Hi everyone I'm new, I'm from Spain, I'm sorry by my poor English,

Hi everyone I’m new, I’m from Spain, I’m sorry by my poor English, I’ll try to express myself as simple as possible.

I have a problem with auto-level, I want to make PCB for SMD components, and I want to make tracks of minimum 0.15mm.

The problem is that if I do a job well I can’t make repeat it, I uses the same G-code, and the same auto-level configuration but sometimes the tool goes deep well and others do not, it stays short scratching the surface or penetrates too much breaking everything

The CNC I use is the standard 3040 modified with Arduino Uno + GRBL 0.9j

The tool is v10º - 0.1mm
The configuration of the auto-level is

I use windows 7 and Chrome

I’ve tried it in many ways but I can not make it stable, I think mechanics is not the problem
If you need more datas, say to me

I show some good pcb tests

Thamks for you help

Hi Nestor
What type of usb chip has your Uno got ?

Your boards look great by the way. Have you tried any fine pitch work like an atmega 2560?

Hi Justin,Yes, it’s a Chinese pcb, it’s weird the chip CH341SE

There are quite a few reports of that chip being unstable.
And quite a few about the 16u2 having firmware issues. The good choice seems to be a ftdi. Do you have any ftdi chips that you could use ? Maybe an IC programmer you could hook up to the serial pins directly?

No, only the Arduino One

I don’t have it, but if you confirm me that using an arduino One with that chip that you tell me is work well I will go to buy it.
I’m desperate hahaha

All things being equal either there is some occasional misstepping on your axes (perhaps some dirt on the screw) or your getting some arbitrary comms problem causing the axis values to change. There are some reports of this.

No guarantees Nestor! What other boards do you have?

The nanos have ftdi chips. But they don’t interface neatly with the gshield (is that what you are using as a motor xontroller?). You need to use jumper leads to connect it.

I have checked the axes, I have done some gcodes where I move the axes in resolutions of 0.005 from 0 to a distance of 20cm about 30 times at a speed of 200 and does not lose any step, if that was the problem, do you think I could have done some pcb ?, the ones I have managed to do are almost perfect

I see that when doing the auto-level using the same configurations, the results of the measurements are different

I think it is possible for missed steps to occur when milling, yes. first from too much stress on an axis caused by the bit being in the material. second from dust and debris getting on to the lead screws and accumulating. But I agree that it is less likely.

on the auto-levelling: are the measurements different for the same board? if so, what is the maximum difference for a point?

Yes, I repeating the auto-level on the same board and without modifying anything at all, I have taken measurements in the same point that goes from 0.010 to -0.036, I put some sample images

Another data, using the bCNC software happens to me exactly the same

@Nestor_D Does it go to a depth that is not in gcode and report that it went to that depth? That is what is happening to me lately.I got a bit sick of it and a little busy, so I have avoided looking at it more.

so 0.04 mm difference.

what type of leadscrew are you using on the z axis?

Hi, Nathaniel.
I don’t know, when I do the work I use the auto-level corrections and what it shows on the screen is the gcode corrected, but I have used different depths, from 0.034 to 0.055

The copper thickness of the pcb is 0.35um

Justin, my machine has in its three axes
ball screw, diameter 12mm thread pitch 4mm
Motor NEMA23- 200step
Pololu DRV8825 1/32 microstep

the standard resolution is 2/100mm for your ball screw. so that difference seems to be within the expected resolution.

but when you add in the microsteps then that’s not so good! Although it is odd that you use 32x microstepping on the z axis. from what i understand, that is quite unusual!

but still the difference doesn’t seem that troublesome in real terms. it’s only 1.6 mil, whereas typical copper thickness is 27mil or so.

I would worry more about gross z differences.

(I assume you are quoting everything in mm of course…)

Yes all in milimeters.
But the strange thing is that there are works that have made it perfect in all path and others that fail and when it fails, it´s do from the beginning, I mean that if it it fail, it`s all at the same height, and if it does the job well does all at the same height, if it were a problem of lost steps in the same job and in different areas you would see parts scratched and parts where the tool has penetrated more and is not the case

Sorry my poor english :frowning: