MK3S+ filament sensor?  

RE:

Pity about the cookout.

The filament sensor is very similar to the Mk3s referenced in that thread but uses a different arm, extruder body and the latest revision sensor board.   The magnets and steel ball and general principle are the same.    Of course, the magnets must be positioned to push away from each other. If one is reversed, the sensor will always indicate that filament is present.  The tab that blocks the sensor is visible if one removes the sensor, and you can see if it moves when filament is inserted.  Likely faults are - pivot bolt too tight, magnet reversed, binding on sensor or foreign material 

Good luck!

RE: MK3S+ filament sensor?

Thanks.  Odd that it worked and then didn't.  I suppose the electronics could have failed.   Time for some exploratory surgery.  Will report back with findings.

RE: MK3S+ filament sensor?

Finally got back to the printer and tested out the filament sensor.  It seemed to work perfectly, paused the print and the menu went to 0 for both PINDA and sensor.  However, the menu prompt asked me to hit the knob to unload.  So, I am hoping there is a setting to "just do it," but I have not found it.  Is there one?

RE: MK3S+ filament sensor?

Finally got back to the printer and tested out the filament sensor.  It seemed to work perfectly, paused the print and the menu went to 0 for both PINDA and sensor.  However, the menu prompt asked me to hit the knob to unload.  So, I am hoping there is a setting to "just do it," but I have not found it.  Is there one?

RE: MK3S+ filament sensor?

Support says no such feature.  I put in a request:  https://github.com/prusa3d/Prusa-Firmware/issues/3445#issue-1171411511

RE:

Releasing the sensor cover a bit (or even adding a small shim, as recommended in those threads) did not help? You can always switch off the IR sensor in the settings.

Regards

Chris

RE: MK3S+ filament sensor?

I'm confused by your response, Chris.  The design seems to work perfectly.  It detected when the filament ran out.  It just didn't unload or turn off the heat.  Are you saying the heat would turn off if the case was loose?  I had thought the problem was that it did not detect the end of the filament, but it did.  It just kept cooking the filament that was still in the hot end.   It seems like and auto-unload would be perfectly reasonable and helpful at that point.

RE:

Maybe I misunderstood your problem. I thought you had an issue with filament still being physically loaded but the sensor indicating 0 for unloaded status.

Following the completion of a print, the printer should cool down automatically, whilst the filament is still loaded. If it does not I wonder whether this is an issue of a wrong gcode, but I don´t know enough about gcode.

Cheers

Chris

RE: MK3S+ filament sensor?

Sadly, the job did not pause

The design seems to work perfectly.  It detected when the filament ran out.

These two statements seem to contradict each other.

If it detected the run out and paused the print, the problem could have simply been that there was something about the end of the filament that prevented it from backing out (e.g. bend from the spooling).

If it didn't pause the print, that would indicate that either the filament sensor is disabled (Settings menu) or it for some reason didn't recognize the filament had run out. Are you 100% sure that the filament passed the steel ball? It could be that the jam came first and the filament simply hadn't advanced far enough into the extruder to trip the sensor? Have your been printing much since the last time you took apart the extruder? Maybe there is some debris that is/was preventing proper operation of the sensor mechanism?

Given that you report manual testing works, that would seem to indicate the IR board itself is probably healthy (still could be heat or movement related which you wouldn't see in a static test). Another good test of more real world conditions would be to start a print and once it's been going for awhile, use your flush cutters to snip the filament just above the extruder and watch how it reacts.

As far as the process goes, when it pauses, it asks you to unload it first (I assume to prevent a false runout from letting your filament wind back and cause a tangle?). There is a set timer (not configurable as far as I'm aware) where the hotend heater will turn off if it is idle too long (I want to say either 10 or 30 minutes) and that isn't long enough to cause issues with PLA. It leaves the heater on initial with the expectation that you are going to change the filament promptly and don't want to have to wait for it to reheat.

RE: MK3S+ filament sensor?

I'm sorry, I did not report the events very clearly.   I was describing two different events.   Event 1 - the unattended job that clogged my hot end.  Event 2 was when I tested the sensor (after clearing the hot end).  Event 2 demonstrated that the sensor worked perfectly.  Event 1, I misinterpreted as the sensor not working.  Event 2 showed that the sensor detected end-of-filament (EOF), paused the job, and set the codes to 0. 

The fact that the hot end was clogged on event 1 I intrerpreted as EOF not detected, a common event on my mk3s.  That seems to have not been true per the results of event 2.   However, the filament did continue to burn on event 1, so it would seem the heat did not turn off.   

In event 2, I was prompted to push the knob to unload the filament.   Since I was not attending the printer at event 1, I must have received the prompt and when no action was taken, the filament did not move out of the hot end and cooked/burned (I assume).  BTW, I was not there when the event 1 was discovered, so I don't really know the state of the display at that time.  I assume it was requesting the knob be pushed, but I can't be 100% sure of that.

I asked support if an unattended unload was possible when the sensor detected EOF.   They said no, but that I could request as such in github.  The current behavior seems a wrong design to me.  In what case would you not want the filament to unload on EOF.  Why does it need to be attended?  Also, if EOF is the state, why keep the heat on?   At the very least it should time out on the prompt and turn off the heat.  

See what I did there with EOF 🙂

RE: MK3S+ filament sensor?

" There is a set timer (not configurable as far as I'm aware) where the hotend heater will turn off if it is idle too long (I want to say either 10 or 30 minutes) and that isn't long enough to cause issues with PLA. It leaves the heater on initial with the expectation that you are going to change the filament promptly and don't want to have to wait for it to reheat."

Hmmm, I'll test this timeout.  It would seem my timeout didn't work, because on my event 1, the hot end was really jammed and black material came out when I finally got it clear.   It's a little hard to do as the menu is fixed with the prompt to hit the knob, so I can't look at the temperature reading, but I'll figure out a way to measure the nozzle temperature after 30 minutes.

RE: MK3S+ filament sensor?

Posted by: @ray-of-hope

Since I was not attending the printer at event 1, I must have received the prompt and when no action was taken, the filament did not move out of the hot end and cooked/burned (I assume).  BTW, I was not there when the event 1 was discovered, so I don't really know the state of the display at that time.  I assume it was requesting the knob be pushed, but I can't be 100% sure of that.

If it detected a filament runout, it would have prompted for the unload. That prompt does not timeout (I've had mine wait about 36 hours (ran out of filament before the new spools arrived)), it simply turns off the nozzle heat if it is idle too long. After you press the knob to start the unload process it will then pause the process while it reheats the nozzle.

If it had been left for a significant amount of time and the nozzle was still hot, then it's likely that it thought it was still printing (e.g. it clogged close to the end of the filament, but before the sensor was tripped). Without direct evidence of someone familiar with the printer being there when the event happened, it's hard to jump directly to a firmware/hardware issue (especially with the results of "event 2") when other possibilities exist.

The current behavior seems a wrong design to me.  In what case would you not want the filament to unload on EOF.  Why does it need to be attended?

I'm sure there are others, but while (my experienced with) the IR sensor is reliable in the stock form (with the MMU is a different story) there can be false events. If it incorrectly sent a false event while there really was still filament present and automatically unloaded, that could allow the filament to snap back to the spool and get tangled. Such tangles aren't always obvious and don't immediately exhibit problems, so it could end up ruining a later print (or the same one after multiple more hours of printing). I would much rather have to press the button to confirm I'm ready for the unload rather than deal with the results of a tangle.

 Also, if EOF is the state, why keep the heat on?   At the very least it should time out on the prompt and turn off the heat.

Unless you stumbled onto a bug in the firmware (always possible), it does turn the heat off after a relatively short period of time.

RE: MK3S+ filament sensor?

As I did not attend to the event 1 failure, I can't really say what happened and so your hypothesis of a "false event" at least explains why my hot end was clogged.   Hmmm, a configurable heat timeout would be a nice addition.   I'd set mine to 5 minutes.  We don't attend to jobs like we should.  

I get the spool tangle argument.  Done that, don't want to do it again.

Testing my timeout experience now.

RE: MK3S+ filament sensor?

Posted by: @ray-of-hope

Hmmm, I'll test this timeout.  It would seem my timeout didn't work, because on my event 1, the hot end was really jammed and black material came out when I finally got it clear.   It's a little hard to do as the menu is fixed with the prompt to hit the knob, so I can't look at the temperature reading, but I'll figure out a way to measure the nozzle temperature after 30 minutes.

Unfortunately because it was unattended, event 1 is not a reliable data point for you to base this on unless you can figure out how to recreate it while it is monitored so you know what the exact order of events is.

I can't find an explicit definition (e.g. on Prusa's site), but I'm finding references that the timer is 30 minutes. Cause a run out event, leave it at the "press to unload" prompt, and then come back after an hour and the nozzle should be cold. If it's not, then there is something up with the firmware.

RE: MK3S+ filament sensor?

"come back after an hour and the nozzle should be cold"... in progress.     Wonder what the timing is for when the hot end clogs, filament burns?

RE: MK3S+ filament sensor?

Posted by: @ray-of-hope

"come back after an hour and the nozzle should be cold"... in progress.     Wonder what the timing is for when the hot end clogs, filament burns?

There is no time out for a clog as it has no way to detect that (the original laser sensor on the MK3 could, but talk about false positives!). So as long as the printer is still processing gcode, it will keep the nozzle heated and move the extruder as if it was still printing.

RE: MK3S+ filament sensor?

No, I meant, say the time out fails and the filament keeps being heated.  In theory 30 minutes of heating is ok as that is the known timeout.  When does the burning begin where the hot end is clogged?  1 hr?  3 hrs?  Just curious.  Don't want to try to find out by experimentation.

Interesting.  I tested the EOF again, set a clock timer for 40 minutes and then, not having a thermometer, decided to get near the nozzle with my finger.  I was able to touch it.  The heat had shut down.  The bed stayed hot.  I was able to push the knob and unload, so no clog.  

I guess the false event hypothesis is all that is left.   This is interesting. We will need to monitor the EOF more closely as with the MK3S.  Unless we find we enjoy clearing the clog (said no printer operator ever).  We had hoped the mk3s+ would prevent this particular problem.

So, the 

RE: MK3S+ filament sensor?

Posted by: @ray-of-hope

No, I meant, say the time out fails and the filament keeps being heated.  In theory 30 minutes of heating is ok as that is the known timeout.  When does the burning begin where the hot end is clogged?  1 hr?  3 hrs?  Just curious.  Don't want to try to find out by experimentation.

There are many factors that would go into such a time (type of plastic, additives, temp, etc..), so I don't know if there is a simple estimate

I guess the false event hypothesis is all that is left.   This is interesting. We will need to monitor the EOF more closely as with the MK3S.  Unless we find we enjoy clearing the clog (said no printer operator ever).  We had hoped the mk3s+ would prevent this particular problem.

I would start with always making sure you have a clean tip at the end of the filament (e.g. clip off the last 1-3cm) with no kinks or bends that could prevent it from going into or being ejected from the extruder. That and not having the sensor enabled are the two big culprits of the IR sensor not doing it's job.

My IR board is about to turn 3 in April or May and the only (non MMU related) failures to detect run out that I've experienced has been due to not giving it a clean end that then gets hung up.