Sharp 90* turn in gcode becomes a large sweeping turn

2.7.15

can you post your gcode or a portion of it that exhibits this issue?

Apologies for the fairly large file, I'm not at my controller computer, so I can't tell you exactly which lines the problem occurs on.
The toolpath is in three passes, and as the photo of the result shows, pass #1 does not superimpose pass #2 or #3.

I changed the max acceleration value to 17. Re-ran the file (no load). The track shown in the preview window conformed to the path plotted (i.e performed as expected)
It would be nice to know if that was the problem, or if it's still lurking.

I have also been getting frequent random shutdowns ("estop reset" is the status) with no error messages. I had written it off as a loose connection... Could it be related?

I cannot duplicate this with either 2.7.15 or 2.9

LinuxCNC 2.7:

LinuxCNC 2.9:


both with accel of 12 ips/s

Not sure about the estop issue, that sounds like hardware to me,
I would put a debounce in this connection

net estop-ext <= parport.0.pin-10-in

My ini file settings. I will be re-doing this shortly as I move to new DM542T drivers. The settings below were from the previous owner.

Y
MAX_VELOCITY = 3
MAX_ACCELERATION = 4.0
STEPGEN_MAXVEL = 7.21
STEPGEN_MAXACCEL = 5.00


Z
MAX_VELOCITY = 1.0
MAX_ACCELERATION = 5.0
STEPGEN_MAXVEL = 1.25
STEPGEN_MAXACCEL = 6.25


X
MAX_VELOCITY = 2.5
MAX_ACCELERATION = 10.0
STEPGEN_MAXVEL = 2.92
STEPGEN_MAXACCEL = 12.50

Unless you are using G64 Pxxx, rounded corners are _expected_ so its not likely your INI
settings are involved (other that the fact that you have very low acceleration settings so you
will get very large radius blends with plain G64). Did you try G64 P0.0005 as cmorely suggested?

Thank you for the troubleshooting help.

Can you please advise me about debouncing pin 10?

I think I need to replace this wscnc.hal code
with

G64 P0.0005 resolved the sweeping 90* turn.

So I am still a CNC newb. When I am in the CAM phase, how would I detect such an issue in the future as to reduce the likelihood that I destroy valuable material when running the gcode? At this point in my learning (about 4 months in), I feel like every other new design I take to the CNC destroys a piece of material. Each time I learn something new, but it is getting a bit discouraging that it feel rare to generate gcode, execute it, and have the piece I designed. Maybe this is part of the hobby cnc router process?

Probably best way is to have your CAM program insert a gcode preamble that includes
important modal gcodes like G64 Pxxx.

Also with new code and expensive material, take time to study the preview,
or run it in simulation to see if its doing what you expect or perhaps
do an run on scrap material