2nd Filament fails to reach extruder after bowden length calibration  

Re: 2nd Filament fails to reach extruder after bowden length calibration

Why is there a need for a bowden length calibration when we have a filament sensor in the extruder?

The sensor at the top of the extruder is disabled at present for the MMU as it is not totally reliable.

Peter

Re: 2nd Filament fails to reach extruder after bowden length calibration

The sensor at the top of the extruder is disabled at present for the MMU as it is not totally reliable.

I see, thanks for the clarification.

Maybe a mixed approach would make sense - something like:
- Push filament until sensor triggers or expected trigger point plus 10mm is reached.
- Turn on extruder and feed 10mm more, so it reaches the extruder gear.

On the other hand, when the next firmware version allows you to calibrate the 5 gears in the MMU separately, that should probably work just as well, if not better.

Thank you!

Re: 2nd Filament fails to reach extruder after bowden length calibration

Yep, exactly the same problem here...

Would greatly appreciate a quick patch-upgrade to the firmware!

That "ram-method" - which I did by accident as gear #1 appears to be the smallest of mine - might work short term, but the pulleys will eventually get filled with rasped material and then other problems arise...

Cheers,
Daniel

Re: 2nd Filament fails to reach extruder after bowden length calibration

...on second thought...

...why not make that figure fully dynamic and calculated on-the-fly...?

Reason being that the number of steps for the stepper is not only dictated by pulley-diameter, but also by filament-thickness!

Since the filament-sensor at the hot-end is still there and operational...
...why not use it to help?
When loading the filament, feed it just like now while counting the steps of the motor. Once the filament is detected at the hot-end-sensor, you know how many steps it took to get it from the MMU to the hot-end-sensor. Since you know the length of the bowden-tube from the MMU to the hot-end-sensor, and you know the distance between hot-end-sensor to the extruder-gears, it is a simple matter of mathematics to find the required number of steps from the sensor to the extruder-gears.
Surely not rocket-science which can easily be handled by the current electronics.
For calibration, all that's needed is the number of steps from the smallest pulley, which is easy to find: Advance each filament as if you would calibrate the tube-length, only instead up to the extruder-gears just to the filament-sensor and then safe the biggest number. That way, you can always fast-advance the filament for this number of steps minus safety-margin of say ten steps, then advance the filament slowly until the sensor triggers. Once triggered, continue to feed slowly while you calculate the required steps to reach the extruder-gears and then stop once that number has reached.
That way, even if the filament-diameter changes on the roll, the MMU will adapt for it.

I hope that gave some food for thought for the developers... 😉

Cheers,
Daniel

Re: 2nd Filament fails to reach extruder after bowden length calibration

Since the filament-sensor at the hot-end is still there and operational...
...why not use it to help?

Since the optical sensor can detect filament movement and not just whether the filament is there or not, you could use it to detect almost every load/unload failure and pause for the operator to clear the problem.

I'm hoping that re-enabling the optical sensor is in their development plan. But it sounds like the current thinking is that the optical sensor is actually less reliable than the MMU2 (which may be true, at least for some filaments).

Re: 2nd Filament fails to reach extruder after bowden length calibration

The different sized wheels seems to be causing another problem while swapping of filament but this time seems to be not retracting sufficient length of filament to do the swap.

I did some testing while using the smallest wheel diameter for my 1st filament, and it seems that from the bowden length calibration the variance of the length inserted seems to be quite big, sometimes ending in the middle of the gear while sometimes entering into the PTFE tube 😯 . If this is indeed the case, I'm not sure if the software can fix this since the length varies.

Re: 2nd Filament fails to reach extruder after bowden length calibration

...on second thought...

...why not make that figure fully dynamic and calculated on-the-fly...?

Reason being that the number of steps for the stepper is not only dictated by pulley-diameter, but also by filament-thickness!

Since the filament-sensor at the hot-end is still there and operational...
...why not use it to help?
When loading the filament, feed it just like now while counting the steps of the motor. Once the filament is detected at the hot-end-sensor, you know how many steps it took to get it from the MMU to the hot-end-sensor.

I like this! And even if the sensor's unreliable, you can take a few stabs at it the first time (or even every X'th layer) to get a good number, pulling it in and out till it is repeatable.

The real question is can you use that to update the flow multiplier as you go along. 😛

The different sized wheels seems to be causing another problem while swapping of filament but this time seems to be not retracting sufficient length of filament to do the swap.

I've heard people say the same . I can say the difference in hobbed gear can account for a couple centimeters of travel, easily (my worst case was 3.5 or so). Put a pretty small one on (maybe not the worst), and recalibrate.

Re: 2nd Filament fails to reach extruder after bowden length calibration

The issue here is that the filament sensor is not 100% with transparent filament, and I have permanently deactivated it on my mk3 as it would "err on the side of caution" and call for help mid-print sometimes even when plenty of filament exists. I get the better safe than sorry approach, but if the sensor is giving false positives, it might be hard to implement it. Imagine, whilst loading filament, the sensor doesnt trigger; the filament will just keep being pushed into the hotend and you could end up wasting a whole lot of filament because the MMU is waiting for the sensor to tell it to stop.

Unfortunately, without a sensor, it will always be an open-loop solution...

Maybe the solution is to have another FINDA at the top of the extruder 😀 I'd have the PINDA face sideways, perpendicular to the filament flow.

Re: 2nd Filament fails to reach extruder after bowden length calibration

Maybe the solution is to have another FINDA at the top of the extruder 😀 I'd have the PINDA face sideways, perpendicular to the filament flow.

Problem with that is that the FINDA requires gravity to operate 🙁 Unfortunately, magnets or springs affect the response.

Peter

Re: 2nd Filament fails to reach extruder after bowden length calibration

Maybe the solution is to have another FINDA at the top of the extruder 😀 I'd have the PINDA face sideways, perpendicular to the filament flow.

Problem with that is that the FINDA requires gravity to operate 🙁 Unfortunately, magnets or springs affect the response.

I'd rather see something like a scroll wheel encoder, so it can sense motion and not just whether the filament is there or not.

Almost every failure mode I've seen in the MMU could be detected with a combination of a reliable motion sensor at the extruder and the FINDA at the MMU.

Re: 2nd Filament fails to reach extruder after bowden length calibration

I'd rather see something like a scroll wheel encoder, so it can sense motion and not just whether the filament is there or not.

Almost every failure mode I've seen in the MMU could be detected with a combination of a reliable motion sensor at the extruder and the FINDA at the MMU.

I agree the almost, but having said that, my MMU now seems to be working perfectly. Not had a fail since I changed the firmware 🙂 and there's a few more photos in the KISS gallery now 🙂

I think I will try the clown fish tonight (with some rather odd colours).

Peter

Re: 2nd Filament fails to reach extruder after bowden length calibration

Support told me to increase the length of my filament for my bowden length calibration all the way down to the bottom of the bondtech gears..
Still didnt work =( I can see the filament go all the way through the bowden tube so it has to be falling short somewhere at the top of the extruder.

By the way, the current MMU firmware version is 0.9.0? Means we're not even running a version 1.0.0? 😮 if so, then maybe its not too shocking to see these issues as early adopters, may be better to revert to single filament printing for now 😉

Re: 2nd Filament fails to reach extruder after bowden length calibration

Support told me to increase the length of my filament for my bowden length calibration all the way down to the bottom of the bondtech gears..
Still didnt work =( I can see the filament go all the way through the bowden tube so it has to be falling short somewhere at the top of the extruder.

By the way, the current MMU firmware version is 0.9.0? Means we're not even running a version 1.0.0? 😮 if so, then maybe its not too shocking to see these issues as early adopters, may be better to revert to single filament printing for now 😉

Tho only thing shocked me was how I was naive to expect MMU will work. 🙂 Like I already haven't spend dozens hours of frustration about MK3 itself. The only thing you can rely on Prusa's products is that these are not products, but just prototypes sent to public. Horrible experience. I like prototyping and solving technical problems, but they should sell it the way -"try our prototype and tune it your way". If I had not expected it to work, I would be satisfied and enjoyed the tuning time instead of hours of frustration.

Re: 2nd Filament fails to reach extruder after bowden length calibration

Maybe the solution is to have another FINDA at the top of the extruder 😀 I'd have the PINDA face sideways, perpendicular to the filament flow.

Problem with that is that the FINDA requires gravity to operate 🙁 Unfortunately, magnets or springs affect the response.

I'd rather see something like a scroll wheel encoder, so it can sense motion and not just whether the filament is there or not.

Almost every failure mode I've seen in the MMU could be detected with a combination of a reliable motion sensor at the extruder and the FINDA at the MMU.

Maybe a simple microswitch attached to the extruder idler would be enough. It would detect a pressure on bondtech gears.

Or if we want it to trigger their rotation before the filament is pressed (not MMU2 use case), there could be 2 rollers connected by a spring of weak tension upper to gears. The microswitch would detect their disconnection, as the filament passed between them. If these were metal, they could act as a switch by just splitting them. (It would not work for conductive filaments in that case - I am not sure, if there are some). Or the already present sensor could watch splitting/connecting the rollers in a dark chamber looking for movement of well-defined printed part and so detecting the filament indirectly. It could be an attached wheel encoder, as you mentioned. Like in my drawing. 😀

Re: 2nd Filament fails to reach extruder after bowden length calibration

I have had the same issue as well. I have attached an image of the measurements that I have. Looking forward to finding the solution to this issue.
Cheers Andrew

Re: 2nd Filament fails to reach extruder after bowden length calibration

Does it mean I should approach support as well?

If you can wait a few days, then there should be new firmware available which will allow you to calibrate each filament feed.

Peter

Would be nice to have that new firmware soon... 🙂

Re: 2nd Filament fails to reach extruder after bowden length calibration

Would be nice to have that new firmware soon... 🙂

You may want to try my suggestion here: https://shop.prusa3d.com/forum/general-discussion-announcements-and-releases-f53/week-3-t24781-s10.html#p109359 (and the post following that one).

Peter

Re: 2nd Filament fails to reach extruder after bowden length calibration

Any update from PR on when the updated firmware will be coming?

Re: 2nd Filament fails to reach extruder after bowden length calibration

Hi,
what strikes me is the great variation of the 5 gears...
is a machined part, think this could be improved...

ron

Re: 2nd Filament fails to reach extruder after bowden length calibration

I measured mine. Two 6.30mm, one 6.35, one 6.40. Measuring is kind of finicky, so I did multiple measurements with similar levels of pressure and averaged.

The variation on mine doesn't seem that much. I reordered them anyway, but I suspect some folks got unlucky on the variation.