Calibrating mm/step for leadscrews

[itchytweed]

Let me start by saying what the equipment is: OpenBuilds LEAD 1010 with X, Y1, Y2, Z axes. This is leadscrew driven using TR8*8 screws. Steppers are 200 steps / rev set to run at 1/16 microstep yielding 400 steps/mm, so one wound think. That is what one is lead to believe being an 8mm/rev ACME screw. Then reality sets in and you realize that nothing is coming out to desired size, especially with inlay work, etc.

I set up a measurement test using a bunch of 1-2-3 blocks giving me a measurement route of 10 inches / 254 mm. I set up the blocks as 1-3-3-3-1 and approached the 1's from the same side to eliminate backlash issues (later dealt with). I was doing the measurements with a Haimer 3D Tester (1 div = 0.01mm). The Haimer tells me the error in motion. What I came to find is that I don't have TR8 leadscrews as they are all unique in their pitch value. I ended up with these numbers: X: 394.92984; Y: 395.10629; Z: 396.17999. All axes use anti-backlash nuts. Since the adjustments have the appearance of being analog, they aren't. They are actually limited by the granularity present from the mm/step value. The kicker here is that the mm/step value changes based on the steps/mm value used. For example, 400 steps/mm works out to 0.0025mm ( 1/[steps/mm] ). So, for my X value, the granularity works out to be 0.002532. This value also comes into play when setting backlash as the backlash setting has to be a multiple of this value (for the particular axis).

Once I created the attached spreadsheet and starting plugging in the numbers, in a matter of 3-4 tries I was ending up with a needle's width error on a 10 inch/254mm move. The machine's mechanics are far sloppier than that value but it is what I was ending up with when stable. The second section is for calculating the backlash values to use. That will take some empirical testing but the section will help with the step granularity.

Hope this helps out.
LeadscrewCompensation.ods

[DBS3262]

If Im reading your post correctly, I experienced this issue with open builds leads. It's not the pitch that changing, its the "lead." The lead is the distance the nut bock moves in one rotation. I brought this to their attention and was quickly trolled, and accused of stretching my leads at installation. In my case, I found that all my OB leads had its own resolution. You can set the resolution for the lead in your config for X,Y, Z, but it doesn't work for your ganged axis because the slave motor takes the same number of steps that the master takes. That subtle difference lead to lead is a fraction of a millimeter, but the error accumulates with the number of rotations. Depending on distance traveled and margin of error lead to lead, this can really add up. In my case I noticed I would gain 2mm on the slave side of my y axis while moving the master 1200 mm.

I've been meaning to bring this to the attention of Terje. The Auto square feature already separates the motor inputs for ganged motors moving the slave motor and additional limit to the "A" output. I think it would be a great idea to allow the resolution to be configured for this motor also. Then you could dial everything in.

In the interim, your choice will be, buy more leads from open builds and hope to find a better match between your new collection of leads, or find an alternative lead vendor. Makerstore claims to have an accuracy of .05-.1 mm over a 1000 mm. The problem is you will need to replace the nut blocks and tensioning nut if you make the switch.

[itchytweed]

We are thinking the same thing. The leadscrews are chinesium quality and are demonstrating it. I even used my digital calipers to measure between the screws and the same turn count gives me different values of distance. The nutblocks have enough slop in them to deal with the slight differences and all of my axes use the ABM blocks. I do though agree that with the X-Y1-Y2-Z setup that the Y1 and Y2 steppers need their own individual steps/mm and backlash settings to prevent racking the X axis (another item to put on Terje's more-than-full plate) on a long run from home. In absence of this, one would have to buy a metric crapton of leadscrews and either do a side-by-side, one end butted comparison or put them into a test jig and do what I did. In defense of the leadscrews and their manufacture, they can only make the tolerances only so tight before it becomes impractical. One could also argue that the OB units couldn't maintain 0.01mm accuracy just on machine flex alone so why go to these lengths? I do it because I can ask for a 0.630" hole and get a 0.630" hole out of my machine. .