Filament sensor question.  

Re: Filament sensor question.

The optical sensor is currently not used with the MMU2.

Peter

Re: Filament sensor question.

ok, turned it off then.

Re: Filament sensor question.

The optical sensor is currently not used with the MMU2.

Peter

It IS used for filament run out detection.

Re: Filament sensor question.

It IS used for filament run out detection.

As far as I am aware, that is the job of the FINDA. By the time the back end of the filament gets to the optical sensor, there would be no way that the "tail end" could be removed. However, if it is detected by the FINDA, it can still be unloaded.

And FWIW, my optical sensor is just sitting on the workbench next to the printer; it has been for some time.

Peter

Re: Filament sensor question.

i read a little more on the sensor. the sensor is only in use with the MMU2 when in single mode.

if you enable auto-deplete, then it will also use the sensor and keep switching between the 5 spools when one runs out.

it seems the sensor in the current firmware is disabled to be used as a filament check when in multi-mode.

but i could be wrong and what i read could be out of date as well.

i turned my sensor off for the last multi color print and the printer worked fine. I've had the sensor tell the machine randomly that it was out of filament when it was not and it just causes more trouble then it is worth. I have mistakenly run out of filament before but the sensor in its current state is not worth the trouble of false readings and fires!

Re: Filament sensor question.

Just asked MMU2 development guys for confirmation on this:

Quick question: Which sensor is used for the "Auto-Depletion" function?
If it's the FINDA, is the optical sensor used at all with MMU2?

Answer: Finda, optical senzor is not used at all...

Peter

Re: Filament sensor question.

awesome thanks!

Re: Filament sensor question.

Filament sensor operation would be fantastic though, especially if it detected the movement of the filament. I think you could prevent false loads and therefore skipped layers. I have tried activating the filament sensor to work (I had no idea about it melting parts? is that true?), and what I have found so far is that it could prevent some fails. The sensor worked alright with black color (I mean no skipped layers, nothing to notice), a bit worse with orange and awfully bad with white, where I had a lot of this backtracking of the filament, beeping, showing the tip (then I had to fix it just in case and put it again in position), then it required the button to be pressed not once but twice (why wouldn't it just nicely retry a couple of times before showing the tip? the filament did not even need to backtrack all the lenght of the bowden tube...), and all this happening for no good reason, because the machine stopped in the middle of a good layer to perform this little dance. Checking should only be done at the start of the load (say the first 20-30 seconds), but it is unnecessary in the middle of a layer (what is it looking for? a filament end situation? then, why does it show the tip?)

I have seen somewhere a modified filament cover (I keep on searching but I cannot find it now) which pressed a metal ball with a small spring against the filament (much like the ball of the FINDA probe but with no need of gravity). The filament sensor is aimed at the ball, not at the filament. Have you seen this mod? Where can I find it? Do you know if it works? Would it detect the movement? Even if it did, I guess the code would have to be rewritten to make it really work alright or perhaps not...

Re: Filament sensor question.

Filament sensor operation would be fantastic though, especially if it detected the movement of the filament. I think you could prevent false loads and therefore skipped layers. I have tried activating the filament sensor to work (I had no idea about it melting parts? is that true?), and what I have found so far is that it could prevent some fails. The sensor worked alright with black color (I mean no skipped layers, nothing to notice), a bit worse with orange and awfully bad with white, where I had a lot of this backtracking of the filament, beeping, showing the tip (then I had to fix it just in case and put it again in position), then it required the button to be pressed not once but twice (why wouldn't it just nicely retry a couple of times before showing the tip? the filament did not even need to backtrack all the lenght of the bowden tube...), and all this happening for no good reason, because the machine stopped in the middle of a good layer to perform this little dance. Checking should only be done at the start of the load (say the first 20-30 seconds), but it is unnecessary in the middle of a layer (what is it looking for? a filament end situation? then, why does it show the tip?)

I have seen somewhere a modified filament cover (I keep on searching but I cannot find it now) which pressed a metal ball with a small spring against the filament (much like the ball of the FINDA probe but with no need of gravity). The filament sensor is aimed at the ball, not at the filament. Have you seen this mod? Where can I find it? Do you know if it works? Would it detect the movement? Even if it did, I guess the code would have to be rewritten to make it really work alright or perhaps not...

the problem with the sensor is it actually causes more issues then it resolves. the same with layer shift detection.
in the beginning these 2 items on my MK3 caused more un-intended failures then anything else. you would be 50 hours into a print and have the entire thing ruined by either of these features. in addition the Sensor does have a problem with internal failure causing it to get hot enough to cause a big mess. there is also the problem of False readings

while these feature on paper are fantastic and would solve some problems they just don't work 100% in the real world. in a complex system if you can run it simpler then you will decrease the chances of things going wrong.

i think i have seen that FINDA cover on thingiverse the problem is the spring pushing against the filament is going to introduce more friction and solves one problem maybe, yet makes another much worse.

a lot of these fixes people tried were before the new spool rollers were posted. with your machine down if you can get it running as a normal MK3 in single color mode and you have not changed to the new rollers on the spools , then start printing these new spool roller parts out convert all 5 and then when the PINDA replacement comes in and you put the machine back together as a MMU2 you should start seeing a big improvement.

Re: Filament sensor question.

My MK3 runs smoothly in the single color mode since day one. The first thing I did was to change that stupid spool holder for one I had designed myself that is virtually frictionless. The reason: the old one did not fit into the IKEA lack enclosure. When I received the MMU2 I just threw away the spool holders and printed 5 new ones of this design, so spool holding is not an issue for me.

So the question could be: if the filament sensor run RELIABLY and SAFELY, would it be useful to prevent false loads and therefore skipped layers?

To find it out I am designing now a new cover for the filament sensor. The sensor will try to detect the movement of a 7.5 mm steel bearing ball (instead of trying to detect the filament), but I will use a small MAGNET to press the ball against the filament instead of using a spring. This method should barely have any friction and if it does, I will just move the magnet a bit further away. It has to be close enough to press the ball agaist the filament and far enough so this pressure force is not an issue. I would say this objetive is easy to obtain, because magnetic force decreases with the distance squared, blah blah

The part, as I am designing it, is very simple, it is just a 12 mm slab that holds the sensor, the ball and the magnet, and it is sandwiched between the extruder body and the actual filament cover (the one that has the festo coupling) which I will keep. I will just use longer bolts to attach it.
I will print it in ABS so it withstands a bit more temp and maybe I could insert some cooling gills so there's some airflow to cool the sensor... If the ABS melts there's alays polycarbonate or even modifyng the filament cover to improve the airflow

Anyway, it seems like it's going to take some time until I can test out this gig... Any ideas?

Re: Filament sensor question.

I've designed something similar to this before.

when you design the magnet area i would setup the magnet to be externally adjustable. use a grub/set screw that you can access from outside to allow you to move the magnet up and down. you can buy gub/set screws that can be 30mm long if needed. not only are magnets effected by distance, but every magnet has a slightly different magnetic field and Gauss strength that will effect sensors differently. You will want to make it very flexible so you can dial it in.

Re: Filament sensor question.

In case you are wondering how did I manage to get the MMU2 inside the IKEA lack. Yeah, I close all sides and put something on the top to cover the hole when I print ABS but it works alright

Re: Filament sensor question.

It looks like that magnet position regulation is unnecessary after all, and it really doesn't matter which of the magnets I use or even if I use them all together. In the photo you can see the high tech gig (NASA style) I have used to measure the constant of the magnets, and the result of the tests suggest that to overcome the inertia due to the maximum acceleration of the printhead during infills (1250 mm/s2) distances from the magnet to the ball have to be very close.

What this means is that this method of filament detecting might not be good for a filament end situation, because the printhead is moving and there could always be some false positives. The ball might move due to the movement of the printhead and this would be understood by the MK3 as if the filament is present though it might not.

But I think the method might be valid if the printhead is static, and that is exactly the situation during the filament load, and what I am looking for is to prevent false loads (skipped layers), so I will give it a try.

There might be also some collateral damage with some noise from this metal ball banging in the printhead much like a baby's rattle... I'll try to give it as little play as I can.
Little room also for those cooling gills, I`m afraid l might have to go on without them. I guess a complete rework of the extruder would be mandatory to do this properly and I have found after a fast look in thingiverse that this extruder designing thing must be some kind of sport. Dozens of designs available. None of them had what I wanted though. In any case I would need two printers to design an extruder. One to print the parts, the other to test them. Otherwise I might go crazy assembling and disassembling the device. Unfortunately I sold my old chinese clone (Kuongshun) so I will leave this complete redesign for other people, though it looks an appealing task.

Re: Filament sensor question.

Easy does it. The new filament sensor fix is done and installed. I only need PRUSA send the parts to repair the MMU2 and I will be able to test it. My first impressions on the thing, the tip goes through much easier than I had thought, and once in operation there's barely any friction, so if the sensor can detect the movement of the ball I guess it should work alright, also for the filament end situation, because in this case the distance between the magnet and the ball is smaller, and magnetic force should be then big enough to overcome the worst case inertia. There's also more air for the sensor to breath than as it was installed previously, so heating might not be such an issue as before.
If I had the stp file for the extruder body, I think it probably would be easy to modify it and do something more elegant to hold the sensor, but as I have installed it now, it's really easy and looks effective enough.

Re: Filament sensor question.

FWIW, I am currently working on a lower sensor:

The extruder body has been modified so that it grips the PTFE tubing, which goes to within about 5mm of the Bondtech gears.

The sensor attachment screws into the holes of the old sensor cover - the optical sensor has been removed and the new FINDA connects to the MMU electronics, using pins 1 to 3 of the spare connector.

The PTFE is 2.5mm ID.

However, my prints have significantly improved with just the extruder body and PTFE; I really don't think I need the lower sensor for anything other than softer filaments (they don't feed from MMU as well and require recalibration).

Anyway, the PINDA is triggered by a 1 (Euro) Cent coin which is a tight push fit into the holder which is part of an amended idler door, so the probe triggers when the filament is loaded into the Bondtech.

Peter

Re: Filament sensor question.

Yeah, I had thought that if the optical sensor failed to detect the moving ball, a pinda could be used for it, but in the system I have designed the ball only moves about 1mm perpendicular to the filament and very little on the other axis, but it does roll. Also, I think that the pinda system would not detect a failed load, that is, when the filament has passed the sensor but does not move anymore.
With the magnet system I have designed, as long as the filament is moving, the ball does wobble slightly up and down and it also rolls. I am eager to see if the optical sensor is able to detect this movements

Re: Filament sensor question.

Also, I think that the pinda system would not detect a failed load, that is, when the filament has passed the sensor but does not move anymore.

Well, the PINDA is triggered by the idler door being opened 😉 so it's not possible to get a "failed load" (unless there is a blockage in the extruder itself, after the Bondtech gears). Having the coin attached to the idler door, It only recognises when the filament is loaded, or for that matter, properly unloaded from the extruder.

Peter

Re: Filament sensor question.

I was thinking simply having a second FINDA where the optical sensor was would be sufficient enough to prevent loading/unloading problems because its easier to ensure that filament moves a precise distance when said distance is short. The MMU2 pushes the filament through the Orange PTFE tube rather fast, and if the filament is going to slip or not go a correct distance, it will happen then.
If it detected filament at the print head, then it might be easier to push it a short distance slowly then a long distance quickly.

Also, I never really had a problem with the optical sensor really, and it seemed to do a great job at detecting the presence of filament. Most issues I have seen are about it thinking there is a jam, which isn't really needed to help improve the MMU2.

Just my 2 cents, though I might just be stating the obvious!

Re: Filament sensor question.

you could even use a momentary switch as long as the filament sends any type of trigger back it would work