scottalford75 / Remora-RT1052-cpp

Remora firmware for RT1052 based CNC controllers. C++ version
10 stars 6 forks source link

LinuxCNC movement commands do not accurately happen physically. #15

Closed Geramy closed 8 months ago

Geramy commented 8 months ago

If I tell my LinuxCNC machine to do G53 G1 X0 Z35 F800 and then G53 G1 X0 Z35 Y200 F800 and have it draw it on a piece of paper using a pencil in my collet we accomplish 195.25mm taken with a accurate expensive dutch micrometer. I have Step Width and Step Length at 8, there is now way i'm missing steps at all. It has to be something inside the firmware or to do with linuxcnc and the motion controller.

Geramy commented 8 months ago

IMG_9179 IMG_9178

Geramy commented 8 months ago

This still happens at F400

Geramy commented 8 months ago

If this was the worst ballscrew in the world I would have a inaccuracy of 0.21mm per 300mm. So its way above even the ballscrew imperfections.

Geramy commented 8 months ago

I have went from 2.1A to 3.7A same thing, I changed from step on Active Low to Active High and vice versa, same thing. I get the same exact output, like we are missing steps inside the DMA engine.

Geramy commented 8 months ago

We are nearly missing a full rotation, 1000 pulses.

Geramy commented 8 months ago

I’m testing other drivers I have which are cheap drivers at 3200 micro steps with crappy drivers I accomplish 200.75mm with a no name no brand driver. My 2DM542-06-JG-2/4P seems to be having some problem.

Geramy commented 8 months ago

I am using the same ghetto driver at double the micro steps and I have the same result. But I did get a direction change had to restart linux CNC.

Geramy commented 8 months ago

I am using the same drive at double the micro steps and I have the same result. But I did get a direction change had to restart linux CNC.

Geramy commented 8 months ago

Using my nice old drivers I removed smooth stepping and out of a 200mm move I can get 199.86mm so maybe that’s it.

Geramy commented 8 months ago

Looks like one of my drivers was way off. we are at about 0.28mm off after a 200mm move which is close to the ballscrew inaccuracies. I cant measure accurate enough at a quarter of a mm in order to accurately tell. I will have to do a machining operation like doing a surfacing operation or something.

Geramy commented 8 months ago

So the outcome i've been testing with my X axis as far as movement tests, which the driver seems to be accurate there. My Y driver was messed up not sure why or how but both seem to have the same accuracy now, im gonna close this ticket unless i find anything else :)