IDEX mirror mode

kazolar

I am trying to get my dual gantry quad extruder to print in mirror mode on the Y axis -- have 2 Ys -- Y and U.
I did the same as duplication -- then set M579 U-1 to flip U. That works, but sorta. The Y then jumped from 0 to something like 140 -- my origin in the lower left, so I don't use 0,0 as center, so I assume that's why this behavior is weird. I then force Y back to 0.3 with G92 -- that works, and the print does appear to work, but the primary tools starts floating the print along both axis evenly, same amount of shift -- not coincidental -- the part was on an angle and for both X and Y to shift with each layer is just weird. The mirrored part is printing fine, I am assuming this is happening because I forced Y to 0.3 with G92 prior to start of the print, but why does the flip of U have anything to do with the Y -- it shouldn't logically (at least not how I understand it's supposed to work)

EDIT:
Turned off mirror -- just pure duplication on the same 2 toolheads -- no drifting (didn't need to adjust Y obviously)

kazolar

@dc42 please help. I can't see how to do mirror printing without reconfiguring the machine from scratch to have 0,0 at center, which is very complicated.

dc42

I recommend you reconfigure your machine to have 0,0 at the centre. It's not that complicated, see https://duet3d.dozuki.com/Wiki/Centering_the_bed_or_setting_the_bed_origin.

kazolar

@dc42 I have 2 y gantries 4 x gantries. It's a rather complicated to keep track of the coordinate system as is. I'll dig into the firmware to see why it's shifting for mirroring. Deltas have 0,0 at center, everything else doesn't. This doesn't appear to be a feature but a bug based on an assumption. I honestly have never seen a cartesian machine reference anything else than front left as 0,0. I understand the benefits, though they don't outweigh the unnecessary complexity. Nothing I slice for the quad would be applicable anywhere else. Different size nozzles, different extrusion miltiplier, different number of extruders.

dc42

All 4 of my printers have 0,0 at bed centre. Two of them are Cartesian. For Idex is makes particular sense to put 0,0 at bed centre, that way you slice all your models centred on 0,0 regardless of whether you are printing one multi-material part, or two duplicate or mirrored parts.

kazolar

@dc42 unfortunately that's not how I setup the machine. I'd have to redo not just config.g -- which would be fairly straight forward, but I'd have to change every pre and post g file to adjust their coordinate system. When I have 5 hours to go through every file to do that, I'll try and make sense of it. For now I won't use the mirror feature since it won't work. If I can find in firmware where the math to do this happens, I'll just fix it for my case.

kazolar

@dc42 what's weird is that it doesn't appear to assume 0,0 entirely.

this my tool definition
G10 P6 X0 Y0 U-296 S0 R0

after I home everything (except Z)
M114
X:-34.300 Y:0.000 Z:0.000 U:588.100 V:639.840 W:641.500 A:-34.000 E0:0.0 E1:0.0 E2:0.0 E3:0.0 Count -1715 0 0 29405 31992 32075 -1700 Machine -34.300 0.000 0.000 588.100 639.840 641.500 -34.000
then I do
M579 U-1
T6
M114
M114
X:640.670 Y:-146.050 Z:0.000 U:-292.100 V:639.840 W:641.500 A:-34.000 E0:0.0 E1:0.0 E2:0.0 E3:0.0 Count -1715 0 0 29405 31992 32075 -1700 Machine -34.300 0.000 0.000 292.100 639.840 641.500 -34.000
U is set to I assume total U * -1 + offset.
But Y is set to half that -- I don't understand why that is.

kazolar

@dc42 i set it differently for mirror to try it out

G10 P6 X0 Y0 U-588.1 S0 R0 -- basically set U offset to be all the way max U
and now it works! It's weird but this works fine