New RepRapFirmware 1.09i-dc42 for Duet electronics

I have just released this at [github.com] (follow the link and then press the Raw button to download it). Improvements in this release:

• Added support for second extruder and cooling fan on Duet 0.8.5 (thanks T3P3)
• Added auto detection between 0.6 and 0.85 Duet revisions
• Added optional P parameter to M115 command to select board type between Duet 0.6, 0.7 and 0.85
• Changed M115 output to report the board type that was configured or auto-detected
• Improved the step ISR and step pulse generation efficiency to allow higher movement speeds, especially when using 0.9deg/step motors on delta printers
• Added XYZE parameters to M569 command to allow stepper driver remapping (e.g. to use external stepper drivers connected to the expansion connector)
• Added R parameter to M569 command to allow enable signal to be reversed when using external drivers (thanks dnewman)
• Removed R parameter from M558 command, This is no longer needed, because board type Duet 0.7 can now be set via M115 instead.
• Moved Fan0 RPM sense pin to expansion connector pin PA14 to avoid conflict with Duet 0.8.5 FAN1 pin
• M408 poll command (used by latest PanelDue firmware) can now be handled concurrently with other commands

See [reprap.org] and [reprap.org] for details of the new M115 and M569 parameters supported.

---

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].

This is really great work David, thanks!

---

DuetWifi.: advanced 3d printing electronics

Thanks for the firmware, David!

I read your blog on Z-probe and calibration.
You mention bed10-7.g file, but what I see now is bed.g in sys folder. How do I run it? If I'm supposed to upload it to printer, why is it in sys folder? Is there some M-code for autocalibration?

Sorry I already found G32 code here

Edited 2 time(s). Last edit at 09/03/2015 03:08PM by pkm.

That's right, G32 runs bed.g. See [reprap.org] for the current instructions.

If you want, you can also create different variants of bed.g with different names, and run them using the M98 P command.

Edited 1 time(s). Last edit at 09/03/2015 04:31PM 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].

Unfortunately I introduced a bug in version 1.09i. When the Pause function was used, sometimes the printer would hang (safely, i.e temperatures controlled) while executing the pause.g macro. This is fixed in version 1.09j. The full change list for this release is:

* Bug fix: pause function sometimes used to hang while running the pause macro file
* Don't wait for all moves to complete when executing a G92 command that just resets extruder positions (this affected prints that use absolute extruder moves, although relative extruder moves are recommended with RepRapFirmware)
* M574 endstop configuration now overrides M558 when determining whether or not to use the Z probe instead of a homing switch
* Added XYZ parameters to M667 command to support CoreXZ etc. machines with paired axis ratios that are not 1:1
* Minor change to step ISR to improve timings and service drives in round-robin order

---

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
dc42
* M574 endstop configuration now overrides M558 when determining whether or not to use the Z probe instead of a homing switch

Great! I was following your blog entry [miscsolutions.wordpress.com] and entered "M558 P1" when setting IR probe... causing a crash on the next homing. Now it's safe.

Quote
pkm

Quote
dc42
* M574 endstop configuration now overrides M558 when determining whether or not to use the Z probe instead of a homing switch

Great! I was following your blog entry [miscsolutions.wordpress.com] and entered "M558 P1" when setting IR probe... causing a crash on the next homing. Now it's safe.

Thanks for telling me. I've updated those instructions so that they are safe even when using older versions of my firmware, or RepRapPro's or zpl's version.

---

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 noticed that the configuration file is shown partially in Settings tab.

Edited 1 time(s). Last edit at 09/07/2015 01:33PM by pkm.

Seeing this error, version 1.09j :
Error: SetPositions called when DDA ring not empty

???

Rick

Quote
llamatrails
Seeing this error, version 1.09j :
Error: SetPositions called when DDA ring not empty

???

Rick

What triggers this message? Is it reproducible?

---

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].

Attached are the original and sliced files. The error was repeated 33 times.

Also, here is the output of the Printer Status :
File Information
Size: 29.6 KiB
Object Height: 200 mm
Layer Height: 0.3 mm
Filament Usage: 1285.2 mm
Generated by:
Slic3r 1.2.9 on 2015-07-14 at 10:23:10

The object height should show 2mm, not 200mm. This also causes the estimated time to completion and number of layers to be off by 100 times.

Whatever is throwing the error, it didn't stop the print.

Rick

Edited 1 time(s). Last edit at 09/08/2015 07:15PM by llamatrails.

Thanks, Rick. I figured out what change I made in 1.09j that is causing that message, and I'll fix it in 1.09i 1.09k.

The object height is showing as 200mm because of the G1 Z200 command in your end gcode. Try changing it to G0 Z200, or remove it altogether as you already have a G28 to move the head away from the print.

Edited 1 time(s). Last edit at 09/09/2015 09:47AM 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].

Quote
dc42
Thanks, Rick. I figured out what change I made in 1.09j that is causing that message, and I'll fix it in 1.09i.

The object height is showing as 200mm because of the G1 Z200 command in your end gcode. Try changing it to G0 Z200, or remove it altogether as you already have a G28 to move the head away from the print.

Appreciate the hard work. Moving from Arduino/RAMPS and using your IR sensor has simplified printing sooo much !!!

Rick

Quote
llamatrails

Quote
dc42
Thanks, Rick. I figured out what change I made in 1.09j that is causing that message, and I'll fix it in 1.09i.

The object height is showing as 200mm because of the G1 Z200 command in your end gcode. Try changing it to G0 Z200, or remove it altogether as you already have a G28 to move the head away from the print.

Appreciate the hard work. Moving from Arduino/RAMPS and using your IR sensor has simplified printing sooo much !!!

Rick

Thanks! You can safely ignore that message. However, it tells me that you are using absolute E coordinates in the gcode, whereas relative E coordinates are recommended with RepRapFirmware if your slicer supports them.

Edited 1 time(s). Last edit at 09/09/2015 09:57AM 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].

I have been taking a look at how other firmwares ( Marlin and smoothie, the pull request) have implemented their autocalibration and they have a delta tower offset calibration in their variants, I was wondering why choose to not have such a feature.

I have been trying to get my delta with duet to print accurately but I still get prints that are off, a cube gets different dimensions in the X- and Y-axis.
I have run the autocalibration routine several times but it just won't get corrected.

When I used a MEGA+RAMPS and Marlin I could adjust that particular offset which would then give prints with correct dimensions.

I was reading earlier that you had some problems figuring out why your effector tilted across the bed, I think that if a tower have a different offset than the others that might be a cause for the tilting.

If you are getting different X and Y sizes, yet the auto calibration is giving you the correct heights all over the bed, then I think there is something significant is wrong with your build. I suspect that you may find that your X:Y size ratio may depend on the size of the print you do and/or the position of the print on the bed. You could try printing calibration cubes of different sizes, and at different places on the bed. Any adjustment you make to the tower positions to correct the X:Y size ratio is going to have an adverse effect on the flatness of the printing plane.

The reason that the effector tilt varies on my printer is that some of the Traxxas joints have a small amount of play in them. I find I can tilt the effector a little by applying very slight pressure. The Bowden tube and associated wiring on my printer causes the effector to be pushed one way when is is close to the Z tower, and the opposite way everywhere else. This not only affects the auto calibration, it also causes prints that extend towards the Z tower to be slightly longer in the Y direction than they should be.

---

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
dc42
If you are getting different X and Y sizes, yet the auto calibration is giving you the correct heights all over the bed, then I think there is something significant is wrong with your build. I suspect that you may find that your X:Y size ratio may depend on the size of the print you do and/or the position of the print on the bed. You could try printing calibration cubes of different sizes, and at different places on the bed. Any adjustment you make to the tower positions to correct the X:Y size ratio is going to have an adverse effect on the flatness of the printing plane.

As I said with Marlin I corrected the issue by altering one of my towers offset.
That also took care of inconsistent first layer-height across the bed.

I took a peek at the pull request for an auto-calibration for smoothie, they have included the tower-offsets in their routine and Marlin has it in their routine, at least in the dev branch.

It's enough that one of your horizontal extrusions differs a couple of 10th's from the others to make the offset different enough to be measurable in the prints.
Not to mention what a couple of printed corners can do for the tower-offsets....

Edited 1 time(s). Last edit at 09/12/2015 03:35PM by Koenig.

Links for the firmwares I mentioned if you would like to take a look at it: For Smoothie.
And for Marlin.

It can't be that uncommon to see this issue if both of them include ways to adjusting the tower offset.

I've said it before but I'll say it again: allowing tower radius offsets to be adjusted, as well as tower angular positions, delta radius, and endstop corrections, achieves nothing other than allowing the origin to be translated. Trying to adjust tower radius offsets as well as angular offsets and delta radius during auto calibration is just stupid, and anyone trying to do that evidently doesn't understand the maths. If you need to move the origin to get your XY ratio correct at the centre of the bed, then you will find it is wrong somewhere else - or very probably everywhere else.

I have considered adding M665 parameters to allow the origin to be translated, in order to support delta printers with deliberately usual geometries. However, placing the origin at the centre fo the circle that the towers lies on work for both both standard and 'square' deltas, so I haven't done this yet.

I suspect the problem you are having with XY ratio not being correct is due to backlash in your joints, or one pair of diagonal rods having unequal spacings at the upper and lower ends, or diagonal rods being not quite the same lengths, or something similar. Any correction you make to tower positions to compensate so as to get equal XY sizes will have an adverse effect on bed flatness. Did you ever try printing calibration spirals when you were using Marlin? Did you try printing calibration cubes of different sizes and at different places on the bed, to make sure that the XY ratio was correct everywhere?

---

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
dc42
I suspect the problem you are having with XY ratio not being correct is due to backlash in your joints, or one pair of diagonal rods having unequal spacings at the upper and lower ends, or diagonal rods being not quite the same lengths, or something similar.

I've "zeroed" all my backlash, using decent quality linear rails, and a set of haydn's rods (magnets in the rods and balls on the carriages and effector), the rods are within a few hundreds of a mm, I have metal corners from robotdigg, I have the belts play me a tone (much like a bass) when I snap them.

Next I'm also going to add springs between the carriages and the effector to really make sure there's enough force to keep the effector in place.

I can also add that from the beginning I used traxxas rod-ends and other linear rails, I had the issue then (could correct it in firmware), but I wanted to make my build as accurate as I could and see if I could find what was causing it, and some time later about the only thing I haven't switched out to get rid of "the issue" is the aluminium extrusions, the horizontals that is, because the verticals have been changed, thought it might be a leaning tower, but now, sandwiched with the linear rails I highly doubt there's any leaning going on, if so that would be errors in the corners (robotdigg again).

The thing left physically is to change the horizontals, but that's really a PITA, and knowing that there's "zero" backlash and there's an fairly easy way to correct it in software + seeing others have implemented such a correction in their version of autocalibration.

Quote
dc42
Any correction you make to tower positions to compensate so as to get equal XY sizes will have an adverse effect on bed flatness.

Fully aware of it, without the offset correction I could see places on the bed (borosilicate glass) where the first layer thinner/thicker, but correcting a tower offset fixed that

Quote
dc42
Did you ever try printing calibration spirals when you were using Marlin?

No, but I printed a ton of these: [www.thingiverse.com]

Quote
dc42
Did you try printing calibration cubes of different sizes and at different places on the bed, to make sure that the XY ratio was correct everywhere?

Well yes, but they were incorrect everywhere until I set the tower offset, when I did that I only tried like 2, they came out OK so I didn't really see the point in printing more at the time.


So you are outright calling Rich Catell and Oskar Linde stupid cause they adjust tower offset in their autocalibration?

Doesn't angular offsets just move the towers along the same perimeter of a perfect circle?
what happens to the math if you put one of the towers slightly inside or outside of that perimeter?

It is real easy to end up slightly inside/outside that circle, an imperfection in a printed carriage, one slightly longer horizontal extrusion, imperfect rollers, imperfect corners......

Edited 3 time(s). Last edit at 09/13/2015 05:29AM by Koenig.

Koenig
I agree with David here. Angular positions + delta radius + origin shift is identical to tower offsets + delta radius.
But nevertheless you don't get the result. I would try another set of calibration points, different number of points etc.

Quote
pkm
Koenig
I agree with David here. Angular positions + delta radius + origin shift is identical to tower offsets + delta radius.
But nevertheless you don't get the result. I would try another set of calibration points, different number of points etc.

You are totally right, this is what I'm asking for here.
Just that you cannot set this in reprap firmware at the moment.

Quote
dc42
I have considered adding M665 parameters to allow the origin to be translated, in order to support delta printers with deliberately usual geometries. However, placing the origin at the centre fo the circle that the towers lies on work for both both standard and 'square' deltas, so I haven't done this yet.

Am I wrong in thiking this:

Quote

Doesn't angular offsets just move the towers along the same perimeter of a perfect circle?
what happens to the math if you put one of the towers slightly inside or outside of that perimeter?

I'm thinking if everything else is correct, rod-lenth etc etc, but one of the towers is acutally a bit off from that perimiter, doesn't the reprap autocalibration try to fit the tower somewhere along the perimeter?
I'm thinking that in some cases that is not possible....?

And I'm real sorry if my tone or anything else comes of as impolite or such, that is certainly not my meaning.

I'm super-appreciative of all the work being done here and have the utmost respect for all that know how to write code and is helping out in this community

Edited 5 time(s). Last edit at 09/13/2015 06:21AM by Koenig.

Can I ask if there is an ETA on the 1.09k firmware release.
or should I just use the 1.09j firmware which is already available.

Any chance I can ask if the 1.09j Raw Firmware files are available for download.

Koenig
I mean changing the points in bed.g file. Put them in a square vertices instead of a triangle/hexagon etc. That might improve the calibration. Or not. Still worth trying.

Quote
pkm
Koenig
I mean changing the points in bed.g file. Put them in a square vertices instead of a triangle/hexagon etc. That might improve the calibration. Or not. Still worth trying.

I can tell you that I have identified what makes me want a adjustable tower-offset.

Two of my horizontal extrusions are 301mm long whilst the others are 300, both the 301 sits on the same side, the side opposite to the X-tower.

This makes Y- and Z-towers have the same offset while X-tower is slightly closer the center and inside the desired perimeter but still at a perfect 120 degree angle (taken from the supposed center and opposite the other towers).

EDIT: These came with a rather cheap kit (Folgertech 2020).
And that sort of makes me think I cannot be the only one seeing this issue, I mean it ships a lot of cheap kossel-kits out there....

Edited 4 time(s). Last edit at 09/13/2015 09:29AM by Koenig.

Quote
3didd
Any chance I can ask if the 1.09j Raw Firmware files are available for download.

You can find it this way : [github.com]

Edited 1 time(s). Last edit at 09/13/2015 07:37AM by Greg_be.

In response to more than one post above:

1. If you want to achieve the effect of moving the X tower a little closer to the centre, you can achieve that as follows. Your towers form an isosceles triangle instead of an equilateral one, with the base longer than the sides. So apply a negative correction angle to the Y tower, and a positive angle correction of equal magnitude to the Z tower. [Alternatively, leave the Z tower alone, apply a negative correction to the X tower, and a negative correction of twice the amount to the Y tower.] Then reduce the delta radius slightly. You can use school trigonometry to work out the size of the corrections you need. You will also get an origin shift of somewhat less than the amount you have moved the X tower inwards.

2. I didn't call anyone stupid. Even clever people occasionally do stupid things, especially when haven't worked out what the problem they are trying to solve really is.

3. 1.09k firmware has been held up a little while I work on increasing the upload speed to SD card over Ethernet. 1.09j is available by clicking on this link [github.com] and then clicking the Raw button.

---

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].

By way of further explanation: adjusting too many parameters during auto calibration (i.e. both tower angle positions and tower radii form origin) is analogous to starting from the following two equations:

a + b + c = 10
a + b - c = 4

and then trying to solve for a, b and c without further information. Of course, the equations are more complicated than this, but the principle is the same. You can add more probe points to get more equations, but if your delta build is accurate (i.e. the only thing wrong with it is that the towers are not quite at the corners of an equilateral triangle), you will end up with equations analogous to:

a + b + c = 10
a + b - c = 4
2a + 2b + 2c = 20

and the extra equation doesn't give you any more information. If your build is less accurate (so that shifting the origin significantly affects the nozzle height), you may get this:

a + b + c = 10
a + b - c = 4
2.001a + 1.999b + 2c = 20

In this case it is possible to solve for all variables, but the ratio between a and b (i.e. the origin position) becomes very sensitive to small changes in probe height measurement. Besides, you haven't really solved the problem. You've increased the accuracy at the particular points you probed, but made it worse at some points that you didn't probe.

Edited 1 time(s). Last edit at 09/13/2015 10:24AM 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].

Quote
dc42
In response to more than one post above:

1. If you want to achieve the effect of moving the X tower a little closer to the centre, you can achieve that as follows. Your towers form an isosceles triangle instead of an equilateral one, with the base longer than the sides. So apply a negative correction angle to the Y tower, and a positive angle correction of equal magnitude to the Z tower. [Alternatively, leave the Z tower alone, apply a negative correction to the X tower, and a negative correction of twice the amount to the Y tower.] Then reduce the delta radius slightly. You can use school trigonometry to work out the size of the corrections you need. You will also get an origin shift of somewhat less than the amount you have moved the X tower inwards.

Thank you for this and for the more "in-depth" explanation in the post above.
I'll have to read up a bit "school trigonometry", to see if I can get it right manually, was like 25 years ago and has not been used since.....
As a note, the auto-calibration seems to want the angles another way: M665 L250.42 R120.78 B105.0 H463.64 X0.43 Y0.01 Z0.00

I'll have another go at getting to the accuracy I want, just ordered the metal carriages and effector from robotdigg, and I'll see if I can find a firm that cut/assemble me some rods that have a precision below 10 mikrons.

Quote
dc42
2. I didn't call anyone stupid. Even clever people occasionally do stupid things, especially when haven't worked out what the problem they are trying to solve really is.

Fair enough!
Not seeing the forest for all the trees....

EDIT: So with that effector and my new E3D v6 I guess I'll also new a 1.1 version of your IR-sensor.
Do you have any left and whats the price on them now, or is this a disucussion only suited for PM?

And again, sorry if I somehow seem offensive or so, english is not my native language.

Edited 4 time(s). Last edit at 09/13/2015 12:18PM by Koenig.

Hi Koenig,

I don't find your language offensive, and I appreciate you are trying to make sense of conflicting approaches to auto calibration. I admire people like you who are not native English speakers, but manage to express themselves in English far better than I could in any foreign language that I have learned.

Your magnet & ball joints should be free of the backlash that I have in my Traxxas joints. However, the spherical cups need only be slightly over-sized to allow some slop. I am considering replacing my Traxxas joints by balls in conical sockets machined in PTFE.

Have you tried measuring the nozzle height when the sensor triggers at the probe points you are using? I have described how to do this a couple of times, in different threads, but you might not have seen them. If you have a set of feeler gauges, then you can also measure the height of the bottom edge of the sensor board when the nozzle is just gripping a sheet of paper, and repeat this for every probe position. It's rather tedious, but it gives you a direct measurement of effector tilt. Preferably, measure the height at both corners of the bottom edge, which will tell you about the tilt in both directions.

Regarding the E3Dv6 and version 1.0 IR board, I published a fan duct and IR board mount for this combination at [www.thingiverse.com]. This is what I was using before I developed the version 1.1 board.

---

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].