Fisher SD image 210715

Hello everyone,

It is great to see so many of you successfully completing the build of your Fisher kits and getting useful prints out. The feedback we have so far received has been tremendously helpful, and the instructions have already improved a great deal based on this. The process is of course not yet complete, and look out for many more updates over the coming days.

As we continue to test and tweak our settings, we are making small improvements to the configuration files we provide. Some people have reported issues with setting the height of Z zero relative to the bed. Along with some advice added to the troubleshooting page, we have published a new SD image which seems to work well from a cold start. Until I organise our github repositories, you will find the updated zip file within the commissioning instructions here [reprappro.com]

Improvements include better height calibration when running 'Auto bed compensation.' From a cold start, simply home all axes, run the auto bed compensation, then hit print. I have also updated the filament load and unload macros.

Thanks again for all of your great feedback, and if you get any problems with the build or use of the Fisher, please do get in touch but remember that all warranty issues are best dealt with via email to support (at) reprappro (dot) com

Happy printing!
Jean-Marc.

My first feedback after some prints is that the bed calibration is good, I had always various distances with the latest, now after setting M665 and M666 based on the last script, my auto alignment is now working good
G32 Calibrated 4 factors using 4 points, deviation before 0.322 after 0.000

I had no specific problem with load / unload filament, so there is no change for me.

Thanks for the improvement.

---

Greg
3D Newbie

Quote
Greg_be
My first feedback after some prints is that the bed calibration is good, I had always various distances with the latest, now after setting M665 and M666 based on the last script, my auto alignment is now working good
G32 Calibrated 4 factors using 4 points, deviation before 0.322 after 0.000.

The deviation after will always be reported as zero if the number of factors being calibrated is equal to the number of probe points, To get a better idea of how flat the printing plane is, add more probe points to the bed.g file, or set up a separate bedcheck.g file. See [forums.reprap.org] for more.

---

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].

At the end of bed4 there is one line with G31 Z0.1
In the bed4meas, the same line is there but with negative Z value (G31 Z-0.1)

Is there any reason to invert the value or which one is the correct one ?

Edited 2 time(s). Last edit at 07/24/2015 06:20AM by Greg_be.

---

Greg
3D Newbie

With 7 points like you propose, I got these values.
Is it looks good ?

Endstop adjustments X0.94 Y0.27 Z-1.21
Diagonal 160.00, delta radius 82.64, homed height 180.44, bed radius 75.0, X 0.0°, Y 0.0°

Bed probe heights: 0.040 -0.040 -0.555 -0.196 -0.207 -0.459 -0.149, mean -0.224, deviation from mean 0.198
Bed probe heights: 0.029 -0.040 -0.555 -0.196 -0.196 -0.447 -0.149, mean -0.222, deviation from mean 0.194
Bed probe heights: 0.040 -0.040 -0.555 -0.196 -0.184 -0.447 -0.149, mean -0.219, deviation from mean 0.197
Bed probe heights: 0.040 -0.040 -0.543 -0.196 -0.184 -0.447 -0.149, mean -0.217, deviation from mean 0.194
Bed probe heights: 0.051 -0.017 -0.555 -0.184 -0.184 -0.447 -0.149, mean -0.212, deviation from mean 0.202

Calibrated 7 factors using 7 points, deviation before 0.360 after 0.072

Thanks

Edited 1 time(s). Last edit at 07/24/2015 06:50AM by Greg_be.

Quote
Greg_be
With 7 points like you propose, I got these values.
Is it looks good ?

Endstop adjustments X0.94 Y0.27 Z-1.21
Diagonal 160.00, delta radius 82.64, homed height 180.44, bed radius 75.0, X 0.0°, Y 0.0°

Bed probe heights: 0.040 -0.040 -0.555 -0.196 -0.207 -0.459 -0.149, mean -0.224, deviation from mean 0.198
Bed probe heights: 0.029 -0.040 -0.555 -0.196 -0.196 -0.447 -0.149, mean -0.222, deviation from mean 0.194
Bed probe heights: 0.040 -0.040 -0.555 -0.196 -0.184 -0.447 -0.149, mean -0.219, deviation from mean 0.197
Bed probe heights: 0.040 -0.040 -0.543 -0.196 -0.184 -0.447 -0.149, mean -0.217, deviation from mean 0.194
Bed probe heights: 0.051 -0.017 -0.555 -0.184 -0.184 -0.447 -0.149, mean -0.212, deviation from mean 0.202

Thanks

I presume that's the output from running bedcheck.g before you ran auto calibration. Those results are nicely consistent, but your endstop corrections are obviously out. Try running auto calibration once or twice, then run bedcheck.g again and see what you get. Then run 6-factor auto calibration (like your bedcheck.g file but with S6 on the final G30 command instead of S-1), and see if that improve things.

---

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].

My auto calibration has been replaced with the 6-factor macro.

Calibrated 6 factors using 7 points, deviation before 0.356 after 0.038
Calibrated 6 factors using 7 points, deviation before 0.200 after 0.099
Calibrated 6 factors using 7 points, deviation before 0.103 after 0.102

Diagonal 160.00, delta radius 82.57, homed height 181.34, bed radius 75.0, X -1.0°, Y 0.4°
Endstop adjustments X0.57 Y0.23 Z-0.79

Bed probe heights: -0.002 0.014 0.050 -0.127 -0.105 -0.112 0.184, mean -0.014, deviation from mean 0.104

Are the endstops better ?

Edited 1 time(s). Last edit at 07/24/2015 07:30AM by Greg_be.

Result after setting M665 and M666 from the last run

Quote

Diagonal 160.00, delta radius 82.57, homed height 181.34, bed radius 75.0, X -1.0°, Y 0.4°
Endstop adjustments X0.57 Y0.23 Z-0.79

Calibrated 6 factors using 7 points, deviation before 0.590 after 0.048
Calibrated 6 factors using 7 points, deviation before 0.208 after 0.096
Calibrated 6 factors using 7 points, deviation before 0.101 after 0.100

Diagonal 160.00, delta radius 82.59, homed height 181.34, bed radius 75.0, X -1.6°, Y -0.1°
Endstop adjustments X0.53 Y0.14 Z-0.66

Bed probe heights: 0.030 0.022 0.049 -0.094 -0.094 -0.101 0.199, mean 0.001, deviation from mean 0.101

Hello,
David, what shall we do with the column angles reported by the 6 factors test ?
Could they be reintroduced permanently in the parameters ?
Note that 7 point calibration was removed in that last image supplied by RRP.

Regards, Pierre

Quote
Greg_be
Result after setting M665 and M666 from the last run

Quote

Diagonal 160.00, delta radius 82.57, homed height 181.34, bed radius 75.0, X -1.0°, Y 0.4°
Endstop adjustments X0.57 Y0.23 Z-0.79

Calibrated 6 factors using 7 points, deviation before 0.590 after 0.048
Calibrated 6 factors using 7 points, deviation before 0.208 after 0.096
Calibrated 6 factors using 7 points, deviation before 0.101 after 0.100

Diagonal 160.00, delta radius 82.59, homed height 181.34, bed radius 75.0, X -1.6°, Y -0.1°
Endstop adjustments X0.53 Y0.14 Z-0.66

Bed probe heights: 0.030 0.022 0.049 -0.094 -0.094 -0.101 0.199, mean 0.001, deviation from mean 0.101

I'm surprised to see that the error is so high at the start, and still take 2 calibration iterations to converge. Are you sure you entered the correct M665 and M666 values in config.g? Have they converged on the same values?

I guess the other possibility is that the tower angle corrections are important - see my next post.

---

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].

Quote
PRZ
Hello,
David, what shall we do with the column angles reported by the 6 factors test ?
Could they be reintroduced permanently in the parameters ?

Currently my firmware fork does not support entering the tower angle corrections in the M665 command. However, it has been my intention to add X and Y parameter to set them - that's why I report the tower positions in terms of delta radius and X and Y angle adjustments. But I don't get paid for the firmware changes I make, and it may be a while before I find time to implement it.

---

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].

Hi

Has anyone got a copy of bed7.g I could look at? as PRZ said, it has been removed from the latest Fisher image.
cheers

Quote
jmg123
Hi

Has anyone got a copy of bed7.g I could look at? as PRZ said, it has been removed from the latest Fisher image.
cheers

Here's a copy of the previous Fisher image, with the firmware binary removed to meet the forum size limit. It includes bed7.g in the macros folder. But as I said before, I don't advise 7-factor calibration any more. The least squares solution matrix becomes somewhat ill-conditioned when solving for the diagonal rod length, which means that small differences in the bed probe height readings cause relatively large jumps in the solution for diagonal rod length. It's better to leave the diagonal rod length alone, unless you need to adjust it to get the correct the X and Y dimensions of your prints.

One of these days I'll write a document on how to use the delta printer calibration facility in my firmware fork. For now, here are the essentials:

- In bed.g you define N bed probe points (where N <= 16), using G30 commands with the P parameter ranging from 0 to N-1.
- On the final G30 command you add parameter Sm where m is -1, 0, 3, 4, 6 or 7.
- m=0 is equivalent to m=N, and m=-1 means just display the height errors bit don't change anything.
- Otherwise, the firmware will adjust m factors. m=3 adjusts the endstop corrections, m=4 does that + the delta radius, m-6 does all that + the tower positions, and m=7 does all that + the diagonal rod length.
- If N < m then there are too few data points and an error is reported. If N = m then there are just enough points, so the expected residual height errors will be reported as zero (because the firmware is solving N equations for the same number of unknowns). If N > m then the firmware will minimise the sum of the squares of the height errors.

If a delta is built very accurately then it should only be necessary to do 4 factor calibration. If the tower positions do not exactly form an equilateral triangle, then you need 6 factor calibration. If the towers lean slightly, then 6-factor calibration will work quite well, although it will not compensate perfectly for the leaning towers.

My usual recommendation is to set N=7 or 10 and m=6 for small delta printers, and N=13 and m=6 for large printers. RepRapPro seems to have settled on N = m = 4 for the Fischer, which is OK for a small printer if the design ensures an accurate build. I don't have a Fisher so I can't check this. To test the calibration of your printer, head over to thingyverse and download and print one of the calibration spirals or 6-pointed delta printer calibration shapes. With 4-factor calibration, the print height along radials from the centre to each of the towers should have a consistent height, but the print height along radials from the centre to points midway between towers will only be good if the build is accurate.

If the initial machine parameters set in the M665 and M666 commands are close to the true values, then one calibration cycle is sufficient. Otherwise, more than one may be needed. Therefore, I recommend running several calibration cycles initially, then read off the resulting M665 and M666 parameters and copy them into config.g,

HTH David

Edited 2 time(s). Last edit at 07/25/2015 05:53PM 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].

Thank you for the file, and all that useful info your reply.

I shall give some 6 factor calibration cycles ago until M665 and M666 settle down and then try out the suggested calibration prints.

Kind regards
Jeffrey

Edited 1 time(s). Last edit at 07/25/2015 06:12PM by jmg123.

The reason we updated the bed levelling is because of an error that gets introduced by the way the bed is measured in the centre. There are three springs under each edge of the bed, closest to the towers. There's no problem checking the Z height close to the towers, but in the centre you have the pressure of all three springs to push down on, before one of the contacts breaks. It was causing an error of around 0.1 to 0.2mm in the centre, and making the nozzle not level closer to the edges of the bed. The new bed levelling helps this, by adding an offset to the centre point. We played around with all the other various probe point permutations, but this came out the most reliable. But we're happy to hear from anyone who thinks differently!

Ian
RepRapPro tech support

That's pretty interesting actually. I was having calibration issues so having read some of dc42's posts I decided to start using 6 factor calibration using 7 points (towers, tower opposites at 50 mm radius, and the centre). As it happened, the problem turned out to be a loose ball on one arm, but I stuck with 6 factor calibration anyway. I did notice that while the calibration was pretty good after I sorted things out, the centre ended up being closer than the edges. This difference isn't huge, perhaps 100 um or so, but its enough that it takes some fine tuning of the probe offset to prevent either the edges peeling with larger prints or the centre being smashed into the bed. I found that using a probe offset (G31) of Z-0.22 is the sweet spot. I guess this explains what I've been seeing.

While I had realised that the probing method required an offset (as the nozzle has to physically displace the bed), it hadn't dawned on me that the offset could be varied per point. I shall definitely have to look in to this tomorrow to improve see if I can flatten out the build plane.

The only other tip I'd add on the subject of calibration is that I heat the nozzle up before doing the probing to ensure that any remnants of material on the bed or protruding from the nozzle squash out of the way rather than impact on the reading.

Ian,

I've been having another look at this this morning but I am confused by one of the commands your sending at the end of calibration, I think it may be an error. On line 29 of bed.g, you send "G31 Z0.1", but I think it should be "G31 Z-0.1". In the config I see that you now set it to -0.1, presumably to account for the fact that the bed needs to be displaced by the nozzle in order for the probe to register. This means that the first 3 points (the towers) of the first calibration run have offsets of -0.1, the offset is then changed to -0.2 before probing the centre point, but in then changing it to 0.1 afterwards you'll surely screw up the offsets for the tower probing on the next iteration? So, unless I'm missing something, I suggest this should be -0.1.

Also, in the default config.g you provide, the endstop offsets are set with some pretty weird values. I would have thought that most users would have more benefit with them being set to 0 as an out of the box configuration. My calibration runs settle with pretty tiny values for the endstop offsets (X-0.08 Y0.01 Z0.07), so the calibration converges in far fewer iterations if I start with 0. I guess it's not a big issue since it's expected that the user updates their config with their values after an initial calibration, but I still think it would make things easier for new users if the initial offsets were 0.

Cheers,
Chris

Hi Chris

Thanks for your feedback. Can you also email your observations to me on support@reprappro.com ? Jean-Marc, the designer of the Fisher, and director of RepRapPro, did the calibration, but is away on holiday this week. I haven't dived into the configuration yet to know enough about it (I only finished by Fisher over the weekend, which is why he's been doing most of the support), so can't really comment. Though I'm pretty sure what he's put will be correct; he's been printing on Fisher for the last couple of months very intensively!

If you feel like trying out your suggested changes, let me know your results.

Ian
RepRapPro tech support

Sure, will do! I already emailed a few times last week with some suggested observations and possible improvements but I'll forward these too.

Thanks Chris, got the email. Just a quick note, we added a bit about bending the microswitch levers to the instructions on Friday: [reprappro.com]

Ian
RepRapPro tech support

Ah, great. Yeh, just putting a 45 degree bend in the microswitch arm makes a tremendous difference to the calibration... at least, it did for me.

Quote
bluesign2k
Also, in the default config.g you provide, the endstop offsets are set with some pretty weird values. I would have thought that most users would have more benefit with them being set to 0 as an out of the box configuration.

I agree. Unless the design causes the 3 microswitches to trigger at different heights, it would make more sense to start from zero.

Regarding the G31 Z offset, I would have thought it should be:

1. G31 Z-0.2 in config.g. This should mean that executing G30 at the centre gives an accurate result.

2. G31 Z-0.1 in bed.g before the first G30 command

3. G31 Z-0.2 in bed.g before the last G30 command.

All of this assumes that an offset of -0.2 gives an accurate height in the centre, and -0.1 gives accurate heights at the edges. I can't see why you would ever need to use a positive Z value in the G31 command for the Fisher. But as I don't have a Fisher, this is all speculation on my part.

---

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].

Yeh, that was exactly my thinking.

I have set up a wiki page on configuring and calibrating a delta running RepRapFirmware at [reprap.org]. My latest firmware release allows you to copy the tower position adjustments into config.g.

---

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].

Hi all

Sorry for the problems with bed.g. Now Jean-Marc is back from holiday, we have found out that we did manage to upload the wrong version of bed.g to the 'updated' version of the SD card files. This looked like:

; Probe the bed and do auto calibration
G1 X-64.95 Y-37.5 F12000
G4 P300
G30 P0 X-64.95 Y-37.5 Z-99999	    ; X tower
G4 P300
G30 P1 X64.95 Y-37.5 Z-99999	    	; Y tower
G4 P300
G30 P2 X0 Y75 Z-99999			; Z tower
G4 P300
G31 Z-0.2
G30 P3 X0 Y0 Z-99999 S0		; centre, and auto-calibrate
G31 Z0.1

The last 'G31 Z0.1' is wrong, and puts the nozzle too close to the bed. This was a transposition mistake, and, as many of you have pointed out, should be Z-0.1. Sorry! This is the same as is set in config.g. This gives 0.1mm of movement of the head to disconnect the bed probe, which seems about right.

The 'G31 Z-0.2' is the one to play around with, as it's the offset for the centre of the bed; it will require a little more force at this point to disconnect one of the bed probe contacts. We've been playing around with this, and it looks like 'G31 Z-0.3' is just about the right. So the last couple of lines should be:

G30 P2 X0 Y75 Z-99999			; Z tower
G4 P300
G31 Z-0.3
G30 P3 X0 Y0 Z-99999 S0		; centre, and auto-calibrate
G31 Z-0.1

We'll update the downloadable file shortly. And we've removed the endstop offset from config.g!

Ian
RepRapPro tech support