duet auto-level result not making sense

Here is the message queue from my actions (browse interface for Duet).
First I run auto-cal with good results, but when I run the exact same bed.g macro with the last M30 command using S-1 to just see the actual offsets, I get large errors that don't make sense.
The large errors are the real ones when the nozzle is touching the head, so the auto-level says it all looks good, 'cept it isn't.
(Read from bottom up)

5:38:51 PMM98 P/macros/bed_justreport.g
Bed probe heights: -0.717 -0.486 -0.092 0.063 -0.079 -0.520 -0.640 -0.172 -0.227 -0.323, mean -0.319, deviation from mean 0.248

5:36:36 PMG32
Calibrated 6 factors using 10 points, deviation before 0.080 after 0.020

5:32:53 PMM98 P/macros/bed_justreport.g
Bed probe heights: -0.649 -0.431 -0.003 0.010 -0.040 -0.471 -0.582 -0.032 -0.190 -0.261, mean -0.265, deviation from mean 0.239

5:08:10 PMG32
Calibrated 6 factors using 10 points, deviation before 0.034 after 0.014

Edited 2 time(s). Last edit at 08/22/2015 09:27PM by shadowphile.

I agree, those figures don't make sense. The "deviation before" figure in the G32 command at 5:36:36 doesn't bear much relation to the mean and deviation figures from the bed probing at 5:32:53.. Please can you:

1. Attach your sys/bed.g and macros/bed_justreport.g files;

2. Confirm visually that both files are probing the same 10 points;

3. Attach Pronterface, send M111 S1 P4, run bed_justreport.g, then run G32. Then post the output captured by Pronterface.

Edited 1 time(s). Last edit at 08/23/2015 03:15AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

For some reason I have to disconnect the LAN if I reset the board or Pronterface won't connect,
error message: Connecting...could not open port COM9: [Error 2] The system cannot find the file specified.
I am running the recent RC firmware.
Probe points appear to be close to the same.
I also notice that after every probe the head is jerked upward very hard, more like a snap. I have the max speed in config file set to 8000 which works in other moves. That snap seems unnecessarily hard and pushes the limits of my system.
In fact, it even broke away once and started probing way up high so something got messed up in the servo vs tracked position and I had to do an emergency reset.

Anyway, attached are my two files (I threw in the config file too).

Here is pronterface output

>>>M111 S1 P4
SENDING:M111 S1 P4
Debugging enabled for modules: Move(4)
Debugging disabled for modules: Platform(0) Network(1) Webserver(2) GCodes(3) Heat(5) DDA(6) PrintMonitor(7)
Starting network
Network up, IP=192.168.1.14
Bed probe heights: -0.759 -0.525 -0.203 -0.030 -0.121 -0.597 -0.678 -0.216 -0.311 -0.406, mean -0.385, deviation from mean 0.235
>>>g32
SENDING:G32
  Z probe offsets: -0.159 -0.095 -0.183 -0.090 0.009 -0.147 -0.188 -0.166 -0.091 -0.146, mean -0.126, deviation from mean 0.057
  Endstops X-1.45 Y0.41 Z1.03, height 245.07, diagonal 201.20, radius 111.49, xcorr 0.00, ycorr 0.00, zcorr 0.00
  Derivative matrix
    0.767   0.117   0.117   0.451   0.000   0.089
    0.523   0.523  -0.047   0.326  -0.377   0.377
    0.094   0.854   0.052   0.339  -0.100   0.045
   -0.042   0.526   0.517   0.358   0.001  -0.356
    0.117   0.117   0.766   0.451   0.089  -0.089
    0.508  -0.015   0.507   0.444   0.291  -0.000
    0.648   0.186   0.166   0.554  -0.013   0.093
    0.164   0.619   0.217   0.573  -0.068  -0.032
    0.222   0.222   0.556   0.610   0.073  -0.073
    0.333   0.333   0.333   0.666   0.000   0.000
  Normal matrix
    1.751   0.809   0.773   1.621  -0.052   0.313   0.477
    0.809   1.885   0.787   1.562  -0.304   0.030   0.476
    0.773   0.787   1.626   1.589   0.252  -0.289   0.304
    1.621   1.562   1.589   2.404   0.011  -0.001   0.610
   -0.052  -0.304   0.252   0.011   0.255  -0.159  -0.019
    0.313   0.030  -0.289  -0.001  -0.159   0.301   0.033
  Solved matrix
    1.751   0.000   0.000   0.000   0.000   0.000   0.280
    0.000   1.511   0.000   0.000   0.000   0.000   0.306
    0.000   0.000   1.162   0.000   0.000   0.000  -0.208
    0.000   0.000   0.000   0.111   0.000   0.000   0.015
    0.000   0.000   0.000   0.000   0.091   0.000   0.029
    0.000   0.000   0.000   0.000   0.000   0.075  -0.006
  Solution:   0.160   0.203  -0.179   0.131   0.320  -0.081
  Residuals:   0.018  -0.006   0.005  -0.007   0.009  -0.008  -0.015   0.002  -0.001   0.003
  Expected probe error:   0.018  -0.005   0.004  -0.006   0.010  -0.007  -0.015   0.002  -0.001   0.003
  Derivative matrix
    0.767   0.116   0.117   0.451   0.001   0.088
    0.522   0.523  -0.045   0.328  -0.376   0.376
    0.093   0.854   0.052   0.339  -0.100   0.044
   -0.042   0.528   0.514   0.361   0.001  -0.356
    0.119   0.118   0.763   0.456   0.089  -0.088
    0.510  -0.015   0.505   0.446   0.291  -0.000
    0.649   0.185   0.166   0.555  -0.013   0.093
    0.164   0.620   0.216   0.573  -0.069  -0.031
    0.223   0.223   0.554   0.613   0.072  -0.072
    0.334   0.334   0.332   0.667  -0.001   0.001
  Normal matrix
    1.754   0.809   0.776   1.629  -0.050   0.312  -0.001
    0.809   1.890   0.787   1.567  -0.304   0.028  -0.001
    0.776   0.787   1.612   1.592   0.250  -0.285  -0.002
    1.629   1.567   1.592   2.420   0.011  -0.000  -0.001
   -0.050  -0.304   0.250   0.011   0.254  -0.158  -0.000
    0.312   0.028  -0.285  -0.000  -0.158   0.300   0.000
  Solved matrix
    1.754   0.000   0.000   0.000   0.000   0.000  -0.001
    0.000   1.517   0.000   0.000   0.000   0.000  -0.000
    0.000   0.000   1.147   0.000   0.000   0.000  -0.002
    0.000   0.000   0.000   0.110   0.000   0.000   0.000
    0.000   0.000   0.000   0.000   0.091   0.000   0.000
    0.000   0.000   0.000   0.000   0.000   0.075   0.000
  Solution:  -0.001  -0.000  -0.002   0.001   0.001   0.000
  Residuals:  -0.159  -0.096  -0.183  -0.091   0.008  -0.147  -0.188  -0.166  -0.092  -0.146
  Expected probe error:   0.018  -0.005   0.005  -0.007   0.009  -0.008  -0.015   0.002  -0.001   0.003
  Endstops X-1.35 Y0.56 Z0.79, height 245.33, diagonal 201.20, radius 111.62, xcorr 0.32, ycorr -0.08, zcorr 0.00
.

Quote
shadowphile
For some reason I have to disconnect the LAN if I reset the board or Pronterface won't connect,
error message: Connecting...could not open port COM9: [Error 2] The system cannot find the file specified.

There appears to be a bug in Pronterface, or the the Arduino Due device driver, or some other part of the Windows driver stack, that has the following effect. If you load Pronterface, connect to the Duet, and then reset the Duet or power it down/up without first disconnecting Pronterface, then the USB port doesn't get closed properly and you are unable to reconnect with Pronterface. To avoid the problem, always press the Disconnect button in Pronterface before resetting or powering down the Duet. If you forget, then you generally need to exit and reload Pronterface to recover. You may also need to disconnect and reconnect the USB cable too.

It's late here so I'll look at your debug output tomorrow.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I need to get out of bed earlier!
BTW, I found a scrap of junk text in the config file but I deleted it and reran auto-level + justreport.g and it didn't make a difference.

more diagnostic feedback:
-I removed the homing command in the justreport.g file and run it after a calibration run. It reports expected values. If I home the machine after a cal run, then do another justreport.g run, the numbers are back to really bad. So the homing routine seems to be the culprit. I can repeat this behavior.
The limit-switches are good and they seem solid in action and mounting.
I am also running release e because I retro-graded to see if these problems were firmware related. There was no change.

-Occasionally, while probing the bed height pattern the head suddenly jumps down onto the bed and scrapes along. It happens very fast but I get the impression that it happens during the retract after a probe is finished. I followed repeat runs with a video camera but of course it wouldn't fail when I need it to.
I was about to ask you why the retract is so hard, more like a bang on my machine. Certainly more than necessary and rather hard on everything. I wouldn't be surprised it is exceeding my electro-mechanics and losing position. I set my max speeds to 3000 in the config file but it it seems to have no affect.

Edited 4 time(s). Last edit at 08/24/2015 01:16AM by shadowphile.

Clearly the printer is either not homing consistently, or (more likely) is losing steps somewhere during the auto cal procedure. I suggest the following:

1. Use version 1.09g-rc1 of my firmware fork.

2. Reduce your M201 acceleration values to 2000. For now, also reduce your M203 XYZ speeds to 5000 or lower and the M566 jerk speeds to 600.

3. In your homedelta.g file, use an initial homing speed of 2500 or lower.

4. Add command M210 Z200 to config.g to define the initial Z probing speed. You can increase this to Z500 later if you are using my IR probe.

If that gets is working (i.e. after autocal and then homing, bed_justreport reports low deviations), then you can try increasing the M201/M203/M566 speeds again and see which one is critical.

EDIT: also I am assuming that in your homedelta.g file, you have the XYZ parameters in the initial G1 S1 command high enough so that the carriages always reach the switches, no matter what the start position. Please check.

Edited 2 time(s). Last edit at 08/24/2015 03:20AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

thanks, that fixed it right up, get very tight numbers now.