Help needed with Duet firmware and 3 extruder retraction.

Hi,

I have a RepRap Pro version of a Mendel 3.0 Tricolour to which I have managed to graft a Diamond Hot End. The Reprap firmware is what came with machine (version 1.09) and the web interface (which I love) is version 1.06. The only software modifications I have made are to the config.g file. Basically this is just redifining the tools to have individual extruders but share the same heater and a tweak to the heater time out which someone on these forums kindly resolved for me.

All is working well apart from a problem I have with filament oozing on large non print moves which I can't cure using retraction settings. On talking to "RepRapMe" (the makers of the Diamond hot end) it seems that what I need to do is retract all three extruders simultaneously. Because they share a common chamber, retracting one filament does not create a "vacuum" and just pulls filament from one of the other inputs. Their suggestion is to use Repetier Firmware which supports this feature.

However, after many hours of research it seems that Repetier does not really support Duet electronics and in any case, it looks like I'd lose the web interface.

So, my question is, how can I set my machine to retract all three extruders, not just a single one? I thought maybe I could define the tools so that they all shared all 3 extruders (M563 (P0 D0:1:2 H1 etc) then use M567 and M568 to set and enable tool mixing with each one using 100% of a different extruder. ie, M567 P0 E1:0:0, M568 P0 S1, M567 P1 E0:1:0, M568 P1S1, M567 P2 E0:0:1, M568 P2 S1. I tried putting those numbers in my config.g but when I tried to do a print, all that happened is that at the start, it set the bed temperature to active 205degC (same as the tool temperature). Of course, it would never reach that temperature and fortunately I noticed what was happening before the bed caught fire, so I put the config.g back to as it was.

Will a firmware update give me the features I need and if so which branch should I look at? I'd be loath to lose the web interface. The more I read, the more confused I am becoming.

Many thanks in advance.

First, I suggest you do upgrade your firmware, either to my 1.09x release or chrishamm's 1.09x. At the same time, update the web interface to DuetWebControl 1.10. There have been numerous improvements to both the firmware and the web interface.

However, that won't solve your retraction issue. But I am willing to add a solution in my fork if we can work out a proper specification for it.

Using the tool mixing option will be a good start. Please try the tool mixing config with my release and see if there is still a side effect on the bed heater.

The problem is that retraction and un-retraction will drive the extruders in the same ratio as when printing. I can see two ways of making them all retract the same amount:

1. Recognise retraction and un-retraction moves somehow. I am not keen on this because too much could go wrong.

2. Implement firmware retraction. I know that some slicers have a "Use firmware retraction" feature. I presume this generates special gcodes to tell the firmware to retract or un-retract, but I haven't checked. Does your slicer have this option, and if so, what effect does it have on the generated gcode?

EDIT: looks like slicers generate G10 and G11 codes for retraction and un-retraction. Please see if this is what your slicer does. This unfortunately clashes with how NIST defines G10 and how RepRapFirmware interprets G10. But all is not lost, because we could still recognize G10 without any parameters as meaning "Retract" if the retraction parameters have been set previously.

Edited 1 time(s). Last edit at 03/19/2016 04:24AM 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].

Something I thought some time ago to transfer info from the slicer to the firmware is to give simple values which could never occur in a real slicing, for layer thickness, retraction, temperature, etc.

So, if you use a value of 99 or 999 for retract or unretract, it could easily be recognised as a such by the firmware.

It might be interesting to define a 'standard' to have the slicer transfering some responsibilities to the firmware, which can open a lot of possibilities and may be a step toward a slicing independant from the printer.

Edited 1 time(s). Last edit at 03/19/2016 09:11AM by PRZ.

Many thanks for the help so far.

@DC. I should point out that I'm a 62 year old carpenter so although I do my best to keep up with technology, I do find myself floundering at times. Is there a real idiots guide to updating the firmware? All I have to go on is the instructions that RepRapPro provided when I build the kit and the supplied firmware was the latest available so I have never attempted to upgrade it. I'll give it go but I'm really nervous about doing it.

I'm using Slic3r. A quick squint at the gcode files it generates show that it uses G1. To be precise the command looks like this G1 E-5.00000 F2400.00000.

Interestingly, firmware retraction is how RepRapMe do it with Repetier firmware. That is to say that the Repetier Host uses Repetier Firmware with any of a number of Slicer options and their way of setting it up is to tick the check box in Slc3r which is entitled "Use firmware retraction". I just tried a quick test with Slic3r set to use firmware retraction and looking at the resultant gcode, the previous G1 commands have been replaced with "G10 ; retract", then a small x and y movement of the head followed by "G11;unretract".

This is a link to their wiki [reprap.org]

Near the bottom it states the following

Quote..........

"Since version 0.92 of Repetier-Firmware there is the option to support a mixing extruder. In fact this feature enables the firmware to handle extrusion from more than one extruder at the same time, in effect allowing for true color mixing capability.

Another neat feature of Repetier-Firmware is that when using firmware retraction Repetier-Firmware can simultaneously retract all three filaments, thereby lowering the retraction length needed (to 5 mm or even lower), in turn allowing for a cleaner, faster, more efficient retraction sequence".
............end of quote.

What isn't clear is how that retraction works when one is using a single filament -i.e not "mixed" but the implication is that it retracts all 3 extruders by the same amount as they claim it lowers the retraction length needed to 5mm or lower (with 500mm Bowden tubes) but they don't specifically state that this is regardless of mixing ratios.

Also, on their wiki is a link to their version of Marlin [github.com]. It might give an idea of how they implement things??.

See [reprappro.com] for how to update the firmware. You can download my 1.09x release by following this link [github.com] and then pressing the Raw button. Chrishamm has recently posted a thread in this forum about his new release.

If you can set up your tools as mixing hot ends using 3 extruders (even if the mix ratio is 1:0:0), then I can implement G10/G11 to retract all three extruders equally.

---

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 dc42,

Very many thanks for you help so far. I've managed to update to your latest firmware and Chrishamm's latest web interface (many thanks to him also). That was scary - I'm fine with circular saws and the like, but this stuff is well outside my comfort zone. Anyway, all seems to have gone well.

So, having done that, changed my config.g to set the tools for mixing as before (tool one all filament A, tool 2 all filament B etc). I also added M207 and M208 to set the firmware retraction and recovery. Then I sliced my test piece in Slc3r (just using tool 1) with the firmware retraction box ticked. When I came to print it, I had the same problem as before. i.e. the active bed temperature gets set to 205 degrees C (same as the tools). I've attached bot my start gcode and my config.g files. The bed initially gets set to 57 degrees C (which is what Slic3r tells it to do), then it gets reset to 205 after it has done the homing. So, I'm guessing it has something to do with the G10 commands in the start config conflicting with the tool definitions in the config.g file. If I put the tool definitions back to their "non mixing" settings (i.e all using H1 but each one using a different extruder) and leave the start gcode alone, I don't have the problem with the bed temperature.

Thanks for trying this. I'll try to reproduce your configuration and see if I get the same problem.

---

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 just tried to reproduce the problem, but I was unable to. If I configure the tools the same as yours and then run your start gcode, the bed temperature setting does not change. Perhaps the problem you have is triggered by something else in the gcode file you are trying to print?

Please can you do the following:

1. Remove the M207 and M208 commands you added near the end of config.g. M207 is not yet implemented, and M208 means something different in RRF. Then restart.

2. Uncheck the "Use firmware retraction" box in your slicer, until I provide you with a version of RRF that implements firmware retraction.

3. Test whether you get the problem with the bed temperature if you "print" just that start code you attached in your previous post.

4. If no, slice a small STL model and print the resulting file. If the problem reappears, attach the gcode file to a 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].

@dc24,

I've done all you asked and have been tearing my hair out because my old config.g works fine, the new one (which I previously attached to my post) has the problem. Just trying to "print" a gcode file which had only the Slic3r start code was setting the bed to 205 degrees C. I even put the new config.g back to the same values as the original config.g and still the problem was there!

I compared them both side by side using notepad++ and eventually saw the problem. Line 43 which M563 P1 D0:1:2 H1 had a colon before the comment instead of a semi colon!. I put that error right and a quick test showed that the bed is not now being set to 205 deg C. It's now nearly 11.00 pm and my brain is fried so I'm going to crash out. I'll try to do some more testing tomorrow.

Thank you for being patient with a dozy old f*rt like me.

I'm glad you found the problem.

I wrote the code to support firmware retraction earlier this evening, and I may get time to test it 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].

Did some more testing this morning. I can confirm that when setting tool mixing in config.g then slicing with Slic3r without firmware retraction being set, retraction happens on all 3 extruders but is proportional to the mix ratio. i.e. setting the ratio to 1:0:0 results in only one extruder retracting while setting the ratio to 0.34:0.33:0.33 results in all 3 retracting by the the same amount (as perceived by eye).

This does appear to be the route to go down with a diamond hot end. Using 5mm retraction at 40mm/sec on all three extruders gave 90% less "stringing" on my test piece than the silly settings of 10 mm or more at up to 100mm/sec with a single extruder.

I've sent an email to the makers of the Diamond Hot End asking them to confirm that Repetier does in fact use a fixed firmware retraction length for all extruders regardless of "mix ratio". (I can't see how you could prevent filament being drawn from an unused input unless you do it this way). The only drawback I can see is that continuously retracting and then recovery filament without actually extruding anything, might end up with the filament getting chewed or worn in one spot. I guess setting a mix ratio of something like 0.96:0.02:0.02 might be a possible solution as it would move the filament forward slightly without affect the colour too much. That's a whole new world to explore.....

As an aside, I have to say that I'm loving the diamond hot end as a tri-colour compared to the 3 separate hot end set up of the RepRapPro Mendel 3 Tricolour (apart from the current oozing problem that is).

I hope to have a version for you to test tomorrow. The code I have written applies the degree of retraction defined by M207 to all the extruders associated with the current tool.

---

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].

Brilliant! Out of curiosity, will it also use M208 for the recovery? It's no big deal if it doesn't but I came across this (http://prusaprinters.org/slic3r-and-marlin-configuration-for-reprap-firmware-retraction/) and my understanding of it is that one could use a different (perhaps slower) speed for the recovery which might help to stop the filament getting chewed up?

As an aside, I had a reply back from RepRapMe about how Repetier handles retraction for multiple extruders. Basically, they don't know but I think it must do all 3 by the same amount, or if it doesn't then it won't work if the mix ratio is 1:0:0. So I think your plan is the best approach. As another aside, RepRapMe mentioned that they would be quite keen to use Duet electronics in their products. (Might be worth contacting them as they market a range of printers as well as their Diamond hot end?).

M208 is assigned to setting axis limits, so I won't be using that. I have added an extra R parameter to M207 to set the additional length to un-retract, see [reprap.org]. I haven't heard of any reason to use a different un-retraction speed before, so I haven't made provision for different retract/un-retract speeds.

Edited 1 time(s). Last edit at 03/22/2016 06:08AM 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 put a copy of RRF 1.09zb1-dc42 at [dl.dropboxusercontent.com]. This supports firmware retraction and should solve your retraction problem. You will need to:

- Configure your print head as a mixing tool, even if you use mix ratios of 1:0:0 or something like that

- Enable firmware retraction in slic3r when generating gcode

- Set the retraction parameters you want to use with M207 (do not use M208, that sets the axis limits)

You should be able to install it just by uploading it on the Setup page of chrishamm's web interface.

I haven't run a print using firmware retraction, because slic3r has a bug that causes it to crash on my machine (and those of several other users) which the developer seems to be ignoring, and my replacement slicer (S3D) does not support firmware retraction without a hack. But I have tested the G10 and G11 retract/unretract commands in conjunction with the M567 and M568 mixing commands and they are working properly

Edited 1 time(s). Last edit at 03/22/2016 04:15PM 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].

Many, many thanks. I'll upload it and try it tomorrow. Ref the slic3r bug, I saw something about it here. [reprap.org]. Towards the bottom they say

Quote..........

Note: Although firmware retraction works perfectly for print jobs that does not actually use the colormixing feature of Repetier-Firmware, Slic3r 1.1.7 has a bug that sometimes G11 firmware unretract is not issued after a toolchange. Of course the colormixing functionality works with oldschool retraction, but on the cost of improper retraction.

This issue seems to be fixed in Slic3r 1.2.9 now with slightly better results --RepRap.me (talk) 06:31, 10 February 2016 (PST)

..........end of quote.

My version of Slic3r is 1.2.9 so fingers crossed..........

I'll let you know how I get on.

Thanks again.

Hi dc42,

Sorry to report this but my machine does not like that experimental firmware. The problem I have is that whenever I try to print a file (even one that has previously printed OK) using the web interface, I get an error as follows

"G codes : String expected but not seen
Gcode file M32 xxxx Gcode not found.
Failed to open file M32 xxxx.Gcode"

I substituted the file name for "xxxx" in the above.

Also, if I tried to upload a config.g file while this firmware was installed, I got an error

"Communication Error
An Ajax error has been reported so the current session has been terminated
Please check if your printer is still online and try again.
Error reason timeout"

This latter error persisted even after I changed the firmware back to your stable version but clearing the browser cache eventually fixed it.

It was as if the sd card couldn't be read or written to, yet I could get a list of gcode files from the web interface. Connecting via usb and checking it with the Arduino interface also worked fine.

I did change a few things all at once. i.e As well as the firmware I added crishamms iap file, changed the tool definitions in config g and added a line for firmware retraction. So, it took me a while to remove all of these changes and put them back one by one. The only thing that cured the problem was reverting back to the stable version of your firmware.

Edited 1 time(s). Last edit at 03/23/2016 09:04AM by deckingman.

Sorry about that! I've just fixed it and uploaded a new build to the same location. I am currently doing a test print - I'll post again when that and some more tests are complete.

EDIT: I need to make one more change, working on it...

EDIT 2: new build uploaded, and a test print using firmware retraction is looking good.

Edited 2 time(s). Last edit at 03/23/2016 10:27AM 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].

No probs. Actually I am so relieved that you found an error and fixed it. This is all well outside my comfort zone and I thought I'd done something stupid again.

I'll give it another go as soon as my heart rate and blood pressure return to something like normal. I'm much more confident using a circular saw, or even a chain saw (not that I use chain saws for building decks). I know what I'm doing with power tools but messing around with firmware is so not my thing.

YEAY, IT WORKS!!

So, updated the firmware to your latest (experimental) version. Defined the 3 tools in my config.g as having the same extruders (D0:1:2) and sharing the same heater (H1). Set the mix ratios to 1:0:0, 0:1:0 and 0:0:1. so that each would use 100% of only one filament. Added M207 S5.0 F2400 to the end of my config.g (didn't bother with Z lift or extra length on recovery) to give 5mm retraction at 2400mm.min (40mm/sec). Added my test piece to Slic3r and set the object to extruder1 (effectively tool 0). Sliced it with firmware retraction ticked. Checked the generated gcode file and noted that it contains G10 ; retract and G11 ; unretract in place of G1 retract commands. Printed it and observed that although it was only using extruder 1, all 3 extruders were retracting. Better still, there was no sign of the stringiness or oozing that I had when I could only retract one extruder. The test piece I use is a harsh test and has non print moves between 5 and 110mm. for every layer.

The object I printed was 83 layers tall and each layer had 6 retractions. I was concerned that the filaments which were being retracted without any forward movement might get chewed or worn but even towards the end of the print, I could see that the filament was still being "tensioned and relaxed", even after 250 retractions. I'm guessing this is because there was never really any forward pressure. When the print stopped, with the hot end still up to temperature, I could easily extrude filament from the "unused" extruders so it hadn't worn the filament as I feared it might.

I guess now I have this whole new world of mix ratios to explore........

Very many thanks for your help David. Let me know if you'd like me to run any more tests with this Diamond Hot End set up.

I'm glad it's working for you. The Z hop parameter in the retraction settings is only really useful on delta printers. I've never had to use the "Extra length on restart" parameter in my slicer settings, but I decided to include it as an option in M207 before someone complains that it is missing.

I already released that binary as 1.09z-rc1-dc42 and I will do a proper release 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].

dc42. I just tried uploading a config.g file via the web interface. It seemed to work OK and asked if I wanted to do a software reset which I said yes to. After about 8 seconds it came back with the AJAX communication error (reason timeout), and terminated the session. Is it possible that it could be your "experimental" firmware that I'm running? I'm in the middle of doing a series of tests to fine tune the firmware retraction (which is working fine by the way) and also various "mixing" tool configurations so I'm reluctant to change back to the previous firmware to test it. If there is no possibility that it can be caused by the firmware, then I'll post the issue in the web interface thread. Thanks

Please upgrade to my 1.10 release. During testing, I identified a problem with IP address allocation using DHCP. After a power up or reset, the network stack sometimes failed to acquire an IP address. Chrishamm was able to reproduce it and kindly provided a fix this morning, which I included in 1.10. If you have your IP address set to 0.0.0.0 then that could be the problem.

---

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've updated the firmware but not tested it yet. Quick question, could that cause a general slowdown of Firefox or do I need to look elsewhere?

No, that should not cause a general slowdown of FireFox.

btw there is a bug in both the experimental build and version 1.10 that causes incorrect un-retraction if you have all of z-hop, orthogonal axis compensation and pressure advance enabled. I have fixed this in my current build and I will do a new release in a day or two. In the mean time, do not use all three of those features at once.

---

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 for that. Firefox seems to be fine in safe mode so it must be an "add on" that's causing the problem. For info, I don't use any of the 3 feature you mentioned so not a problem. For further info, I've just printed a 7 colour object using different mixing ratios to give 4 extra "colours" on top of the 3 primary filament colours. I've also reduced retraction from 5mm to 4mm with still no sign of the oozing and stringiness that I had before when retracting only one filament. It's all working well so very many thanks again.

Quote
deckingman
For info, I don't use any of the 3 feature you mentioned so not a problem. For further info, I've just printed a 7 colour object using different mixing ratios to give 4 extra "colours" on top of the 3 primary filament colours.

A photo would be nice!

---

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].

Here you go. I aborted the print at the end because I had forgotten to rotate the final band in OpenScad so it doesn't line up. It's just a container - I'll maybe use it as a pencil holder or maybe just chuck it. The 7 colours are made from the filaments which were already loaded, namely Red, Black and Gold, then 0.5 Red 0.5 Black, 0.5 Red 0,5 Gold, 0.5 Gold 0.5 Black, and finally 0.34 Black 0.33 Red , 0.33 Gold. Slicr can only cope with 4 tools (or Extruders as Slic3r calls them). I sliced it using 3 extruders then edited the gcode file using notepad++. It wasn't too bad to do. I just kept doing "find next" for "Z" to find the required height and insert or change a T(n) command. If I do it again, I won't bother making separate stls and using "load part" in Slic3r. It'll be less bothersome to just create a single object stl and slice with a single extruder in Sllic3r, then add all the tool changes to the gcode file afterwards by doing the "find next" thing. Potentially, I could have a tool change on every layer which would make a smooth(ish) transition from I colour to another in (say) 1% increments. Of course, that would mean having 101 tools. Which actually leads me to my other post where I've asked the question "How many tools can I define?" - see my other post. Or I'm open to any other suggestions as to how I could change mix ratios. I guess I could send mixing commands on the fly but that would mean sitting in front of the printer all the time.

While we are already way off topic, one thing I noticed with the Diamond hot end is that it doesn't actually mix well. The filament tends to come out a bit like a 3 striped toothpaste. So depending on where you view it, the colour is biased towards the extruder that was closest to the face your are viewing it from. Personally, I don't dislike that as it makes some other interesting effects. I'll print a 3 sided pyramid using 3 filaments in equal proportions with each side facing directly towards where the filament enters the hot end, just to see what it looks like.

That is pretty damned impressive if I may say so are you using Full E3D Heatsinks and heatbreaks?

Doug

Quote
dougal1957
That is pretty damned impressive if I may say so are you using Full E3D Heatsinks and heatbreaks?

Doug

Well thank you kind sir.

Basically I bought this (http://www.ebay.co.uk/itm/Genuine-Diamond-Hotend-with-E3D-Heatsinks-for-multicolor-3D-printing-/272073946988?hash=item3f58def76c:g:LlEAAOSwv-NWauos) which is the brass nozzle, 3 off E3D Lite heat sinks and some bits of heat shield stuff. At least, I think they are the Lite version - it's whatever "RepRap.Me" supply in their kits and I seem to recall reading somewhere that the Lite version works better in this application . The other parts like the cartridge heater, thermistor, Bowden tubing, fan etc, I sourced locally. I made a mount to suit my RepRapPro Mendel 3 by grafting bits on to their (RepRap.Me that is) stl file using OpenScad. It's not the prettiest piece of code but the resultant stl makes a mount which includes a bracket to take the bed probe and fits on the existing X carriage mount so no need to take the machine apart. I posted the mount on thingiverse in case any else wants to try it.