inexpensive nema 17 "closed loop" stepper driver

[WhiteHare]

If you move the Z axis up and down randomly for a few minutes, does it return to the same zero point or does it shift?

It depends. If I move it long distances at a high speed, it may not return to the same zero. If I send it short distances or at very low speeds, I think it may be OK, but I'm not 100% sure. That's why I'm waiting to receive the fancy gauge: then I'll know for sure.

[routalot]

If we are talking about steppers, any overshoot you see must be purely mechanical (inertia of the axis). The motion planner does decelerate, but it has no way to know how much deceleration is "enough" - you have to tell it (in the controller settings) what rate to use.

Anyway, you don't normally rapid all the way to the part. I don't think that would end well even if you had a perfectly tuned machine.

I think this is the root of the problem.The steppers may not generate enough torque to overcome the inertia-aided by gravity.If slower speeds give more accurate and repeatable location it narrows things down a lot.The concern I have is that the new tramming tool will be heavy enough to adversely influence the outcome.Which is why I would locate a dial gauge on the table to verify the position.

[WhiteHare]

I think this is the root of the problem.The steppers may not generate enough torque to overcome the inertia-aided by gravity.If slower speeds give more accurate and repeatable location it narrows things down a lot.The concern I have is that the new tramming tool will be heavy enough to adversely influence the outcome.Which is why I would locate a dial gauge on the table to verify the position.

Good point! Thank you for posting it.

I need to investigate whether GRBL allows for different acceleration/deceleration in one direction vs its opposite (to account for different z-axis gravitational acceleration and gravitational deceleration). But if I got it to work with the tramming gauge installed, I'm not sure what would happen when I removed the tramming gauge. Would it somehow over-decelerate? Would it maybe even change directions when it shouldn't because of the overcompensation? In short, on a stepper does deceleration work like rocket fuel burned in a rocket, or does it work differently? I have no idea, but that's what I should probably look into next.

[WhiteHare]

Well, I contacted a different manufacturer who makes these kinds of tramming tools, and they're offering to fabricate something custom if I can specify it. So, hopefully I can dream up something lightweight. If anyone here has specific suggestions along those lines, do let me know. Posting pictures gets bonus points. :-)

[WhiteHare]

Something like this might work:
https://www.cnczone.com/forums/attac...d=447378&stc=1
It claims to have an accuracy of 0.0005". I just wish it had better resolution!

Also, it's too tall for the modest range on my z-axis.

[WhiteHare]

This might be the best so far, except that it comes at a cost of $364:
https://www.cnczone.com/forums/attac...d=447380&stc=1


https://www.grainger.com/product/MITUTOYO-Balanced-Reading-Zero-Setter-54GF31

Great specs though.

[WhiteHare]

OK, I think I may have dreamed up an inexpensive solution: I'll electrify the spring bit, similar to how it's done with a standard bit during auto-levelling. Then, after ramming the z-axis down at high speed by issuing a "return to zero" command, I'll measure how far up I have to move the z-axis before the spring-bit looses electrical contact. Voila: the resulting number will be the length of the overshoot. The aim then would be to calibrate out any overshoot, and have the spring-bit just barely kiss the detection plate after I issue a "return to zero" command.

Not sure about the accuracy of this method, but it's free because I already have the parts. It should also work even with the limited z-axis range of motion. Lastly, it won't add any significant intertial mass to to the z-axis, so that concern is handled completely.

[routalot]

We have a difference to consider between tramming and checking repeatability.For tramming you need a gauge of some sort.For repeatability you can just use a pin in the collet and a hard wedge with a line on it.With the pin lowered to a point where around half of the wedge will slide beneath it-carefully add a mark.Extract the wedge and run the cycle of raising and lowering-then see if the wedge still fits at the same point.If it shows the pin at a location other than where the line is you have a problem.A fine wedge may be needed to identify small discrepancies,not the sort of thing you hold the door open with.

[WhiteHare]

We have a difference to consider between tramming and checking repeatability.For tramming you need a gauge of some sort.For repeatability you can just use a pin in the collet and a hard wedge with a line on it.With the pin lowered to a point where around half of the wedge will slide beneath it-carefully add a mark.Extract the wedge and run the cycle of raising and lowering-then see if the wedge still fits at the same point.If it shows the pin at a location other than where the line is you have a problem.A fine wedge may be needed to identify small discrepancies,not the sort of thing you hold the door open with.

Good idea! Yes, that should work. I suppose a block plus feeler gauges would also do the business.

[WhiteHare]

By the way, if you look at time index 20:00 on this video:


you'll see that he does the same kind of rapid "return to zero" repeatability test on the z-axis as what I'm aiming to do, except that he's doing it on the upstroke rather than the downstroke. Well, at least in his case, he gets it to work spot-on, and with far more inertial mass than what I'm dealing with. And later in the video he reveals that he's driving his at higher velocity than what I am as well. He's running his z-axis at anywhere from 1000mm/min to 2000mm/min, whereas mine, by the default GRBL settings that override g-code, can't be more than 800mm/min. So, he has more intertial mass and greater momentum and great speed, and yet his setup is not over-running on a retutrn-to-zero. Instead, as shown by the dial gauge, it returns precisely where it should at the end of a rapid. So, I just mention all of that because plainly those aren't limiting factors by themselves.