MK3S - New printer model  

Re: MK3S - New printer model

Am I wrong or this is a step back on the filament sensor?
This one will detect only the "presence" of the filament and not if it's actually jammed and not moving for whatever reason....
Your thoughts on this?

Why do you think that? The filament rolls the ball as it moves.. And the sensor detects the ball moving.. So you have the same functionatliy.. But rather than the senor looking at the filament (Which has many different characteristics) its "tuned" just to see the ball rotating.

This technique is not new.. Someone has already come up with this using a bearing, rather than a ball.

Does this help, or am I misunderstanding your questions

The new sensor is just an IR optocoupler ("light-based switch"). It will only report on or off, unlike the laser tracker that reports direction

Re: MK3S - New printer model

I remember PR switched from SCAD to STEP for some parts because they hit OpenScad limitations and had to switch.
Regarding the filament sensor. As far as I know the MK3 sensor was never used to detect jams or moving direction, only presence. The MK3S sensor, as already mentioned, can only detect the presence but not movement. So the functionality is still the same even more "dump" sensor is being used now.

Whoever wants to detect movement with a ball, might remember yourself on the old PC mouse. How many times do you had to clean up that damn axes? And how many times did the ball stuck because of the dirt?

Re: MK3S - New printer model

Mk3 sensor was regularly used to detect jams. That's the cause of the false positives during prints people have complained about.

The false positives have been resolved through the use of a proxy (bearing), so I'm a little disappointed to see Prusa move away from that capability.

I've been thinking about why I'm fairly disappointed by the Mk3s release. There are two reasons.
The first is purely personal, I've got a robust indirect filament sensor on my machine, that works perfectly, detects jams, and is currently working well with the Zerobeast firmware. However, if Prusa switches to a binary sensor, then it's likely that the firmware support for the older sensor is deprecated, so I'll lose out on jam detection.
The second reason is more objective. Detection of jams is a 3d printer feature that has now been let out of the bag. I think it is very likely that it will become a standard bullet point feature used to distinguish and market printers in the future (as it was for the mk3), even if Prusa have stepped away from it for the moment. Prusa laid the groundwork in developing the filament sensor and appropriate code, and the community has found ways to make that sensor applicable to all filament types and viable to use with the MMU. A filament sensor with jam detection is now a feature that any company can easily include on their printers, and will likely become commonplace. Everything is in place and ready, with both software and hardware examples of how to implement it in the wild. I think Prusa has made a mistake here by switching back to the binary sensor, and leaving the gap open for other companies to introduce the feature.
tl:dr. It's likely that one day Prusa will need to have a filament sensor that can detect jams. I think it is a mistake to step back from it now that most of the hard work has been done.

Re: MK3S - New printer model

Mk3 sensor was regularly used to detect jams. That's the cause of the false positives during prints people have complained about.

The false positives have been resolved through the use of a proxy (bearing), so I'm a little disappointed to see Prusa move away from that capability.

I've been thinking about why I'm fairly disappointed by the Mk3s release. There are two reasons.
The first is purely personal, I've got a robust indirect filament sensor on my machine, that works perfectly, detects jams, and is currently working well with the Zerobeast firmware. However, if Prusa switches to a binary sensor, then it's likely that the firmware support for the older sensor is deprecated, so I'll lose out on jam detection.
The second reason is more objective. Detection of jams is a 3d printer feature that has now been let out of the bag. I think it is very likely that it will become a standard bullet point feature used to distinguish and market printers in the future (as it was for the mk3), even if Prusa have stepped away from it for the moment. Prusa laid the groundwork in developing the filament sensor and appropriate code, and the community has found ways to make that sensor applicable to all filament types and viable to use with the MMU. A filament sensor with jam detection is now a feature that any company can easily include on their printers, and will likely become commonplace. Everything is in place and ready, with both software and hardware examples of how to implement it in the wild. I think Prusa has made a mistake here by switching back to the binary sensor, and leaving the gap open for other companies to introduce the feature.
tl:dr. It's likely that one day Prusa will need to have a filament sensor that can detect jams. I think it is a mistake to step back from it now that most of the hard work has been done.

Fully agree, hence why in my previous reply i was clearly "underexcited" about this "upgrade" which looks more like a downgrade to me.
Ok the MK3 filament sensor doesn't work well with some filaments, fair enough at this point I prefer to use the filaments it works well with and switch to an indirect mode rather than going with the MK3s which is a step-back 🙁
What's the point of detecting filamente presence if the filament ain't moving?
It seems just useful for spools runouts....

Re: MK3S - New printer model

Yes... I'm hoping someone remixes the new extruder body and parts to be compatible with the old sensor and/or indirect design. Because I definitely want the touted ease of disassembly/access! 🙂

Re: MK3S - New printer model

I think I have the ultimate Sensor 😀
I use clear filaments in many colors and I have never an error from my filament Sensor 😀
But ok sometimes auto load doesn't work before I restart my printer. But that's OK 😀
This upgraded Extruder was cool but the sensor I think it's not so good. Same thing like the powder coated sheet( I am still waiting for it...) cool feature but doesn't work ^^

Re: MK3S - New printer model

Mk3 sensor was regularly used to detect jams.

well, my mk3 sensor never detected ANY jams. And i've had quite a few ones, i even posted a topic regarding stuck filament which happened on hot summer days because of filament motor became too hot and as a consequence gears became hot, too so PLA filament became soft like a rubber and stuck. But printer never reported jam. Even trinamic drivers didn't report nothing, i just heard those famous clicks from gears when gears move and grind filament.

And, not long ago my sensor died with no reason. Luckily Prusa sent me a new one under warranty.

I think that problem was not because of sensor reporting "stuck filament" but because sensor didn't see very well some colors, so in the middle of the print he decided that filament is no longer present and reported "out of filament" error.

Re: MK3S - New printer model

looking for the new-to-print parts but seem not able to find stl's somewhere.

Does anyone have or found more info on this?

Re: MK3S - New printer model

looking for the new-to-print parts but seem not able to find stl's somewhere.

Does anyone have or found more info on this?

https://www.prusa3d.com/prusa-i3-printable-parts/

https://github.com/prusa3d/Original-Prusa-i3/tree/MK3S/Printed-Parts

Re: MK3S - New printer model

Is it turned off? False jam detection is commonplace. Not detecting jams is unusual (unless turned off).

Re: MK3S - New printer model

Is it turned off? False jam detection is commonplace. Not detecting jams is unusual (unless turned off).

Re: MK3S - New printer model

Yes... I'm hoping someone remixes the new extruder body and parts to be compatible with the old sensor and/or indirect design. Because I definitely want the touted ease of disassembly/access! 🙂

100% agree!!!

Re: MK3S - New printer model

Is it turned off? False jam detection is commonplace. Not detecting jams is unusual (unless turned off).

Really? With Prusa's firmware? I've had the Mk3 almost a year, experienced a few jams, and printed plenty of things, keeping the firmware up to date. The filament sensor is on, autoload works great, but I've never had it detect and warn me (correctly or falsely) about a jam. Is there some setting aside from filament detection and autoload to enable this behavior? Or perhaps it's a new feature (more recent than my last jam)? It has warned me when running out of filament, but never about a jam.

Re: MK3S - New printer model

The ability to swap hot ends quickly is a bonus for me (I have one for "normal" filaments and one for sub 100c print filaments). I've got a MMU2 coming at some point but was thinking of printing the parts now anyway but:
1) What parts do we need to print to do an upgrade (eg all of them?)
2) Can we use the new parts with the existing filament sensor if I disable it?

Thanks
Nathan

Re: MK3S - New printer model

Really? With Prusa's firmware? ...

No jam detection with stock firmware. Not sure why martin is spreading this rumor.

Re: MK3S - New printer model

Are you talking about when linked to an mmu?

Because jam detection is definitely present in the stock firmware on a basic mk3. That’s where all the complaints of false positives during prints comes from.

The filament runout, and jam detection are the same mechanism. They both go off when there’s no filament movement detected.

Re: MK3S - New printer model

The filament runout, and jam detection are the same mechanism.

No. The filament sensor is not being used to detect motion, only presence of the filament:
https://github.com/prusa3d/Prusa-Firmware/issues/505
You can also check the firmware sources and point me to the right direction if I'm wrong.

False triggering is happening when the filament is dirty or the filament is too shiny for the sensor. In this case the firmware assumes there is no filament.

Re: MK3S - New printer model

I think there may be a misunderstanding here about how the Mk3 filament sensor works.

The filament sensor works by taking many rapid images, comparing them with the previously captured image, and reporting back a vector (x and y velocity). As part of the background on my indirect sensor mod, I dug into the Prusa code to see how it was using these values, and whether there would be any problem with sensing an object with partial rotation. Turns out there isn't, and Prusa are using just a single vector (Y axis I think).
So, your claim that "the filament sensor is not being used to detect motion" is not correct. That is literally all it can sense.

This means that the sensor cannot distinguish between no filament, and filament present, that isn't moving. Both scenarios show no movement, so the filament sensor cannot be used to distinguish between the two. Your suggestion that the filament sensor only detects "presence of filament", is incorrect. The sensor can't distinguish between no filament, and filament present but not moving.

The way the sensor works for runout (and jamming, same mechanism), is that the alert flag is set when there should be filament moving (eg: When extruding), but the sensor reports no movement. This is why if the firmware can detect filament runout (no movement reported when expected), then you also have jamming detection (no movement, when expected). They are the exact same scenarios, using the exact same mechanism.

Conversely, if the sensor shows movement, when movement isn't expected, then autoload is activated. A further demonstration of the motion sensing, rather than presence of filament detection is that autoload only works in one direction. If you reverse the filament direction, then autoload doesn't activate.

The example you linked to is a peculiar case where the filament is jammed and "jumping", presumably enough to continue registering as moving. This is likely avoided with either refinement of the error count (how many, and how fast filament sense errors occur before setting off the flag) or through use of an indirect sensor.

As further background, much of the latest work on the Zerobeast firmware has been around deactivating the sensor during filament swaps to avoid setting off the jam/runout flag.
Also, the filament sensing code is here.

Re: MK3S - New printer model

Is there difference in printed parts beside the Extruder body, Y-Axis belt holder and tensioner. My Mk3 was shipped Jan 24, 2019.
If I had of known a new version was about to hatch 2 weeks later I would have waited and got the upgrade.

That is why I am asking. If there are significant printed part upgrades then I will rebuild the printer in whole. If not I will print the few part differences and only replace those that offer marked improvement. Like the Y axis belt holder redesign. That does need improvement.
Another improvement is a better way to handle X belt tension. It would be better with a 4 screw, slotted motor mount with a pair of set screws to push motor parallel to provide belt tension. May be redesigning that myself.

Re: MK3S - New printer model

Thanks for the detailed explanation Martin. It makes fully sense from the technical sensor perspective. The sensor (pat9125) can only detect a movement. But the precision of this sensor is not high enough to detect a jam in most cases. That's why practically it doesn't detect jams (at least for me it never does), but it detects no filament. The head movement, the vibrations, the sensor resolution ... you will always detect movement as long as a filament is inside the scan area. So for me it was always a filament run-out detector only and if the new sensor will work even for fully transparent filament then it's an upgrade in my eyes.

But thanks again for the detailed explanation. I understand your point of view now.