Please help us to test firmware 1.17RC3 and DWC 1.14
-
(in reply to burtoogle) You have a few choices, such as:
1. Run G29 with no tape in place, just once. After that, use G29 S1 to load the height map file before printing. This will work well if your machine is stable and the height errors don't change.
2. Run G29 with the tape in place before each print. The height map will be wrong outside the taped areas, but that won't matter as long as you tape at least as far as the next grid point outside your print in each direction.
3. Decide on a few standard print areas to tape and probe. For each one, set up a macro to define a suitable grid with M557 and probe it with G29.
I will write a wiki page on using grid bed compensation in the next few days.
-
Wonderful addition to G29! Works without any problems on my delta. One thing - it may be slightly inconvenient that the height map also pops up when G29 is a part of a start script in an actual print. On the other hand, it's not like it's hard to just close the window and carry on.
-
It also pops up for G32. It makes it easier to level the bed with a three point macro. Just hit top view before rotating to better see the needed adjustments.
-
Well, the install went flawlessly. Great to see it just do what it is supposed to do.
Going for the grid calibration now. -
Hi David, many thanks for the instructions..
Machine is Kossel XL. Firmwares installed OK. I then started with M557 R110 S25 and a G29 with the homed height set to the distance to the glass and no tape installed. Adjusted the homed height to account for the tape thickness and tried printing a single layer of a test rectangle 110 x 95 mm which is well inside the probed area.
With the compensation turned off the print was light (too high) in the NW corner and down the E edge, could be better but not terrible.
With the compensation turned on it made quite a big difference but the results were not that good. The rectangle was now good and dark everywhere except the NE and especially the SE corners which were very light.
I thought perhaps I needed to probe more points so I tried again with 10mm spacing which produced a 21x21 grid.
Printing from that produced a very different result. Now it is quite dark everywhere except for two patches, one centred in the NW quarter and the other S of the middle. Those patches are approx 40mm long in the NE/SW direction and 20mm long in the NW/SE direction.
Here's the heightmap:
RepRapFirmware height map file v1 generated at 2016-11-22 10:12, mean error -0.02, deviation 0.13 xmin,xmax,ymin,ymax,radius,spacing,xnum,ynum -100.00,100.10,-100.00,100.10,110.00,10.00,21,21 0, 0, 0, 0, 0, 0, 0.051, 0.181, -0.214, 0.064, 0.052, 0.189, 0.186, 0.219, 0.111, 0, 0, 0, 0, 0, 0 0, 0, 0, 0, -0.046, -0.018, -0.024, 0.014, -0.006, -0.150, 0.175, 0.161, 0.161, 0.106, 0.127, 0.132, 0.122, 0, 0, 0, 0 0, 0, 0, -0.100, 0.302, 0.019, -0.005, 0.007, -0.038, 0.004, 0.129, 0.206, 0.298, 0.062, 0.230, 0.236, 0.236, 0.263, 0, 0, 0 0, 0, -0.057, -0.063, -0.158, 0.086, 0.033, -0.067, -0.121, 0.129, 0.100, 0.187, 0.102, 0.079, 0.113, -0.125, -0.126, 0.146, 0.112, 0, 0 0, -0.132, 0.020, -0.119, 0.006, 0.074, 0.064, -0.369, -0.274, 0.044, 0.042, 0.165, 0.180, 0.124, 0.199, 0.230, 0.164, -0.019, 0.199, -0.002, 0 0, -0.046, 0.025, -0.008, 0.066, -0.119, 0.075, -0.107, 0.027, 0.010, 0.098, 0.062, 0.167, 0.088, 0.113, 0.117, 0.030, -0.006, -0.031, 0.099, 0 -0.090, -0.034, -0.006, -0.199, 0.273, -0.039, -0.049, -0.282, -0.023, -0.029, 0.079, -0.046, 0.121, 0.056, 0.168, -0.006, 0.107, 0.239, -0.732, 0.089, -0.058 0.014, 0.121, -0.015, -0.086, -0.120, -0.075, -0.063, -0.045, 0.000, -0.070, 0.101, 0.100, -0.005, 0.196, 0.051, 0.187, 0.176, 0.019, 0.136, 0.062, 0.005 -0.037, -0.256, -0.057, -0.036, -0.030, -0.193, -0.198, -0.160, 0.055, 0.068, 0.052, 0.101, 0.044, 0.008, -0.157, 0.007, 0.092, 0.031, 0.199, 0.175, 0.101 -0.168, -0.042, -0.179, -0.009, -0.087, -0.188, 0.281, -0.085, 0.091, 0.098, 0.095, 0.144, 0.181, 0.057, 0.057, 0.146, 0.032, -0.051, -0.037, -0.055, -0.012 -0.189, -0.036, -0.088, -0.177, -0.288, -0.238, -0.060, -0.107, -0.044, -0.063, 0.032, 0.094, 0.030, -0.008, 0.013, 0.143, -0.075, -0.150, 0.077, 0.045, -0.026 -0.158, -0.102, -0.150, -0.182, -0.195, -0.351, -0.232, 0.007, -0.163, 0.018, -0.000, -0.045, -0.000, 0.074, -0.268, -0.357, 0.012, -0.084, -0.379, 0.552, -0.008 -0.088, -0.060, 0.037, -0.319, -0.164, -0.249, -0.124, -0.381, -0.106, -0.000, -0.008, -0.081, 0.069, 0.087, 0.063, 0.002, -0.031, 0.300, 0.081, -0.057, 0.070 -0.145, -0.055, -0.062, -0.044, -0.111, -0.112, -0.000, 0.018, -0.126, -0.036, -0.269, -0.280, -0.088, 0.045, -0.088, 0.026, 0.162, 0.088, -0.001, 0.000, -0.045 -0.098, -0.081, -0.079, -0.101, -0.320, -0.176, 0.027, -0.100, -0.286, -0.144, 0.075, 0.037, 0.144, 0.061, 0.102, 0.081, 0.056, 0.160, 0.106, 0.080, 0.076 0, -0.098, -0.133, -0.068, -0.118, -0.062, -0.038, -0.082, -0.112, -0.092, -0.070, -0.105, -0.044, -0.018, 0.012, 0.019, -0.189, 0.039, -0.068, 0.013, 0 0, -0.157, -0.106, -0.094, -0.069, -0.076, -0.057, -0.181, -0.046, -0.107, -0.015, 0.055, 0.299, -0.100, 0.049, -0.012, 0.096, -0.432, 0.131, -0.050, 0 0, 0, -0.126, -0.162, -0.099, -0.107, -0.101, -0.088, -0.070, -0.051, -0.039, -0.015, 0.017, 0.025, 0.026, 0.023, 0.013, -0.191, -0.121, 0, 0 0, 0, 0, -0.149, -0.105, -0.118, -0.092, -0.094, -0.064, -0.061, -0.039, 0.001, 0.006, 0.007, -0.092, -0.061, 0.045, -0.317, 0, 0, 0 0, 0, 0, 0, -0.195, -0.107, -0.095, -0.139, -0.130, -0.014, -0.068, -0.108, -0.054, -0.014, 0.042, -0.025, 0.018, 0, 0, 0, 0 0, 0, 0, 0, 0, 0.000, -0.102, -0.094, -0.061, -0.081, -0.108, 0.014, -0.104, 0.056, 0.049, 0, 0, 0.000, 0.000, 0.000, 0.000Any suggestions?
PS - thanks for the Cura gcode recognition tweak, works as expected.
-
Update…
On a whim, I decided to reverse the order of the lines in the heightmap file which (I assume) would have the effect of reversing the Y axis.
Interestingly, I got a much improved print with no light patches at all - looks a little bit low in a swathe from the SW corner to the N but, overall, a much better print than before.
Perhaps I should trying reversing the X values as well?
Surely the Y axis can't be inverted, can it?
-
I really don't think the axes are inverted. I suspect that your Z probe had a trigger height that varies with XY position. If you are using the IR sensor, or another type of probe that is separate from the nozzle, then on a delta this is usually caused by geometrical errors in your delta printer build causing the effector to tilt as it moves in the XY plane. The tilt changes the relative heights of the sensor and the nozzle.
-
Understood.
The sensor is offset from the nozzle by about 10mm in the NW direction. I find it hard to believe that the effector tilts sufficiently to create the amount of error I am seeing in just 10mm. I have measured the tilt before and did adjust the spacing of the arms to reduce the tilt. It doesn't tilt much now.
Is it possible to shift the map to compensate the for the offset of the height sensor from the nozzle?
-
Hi,
I have a macro in which there is :
[[language]] M120 G91 G1 Z-0.01 F6000 M121It gives an error when trying to execute it. M98 P0:/macros/test Push(): stack overflow!
also If I create this macro with DWC and call it "Z-0.01" it appears as "Z-0.01" in the macro list, but it appears as "Z-0" on the machine control tab
-
Hi David,
Thank you and Chrishamm for all your great work!
In the last weeks I've been getting Wiener90 (a Mendel90 derivative) up and running with DWF while playing catch up with the latest firmware releases since I bought my DWF and IR z-probe a few months back.
I'll probably not be able to feedback too much on 1.17RC3 before you release, but FWIW a couple of points:
1. Some of the release notes appear contradictory regarding G29:
line numbers copied from https://github.com/dc42/RepRapFirmware/blob/dev/WHATS_NEW
[[language]] 37 - Implemented G29 S1 (load bed height map from file) ... 45 - G29 grid bed compensation is fully implemented except for G29 S1 (load height map from file) ... 50 - G29 can now be used to do bed probing, however the results are not used yet2. General feedback: In DWC 1.13, Extruder Control > Feedrate (mm/s)
While calibrating my 2.85mm dia extruder running an E3Dv6 hotend I subsequently discovered a) I only had a 30W heater cartridge installed(!), and b) this necessitated a feed rate less than the minimum value available (5mm/s). The only way of achieving that was to temporarily reduce the max E feedrate via the gcode console/config.g. Perhaps after I upgrade to a 40W heater cartridge a 5mm/s E feedrate will be achievable. Nevertheless, that's still a significantly faster filament feedrate for 2.85mm dia material than would normally be used at a print speed of say 50mm/s using a 0.4mm dia nozzle.I look forward to being of more help in the new year - including writing a post about using DuetWifi with Wiener90 as soon as I get the early-day gremlins sorted.
Cheers, and Merry Christmas to everyone on the Duet3D dev team.
-
I like the visual representation after G29.
is it be possible to display it from the existing heightmap.csv (without running G29 again) ? -
M98 P0:/macros/test Push(): stack overflow!
Hmm it looks like it's not related to the macro.
I just tried to issue a G28, the printer does home.
and just after that i issued a G32, here's the result:[[language]] G32 Push(): stack overflow! Attempt to move the head of a delta printer before homing the towers Error: Must home a delta printer before bed probing Error: Must home a delta printer before bed probing Error: Must home a delta printer before bed probing Error: Must home a delta printer before bed probing Error: Must home a delta printer before bed probing Error: Must home a delta printer before bed probing Error: Must home a delta printer before bed probing Error: Must home a delta printer before bed probing Error: Must home a delta printer before bed probing Error: Must home a delta printer before bed probing Error: Must home printer before G29 bed probing Attempt to move the head of a delta printer before homing the towers -
Stack overflow usually means that the macro you are running is directly or indirectly recursive. For example, if your bed.g file contains a G28 command, and your homedelta.g file contains a G32 command.
-
hmm what's wrong with my test macro ? it was working before with 1.16.
Also, the multiple stack overflow errors happened without any changes.
i powered the printer, did upload both firmware and DWC. did a G29, did an emergency stop, did a G29 again.
G32 used to work before.and now G32 now says : Macro file homedelta.g not found.
I have not cold reset my board, but I'm pretty sure if I do it will work again…
(I'm not cold reseting because I had an issue with rc1. using DWC the printer stopped responding after a successful print, I did cold reset and it worked again, I guess I should have tried to connect to usb and try to debug the issue) -
@lolorc - yes you can do that:
Settings->Machine Properties "Download Heightmap"
actually shows the map in the visualizer. Seems like a poor choice of label for this button! Should be Show Heightmap or something similar I think. You can download the heightmap, look at the raw data and even tweak it using Settings->System Editor. I accidentally probed a bed clip so I just edited that one value, very convenient.
-
@dc42 - is Chrishamm's code available anywhere? He hasn't updated his repository with the 1.14-b4 stuff yet. I'd like to create a free-standing heightmap visualizer using his code to make it easy to compare and look at archived heightmaps.
-
@lolorc - yes you can do that:
Settings->Machine Properties "Download Heightmap"
actually shows the map in the visualizer. Seems like a poor choice of label for this button! Should be Show Heightmap or something similar I think. You can download the heightmap, look at the raw data and even tweak it using Settings->System Editor. I accidentally probed a bed clip so I just edited that one value, very convenient.
Oh, i see, thanks
-
So I put some tape on the bed to the S of the centre line and did a scan, here's a picture of the result, I have rotated the picture but you can see the axis arrows and confirm that the huge block is to the S of the centre. It does not extend as far as the E of the bed. BTW - that gives a good indication of how much error the bed has.
Here's a photo of the print of my test rectangle looking in a N direction. Note that the bad print is all on the E side of the rectangle (which did not have any tape to raise the bed height). The S side of the rectangle falls into the region that had been taped and I would have expected that to be badly printed but it's actually fine.
If I didn't know better, I would say that the heightmap has been rotated through 90 deg counter clockwise!
-
Does the ratty right side of your print correspond to the big red peaks in the plot?
-
hmm what's wrong with my test macro ? it was working before with 1.16.
Also, the multiple stack overflow errors happened without any changes.
i powered the printer, did upload both firmware and DWC. did a G29, did an emergency stop, did a G29 again.
G32 used to work before.and now G32 now says : Macro file homedelta.g not found.
I have not cold reset my board, but I'm pretty sure if I do it will work again…
(I'm not cold reseting because I had an issue with rc1. using DWC the printer stopped responding after a successful print, I did cold reset and it worked again, I guess I should have tried to connect to usb and try to debug the issue)cold reset did not fix the stack overflow issue with the simple macro.
I had to revert back to 1.16 (2016-11-08)
all 1.17 version (dev8, rc1 and rc3) have the issue.
