Mendel questions on 0.1 mm positioning accuracy

The x axis seems to be a gear mounted on a stepper with several idlers and directly driving the extruder. Assuming 200 steps per revolution of the stepper, you would get a tenth of a millimeter per step by sizing the gear on the stepper to have a circumference of .1 x 200 mm or a diameter of 20/3.14 = 63.7 mm (which seems reasonable from what I can see)

Am I on target here or are you doing half-steps to get that resolution?

Is the y axis substantially the same?

The Z axis is interesting, and I really like the way you use gravity to overcome blacklash. I know that you are using 8 mm threaded rod, but is the pitch 1.0 or 1.2? Do we have the same resolution on this axis too?

You've skipped a carry or something, (0.1 millimeters * 200) = 10mm (circumference) which is far too close to the belt pitch to be practical - you'd need a gear with only two teeth!

better off using 1/8th stepping which gives you a gear of diameter (0.1mm * 200 * 8) / pi = ~5cm and if that's too big, go for 50um precision and a 25mm diameter gear. Accuracy is usually lower than precision, rather than equal.

The A3977 chip in makerbot's V3 stepper controller has built-in 1/8th stepping.

Would just need to check the speeds at this precision. A step rate of 1kHz at 50um per step is only 5cm/s, and I'm really not sure how fast the firmware can send individual steps.

As for the Z axis, [en.wikipedia.org] will tell you that M8 has 1.25mm pitch and M6 has 1mm pitch. Thus, a full turn of the stepper will only produce 1.25mm of movement- one step gives 6.25um of movement so it has significantly higher precision than the X or Y axes simply because it's geared down something fierce. Microstepping or further gearing would be a total waste of time on this axis.

>you've skipped a carry or something,


(200 * .1) / 3.14 = 6.37 mm diameter. Yes you found my mistake.

Maybe i'm missing a link to all the Mendel parts shown as isometric drawings and measurements. That way I could look at the parts myself and grok any gear ratios. My question was actually "how does Mendel get the .1 mm positioning accuracy? Is it micro-stepping? or maybe there's a gear somewhere that I just can't see?

The unasked question after that is what should I do on my (planned, larger) RepStrap, seeing as the Z axis and the Y platform are both going to be heavier. I'd also like to build enough torque into my design the first time around so I can do some milling with it too.

> As for the Z axis, [en.wikipedia.org] will tell you that M8 has 1.25mm pitch

Well, you've found my second mistake. Thank you. But your linked-to chart has thread spacing on M8 has both 1 and 1.25, not 1 and 1.2. Being in the USA, I have no idea what is most common in the UK. All of the threaded rod commonly available here is in SAE (threads per inch).

Also, from working on my Toyota, I know we in the USA usually express a metric screw as "M8x1.25 60" for a bolt that has a major diameter of 8 mm, threads 1.25 mm apart and a length, (not including the head of the bolt), of 60 mm long.

have never seen the "fine" pitch on anything, ever.. only standard pitch

As for the X and Y accuracy; I have some 20 tooth MXL pulleys I plan on using with a Mendel. I generally use 200 step steppers in half step mode, so that's 400 steps.

MXL teeth have a 80 thou tooth pitch, but just to check, I measured the diameter of my pulley - 12.3 mm. So, 12.3 mm * 3.14 / 400 = 0.0966 mm/step. That's a little better than my old Darwin, which has 0.125 mm/step on the X and Y.

I use M8, 1.25 pitch on my Darwin's Z axis, which with half stepping gives me 0.003 mm/step, which is way overkill, but it works because we don't need to move the Z axis too quickly. On my second Darwin, I set the Z axis to full stepping, as there's a lot of friction on a Darwin Z axis, and the full stepping helps out with the extra torque.

And yes, Toyotas rock. Especially old Tacoma's.

Wade

As far as I can tell, the Mendel has 8 teeth on the drive pulleys, thus with a 5mm tooth pitch, we get 40mm / rev at the stepper. Given a 1.8deg stepper, we have 200 steps / rev, thus 40mm / rev / (200 steps / rev) = 0.2mm / step. Thus to get .1mm precision, we do 1/2 steps which the A3982 does with MS1 tied high as is done on the version 2.3 stepper controller. To do better, you'd have to go with different belts, microstepping, or a fun block-and-tackle system to reduce the ratio.