Has anyone had an el-cheapo CNC PCB mill brick itself?

[I duz avionics]

Hi, I have just started out with a cheap CNC PCB mill. Reasons for wanting to mill rather than outsource are for another thread.
Just wondering if anyone has had any issues with these 'just stopped working'.
Most of the discussions that I have seen are about people that haven't yet bought a mill, or those that have bought a mill but not managed to get it working.
Mine worked at first then just stopped, just short of milling my first test board. This is my situation:

* Received mill from China, it uses an Annoy tools control board, which appears to be an alternate brand for Cronos maker. This has every appearance of being an Arduino embedded in to a 3 axis stepper motor controller board, with extra connections for laser, spindle motor, end stops and a couple of aux inputs of unknown utility. Very generic and clone of a clone as you would expect from Ali Express.
* Assembled half-a-zillion tiny brackets in to a CNC Mill.
* Plugged in and connected to Candle. Obtained manual control of machine, could jog it around using the arrow buttons. The required Ch430 driver was already present from previous Arduino clone projects.
* Exported board design from KiCad.
* Converted this to Gcode usign Flatcam.
* Imported to Candle. Gcode seems to load and display OK.
* Machine still responding to manual control. Zeroised coordinates at start point for milling job.
* Could not find any information at all about where to connect z-probe to control board. Using the z axis end stop input seemed sensible so put it there.
* Initiated a height map build.

The z axis unexpectedly went up instead of down, then stopped about half way up its travel distance.
And that's the last time I have managed to get any motion out of the machine.
The Candle software is able to open a Com port to the machine, and appears to send commands to it, but there is no physical or data response from the machine.
The stepper motors are all making a faint 'stalled motor' humming sound.

So....

Just wondering if this has happened to anyone else, or if anyone has heard of this?
My plan is to refresh the GRBL load on the board to make sure that I have the latest version.

Any other thoughts?

Todd.

[PlainName]

Is there an emergency stop you need to reset?
Are there physical limit switches? If so, any of them activated? If not, or if the limit switch is just used as a position reference, has the software decided the head is out of bounds on some axis? Could be it remembers where it stopped over power cycles, so needs a calibration run.

[Doctorandus_P]

I haven't had a CNC die on me, but I build my own from a "Blue Pill" board and a GRBL version that was ported to it, and it "just works" for me. It uses the hardware USB of the stm32f103. After the "Blue Pill" I've buffered the signals with some old LS TTL IC's and then they drive the optocouplers in separate driver boxes for each axis. Most of the small stuff is soldered on a Matrix board. Designing a PCB for a one-off is more work than soldering a prototype.

[I duz avionics]

Thanks for the replies.

I am posting this update so that anyone else that stumbles across this can see the solution to the problem.

A lot has happened since I made the first post. I have moved to a different country which means that it was a long time before I got back to the CNC mill.

Before I moved, I tried re-flashing the GRBL software, but it didn't help. I now know that was not a good idea without researching which version the board was designed to work with, because some pins have changed functions with newer GRBL versions.

After I moved and I got back to working on the machine, I decided that the AT MEGA processor was non-responsive by flashing GRBL to an Arduno UNO and learning what a responsive board behaves like. I could get a response on the serial port. I could not get same response from CNC controller board.

I built up a replacement controller with the UNO that I had flashed with GRBL, a CNC shield and an LM298 motor controller for the spindle. This has worked sufficiently well to get the machine running and calibrate it as best I can, but it could do with a better spindle motor controller.

After a couple of awful results, I managed to get it running surprisingly well considering I haven't yet done anything to remove the considerable backlash in the X and Y axis.

Other useful things that I learned is that the Arduino CNC shield pinout has changed so that the depth probe input is now connected to the SCL pin, and the motor speed signal is now connected to the Z end-stop.

I also learned that the depth probe works far more reliably with a 10k pull up resistor added to it.