t know what all pulleys he’s using, so I can’t calculate it myself.

Yikes…“I started Zeroing the bed last Tuesday, it’s halfway there”

well the math would follow something like this I believe: to travel the full distance of circumference of the pulley/shaft will take one full rotation on the axle. normally this is divided by the number of steps for one rotation (200 on 1.8 and 400 on .9 steppers) so spectra on a 10mm rod travels 31.42mm for each full rotation (c=2pir) this equals (in a 1:1) 0.1571mm per step. now find the desired worm ratio lets use 10:1 so now we divide by 10 to get our new steps/mm 0.01571

That sounds about right, especially since you remember the correct number of steps for a normal stepper. Oops.

Those of us who want to use belts have a much different situation. 32 GT2 teeth per revolution is 64mm/rev. (64 mm/rev)/10(200 steps/rev)=0.032 mm/step. Not so good.

If I still want 0.005 mm/step, I need a 64:1 ratio. That might be an awkwardly large worm wheel if printed. If I hob a plastic worm wheel with 64 teeth and use M6x1.0 threaded rod as the screw, I get a pitch diameter of ~Ø20. That sounds feasible, but it might wear quickly. ~2kg held by 0.5 mm deep ABS tooth doesn’t give me warm fuzzy.

@Dale_Dunn sorry for the brain dead moment. We forgot to divide by 32 for micro stepping. Or 16 or 8 depending on the jumper setup

.032/32=.001 way better

With 10:1 worm

BTW spectra and 8mm rod would give better resolution than 10 but the math isn’t as easy to do in the head