[Grbl_Esp32 Issue#673] Uncommanded A-axis movements

What version of the firmware are you using?
[MSG:Grbl_ESP32 Ver 1.3a Date 20201101]
[MSG:Compiled with ESP32 SDK:v3.2.3-14-gd3e562907]

Is the problem repeatable?
Yes

Under what conditions does the bug occur?
Running g-code program with A-axis movements in the code.

When running a program with A axis movements, the A axis gains additional degrees beyond those commanded during the program and in excess of acceptable rounding/quantization errors. These tend to occur all at once at some point in the program. It appears to be most repeatable in the lines immediately following the first A axis move in a program.

I have also been able to observe movement of the A axis when manually typing gcode commands to move the Z axis. Again, this appears only to happen when the A axis has been moved from its “pure” zero position. Indeed, if I run a program with no A axis moves in it after a hard reset, there is no movement of the A axis from “0.0”. But if I jog the A axis at all (say forward and back 1 degree) so that it calculates the “true” position (almost never returning to 0.0 after jogging at all) and run the same program, the A axis moves without being commanded.

I’ve seen the A-axis jump anywhere from a single step to 8 degrees! The error is reflected at both the DRO and the physical axis on the machine. Once this error occurs, it carries through the rest of the program even after subsequent movements of the A axis. As far as I can tell, it only occurs once in a given program. See annotations in the serial log from a run of a test program below where the axis moved 0.478 (17 steps) without being commanded.

Observed same effect with code run from two different streamers (bCNC and GrblGru).

Serial log showing an occurrence of the bug. (“ok” lines removed):
“
$G
[GC:G0 G54 G17 G21 G90 G94 M5 M9 T0 F0 S0]
<-------- ? to report position at start of program ? G90 G90G94 G17 G21 G28G91Z0 G90 S12031M3 G4P4 G54 G90 G1A0F3000 <-------- No movement. 0.028 on DRO (acceptable error expected due to step quantization) G0X41.679Y-1.185 Z55.8 <-------- Observed uncommanded additional 0.478 degrees, displaying 0.506 on DRO here. G1X40Y-1.185F1500 G1A90F3000 <-------- Moved to 90.506 on DRO G0X41.679Y-1.185 Z55.8 G1X40Y-1.185F1500 G1A180F3000 <-------- Moved to 180.506 on DRO G0X41.679Y-1.185 Z55.8 G1X40Y-1.185F1500 G1A270F3000 <-------- Moved to 270.506 on DRO G0X41.679Y-1.185 Z55.8 G1X40Y-1.185F1500 G0Z55.8 G28G91Z0 G90 M5 M3 <------- Actually sent M30, but echos M3 for some reason. Probably unrelated. Run ended 2020-11-18 22:42:02.861000 Current: 32 [32] Completed: 100% [12s Tot: 12s ] Grbl 1.3a ['$' for help] <-------- "?" to report position at end of program. Confirms 0.478 degree error from last commanded A position. ? g1 a270 f3000 <-------- manually commanding the last A axis move from the program while in the idle state causes the axis to move as expected. Axis moved 0.478 degrees to commanded position. DRO showed 270.028. $G [GC:G0 G54 G17 G21 G90 G94 M5 M9 T0 F0 S0] <-------- ? to report position after sending above command manually. Confirms axis moved and error is gone. ? `

The machine settings are as follows:
`
$0=3
$1=250
$2=0
$3=1
$4=0
$5=1
$6=1
$10=1
$11=0.010
$12=0.002
$13=0
$20=0
$21=0
$22=1
$23=1
$24=200.000
$25=800.000
$26=250.000
$27=1.000
$30=24000.000
$31=1836.000
$32=0
$N1=
$N0=
$100=320.000
$101=320.000
$102=320.000
$103=35.556
$104=100.000
$105=100.000
$110=3000.000
$111=3000.000
$112=3000.000
$113=3000.000
$114=1000.000
$115=1000.000
$120=300.000
$121=300.000
$122=150.000
$123=200.000
$124=200.000
$125=200.000
$130=275.000
$131=397.500
$132=116.000
$133=1440.000
$134=300.000
$135=300.000
`

Test program:
`
G90 G94
G17
G21
G28 G91 Z0
G90

T19 M6
S12031 M3
G4 P4.
G54
G90
G1 A0 F3000
G0 X41.679 Y-1.185
G0 Z55.8
G1 X40 Y-1.185 F1500

G1 A90 F3000
G0 X41.679 Y-1.185
G0 Z55.8
G1 X40 Y-1.185 F1500

G1 A180 F3000
G0 X41.679 Y-1.185
G0 Z55.8
G1 X40 Y-1.185 F1500

G1 A270 F3000
G0 X41.679 Y-1.185
G0 Z55.8
G1 X40 Y-1.185 F1500

G0 Z55.8
G28 G91 Z0
G90
M5
M30
``