I would like help confirming that my problem is with chilipeppr.com/jpadie and not my

I would like help confirming that my problem is with http://chilipeppr.com/jpadie and not my gcode which came from a program that I wrote myself. Any volunteers? I am happily testing my gcode and all of a sudden my mill is cutting too deep where there does not seem to be any instruction to cut deep.

Originally shared by Nathaniel Stenzel

I need someone to test output from bucket_mill.py. http://chilipeppr.com/jpadie seems to keep going nuts and moving too deep. That or my arduino is going nuts or I am missing some Z steps. I checked my bit and it was not slipping out of the chuck. The two gcode files have comments surrounded by parenthesis and should be cut in the order provided. They were made for a 1/4" (6mm) spiral bit and need at least 3mm of cutting depth per pass at a total of 16mm depth. Please examine the files before using them in case I am overlooking something.

https://raw.githubusercontent.com/nathanielstenzel/bucket_mill/master/example_output/owl-trace-A3mm.gcode
https://raw.githubusercontent.com/nathanielstenzel/bucket_mill/master/example_output/owl-final-A.gcode

I guess alternatively my problem could be with the serial port sender. Below is some Linux top command output.
3693 nathani+ 20 0 31700 28240 5512 S 101.0 0.5 29:52.47 ./serial-port-json-server -regex ACM
I am using version 1.80 of the serial-port-json-server.

I used the download link on Chilipeppr and it got me version 1.88. Maybe that will work better for me. I don’t know.

Well, I tried 1.88 and that was not working well at all. It was not tracking the position correctly and the movement was way off from what it was supposed to be.

+Justin Adie, what version of serial-port-json-server do you use with jpadie and which repository is it from?

I use 1.94 on Mac and 1.88 on windows. Both from the cp repo.

Are you milling in inches or mm?

The grbl widget does not manipulate the active gcode at all. So even if the widget is reporting the incorrect data the gcode being sent will still be accurate. So if the z is changing then it is the gcode or the controller, not the grbl widget.

Having said that there is some manipulation of gcode by some of the other widgets (comments and autolevelling changes). And by sjps (stripping comments and blank lines and the percent character). So CP as an overall solution is not a pure gcode sender in the way that Sonny defines it.

Capture the sjps logs (-v option) and take a note around the time that the z changes. That will give some diagnostic data that might be useful.

What hardware are you using? I know that the electronics based on the arduino mega2560 are not working well with CP. Steps are skipped and when it skips a Z lift command, the sh** really hit’s the fan.

I am doubtful that CP will work with a mega2560 without there being a change to sjps or a hardware mod.

Beyond that there should be no innate issue between CP and mega2560 as all CP does is send gcode a listen for a response from grbl.
Of course it may that the mega2560 port of grbl 1.x is not without bugs. No idea about that.

It is a Shapeoko2. My gcode cuts in mm. It has an Arduino Uno with grbl shield. The comments have parenthesis around them. I use Linux, so the Windows and Mac versions to use do not help as much. I suppose I could try LaserWeb to see if it cuts without going nuts. I suppose the enclosure could be flawed in such a way the the grbl shield gets cooled but the arduino does not. If so, maybe the arduino gets too hot.

did anyone notice that top command said that the seri port server was running at 101%? The thing is, that was a good 15+ minutes after doing any CNC milling with the CNC mill no longer plugged in. It was going crazy when it should have been idle.

sjps does do some code manipulation when there are no g commands on a line. it might be an idea to use fully formed lines of code.
or to push the code through UGS and see whether grbl likes it.
or we could write a trial flow control buffer for SJPS which does not do any gcode manipulation.

UGS = Universal Gcode Sender?
So “X20 Y30 Z-3” may be better than if it gets the parts only when it is not redundant code? I put in the gcode scrubbing to save space to help Chilipeppr not choke on the file size.
Any Linux users car to say which serial port server version they are using?

i meant use G0/G1 on every line. and add a feedrate if you can (I have noticed some odd behaviour in grbl1.1e when no feedrate is supplied).

yes UGS is universal gcode sender. (or any other app - just to give diagnostic colour to whether this is a SJPS issue or a grbl issue)

Okay. I can generate the gcode with G and F redundancies and scrub the rest instead.

worth a try.

also what 'duino are you using for grbl and what usb chip does it have?

The sad thing is it usually takes at least 10 minutes of cutting before it misbehaves on me. :frowning:

yes - been there too …

It has “gshield v5b” on it. It came in a kit from Instructables

that is the motor controller. before that you have arduino of some sort, that has grbl installed on it.

Oh right. Sorry.