Ideas for recovering failed prints

[schafe 3]

I have had 2 prints fail because the filament stopped. Once because it got bound up on the spool and once due to a clogged nozzle. The printer keeps on going assuming everything is normal. If this happened at the end of a really long print it would be a bummer.

I have 2 ideas I think make a good solution.

Add a pause function to the print routine. This would be nice when you need to remove a stray fiber or goober on the nozzle. Once paused there is an option to rewind the print using the control wheel. The rewind function would stop extruding but the print head would run backwards and you could return to the point where the filament stopped flowing and take off again. There should be a speed option so you could rewind really fast until the head got close. It would also be nice to have the head travel to the front of the machine for cleaning. Once clean, have a manual extrusion option to verify the filament is again flowing properly. Seems easy to me with no parts to add, just a software upgrade.

The second option is to add an encoder somewhere along the filament path. As the filament travels along, the encoder verifies it is actually moving and allows the print to continue. Once the encoder detects no movement it will pause the print and move the head away to allow for cleaning. Or it might sound an alarm but continue printing so you could check what is going on. A few minutes after the alarm it will then pause. Seems more difficult adding another component which could create problems and is likely impossible on existing machines.

 

[gr5 2,069]

Several times I have been able to continue a print after a nozzle clog.

In fact I did one today. It works quite well.

I've done it on both UM Original and UM2. You have to know exactly where the print failed and where to continue it after cleaning the nozzle. Usually I can tell because I was there when it clogged and look at the Z height on the ulticontroller or there is something distinctive on the layer that I can compare to the layer height view in cura but today it was too difficult to do that so I Homed the head and platform and connected to the printer with pronterface and set the Z to verious heights with the G0 command:

G0 Z23.3

I used pronterface to move the head in X and Y until it was right over part of the print and was able to figure out *exactly* which layer it was printing when it failed and I started at the next layer up.

To continue where the print left off - the short answer is you delete the gcode up until the layer you want but in reality you have to keep some begining stuff and also add a G92 e.g.

G92 E7876.1

sets the extruder to 7876.1 mm as the *current* position so that when it moves to E7876.2 it only extrudes .1mm of filament and not 8 meters of filament.

 

[schafe 3]

That's easy for you to say. In other words, "Um, what?"

 

[gr5 2,069]

Look at the gcode of one of your existing prints. It's pretty easy to understand. There are comments. The gcodes are described here if you scroll way down:

https://github.com/ErikZalm/Marlin/

 

Next time you get a failed print, come back to this post and read this post and look through the gcode and try to figure it out and you can explain to the next person (by posting here) whatever it was that was unclear in my post above.

 

[gr5 2,069]

Maybe the part that is unclear is about the E axis. Marlin prints only in straight 4 dimensional lines where the 4th dimension is the extruder. On some machines (not 3d printers) when you say go to 0,0 then goto 1,1 it won't necessarily go in a straight line - it might move one axis faster than the other. Marlin always accelerates and travels in such a way that it moves in a straight line. Including the E axis. That way if it starts printing slowly, it extrudes slowly. As it accelerates, the E drive accelerates and all axis get to the end of the line at the same moment.

To take advantage of this all extruder moves (E axis) are in absolute mm when output from Cura. So if your print uses 100mm of filament the final move will end at E100.0. This is so much more obvious and clear if you just look at the gcode file yourself and look at the X,Y,Z and E values. Notice how the E values always increase whereas the X axis might go back and forth.