Closed loop servo control with custom driver board

Hmm, i'm close.

I am looking through the hm2_eth driver source, but i can't find exactly where packet data is inserted/parsed...

Alright, i finally have the ability to exchange packets with LinuxCNC.

Everytime i start LinuxCNC it stops right away with an error message after sending dozens of requests. When looking through the source code, i found there is a

" if (debug) "

and lots of log levels everywhere, which prints additional debug information, which would be really valuable to me.

Is there a way to launch LinuxCNC in debug mode to print all messages or to increase the log level?

I'm not sure about the log level, but I know of two ways to increase the logging verbosity:

* Launching linuxcnc with the -v flag
* Setting the DEBUG variable in the .ini file. This is a bit mask that selects output from different parts of the code, if I have understood correctly; I believe 0x7FFFFFFF gets you everything. The individual bits are defined in a header somewhere.. yes, in here: ./src/emc/nml_intf/debugflags.h. I have not looked into how (or even if) this actually works though.

The STMBL servo drive does a lot of what you are describing and even uses an STM32 to do it.

Making your own motors and encoders sounds like a fun project. I did it myself once, with mixed success. (I needed a servo-motor with a very large through-bore)

An easy and (now) relatively cheap way to get a target for an optical encoder might be to use a PCB fab house and order a custom solder stencil. They are laser-cut from thin stainless steel, and can be made to any artwork. This is how I had mine made for my motor:

photos.app.goo.gl/UJhu3RTKrto2yRXv7

@arvidb: Thank you for the suggestion, but unfortunately it didn't help anything. It printed more, but entirely not what i was looking for.

I now took the time and compiled LinuxCNC from source and then edited the functions to always print, no matter what log level. Now i get all those prints and i am adding more print statements to further reverse engineer the interface and fully understand it.

@andypugh: Thank you also for your suggestions: STMBL surely looks interesting.

Also, the solder stencil is a really good idea for manufacturing. Up to this point i cut mine manually with a carpet knive in laminated paper and reached up to 200 pulses per revolution. Now i used transparent film and printed the lines with a laser printer, which looks really good and has great accuracy, but it still needs to be tested, if the ink actually blocks infrared light...

HerrNamenlos123 wrote: but it still needs to be tested, if the ink actually blocks infrared light...

I made a laser-printed reflective encoder target once.
Then I laminated it, and it stopped working. Once laminated the IR reflectance was the same for both black and white.

So, I would not like to predict if your laser printed transparency will work or not.

I like this idea, a very robust and accurate slotted disk for cost of a PCB , which PCB house did you use?

nickom44 wrote: I like this idea, a very robust and accurate slotted disk for cost of a PCB , which PCB house did you use?

It was a long time ago, and there turned out to be a company less than a mile from my house that _only_ made PCB stencils.

If I was getting one now, I would probably use seeed. www.seeedstudio.com/stencil.html
The challenge might be getting the pattern needed as a Gerber. But you would get quite a few encoder discs on an 10 x 15 sheet.