I've been having issues with losing steps on rapid moves in my programs. This appears on all axes, and seems to be highly repeatable re-running the offending programs.
Previously this has happened on plunges in Z at the beginning of a program, stepping through the code in single block mode works fine, and this was enough to get me going and move on.
I had a thread milling program this week that calls a sub-routine which was stalling on every traverse in X, so it forced me to spend some more time looking into the issue.
I know that it is not the slides binding. I can run a program with a different Z offset and see identical behaviour. Manual jogging rapids and single block rapids both work fine.
This has to be something to do with how Mach3 processes code.
I tried all kinds of things with the subroutine to try and find the source of the bug, thinking this is something to do with my code, or the complexity of interpolating the helical moves as it reads ahead.
Eventually I stepped through the entire thread milling program just to get the part off the machine.
Today I've finally got to running a program that mills away webbing between parts in a billet. I had this stall in Z issue meaning my webs were left intact and rerunning the program gave me the same results.
This prompted me to look more at the read-ahead settings.
Conspicuously, the "( MILLING POCKET )" comment/message that TurboCADCAM puts in the code was 19 lines ahead of the rapid move which stalled. Changing the read-ahead settings to 10 lines gave me a stall in Z at a different level (convenient coincidence that my cutting passes were 9 lines long and there was a rapid move in Z).
So far, I've not observed this behaviour where a feed-rate move is at the appropriate position in the code, but a global replace of '(' with '%(' to make the headings real comments appears to have fixed the issue.
Does anyone have any experience with this kind of thing or anything else to add on the topic of comments/messages in code, and how Mach3 (doesn't) handle them?
I can post some example code or provide more information next week if anyone would like to look into replicating the behaviour themselves.
It's been a long week, far too many nonsense errors. The weekend came just in time!
Thanks,
- Ian.