Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
I have been reading through this thread and looking at it from the viewpoint my pre-retired life as an engineer.
This issue with the filament sensor and the 3.4.0 firmware smells to me like a QC problem with the sensor; in short the there is too much variation in the output from the sensor that pass QC checking from their manufacture and the old firmware was taking that variation in sensor output into the way it dealt with it... The solution would seem to be much tighter QC on the sensor starting with testing every shipping unit to be sure it met the needed tight tolerance. Will this increase the unit cost of the sensor? in the short term yes, but given that the 3.4.0 firmware seems to require tighter output from the sensor to support the MMU2.0 this might be the best way forward and in a relatively short time frame the cost of the shipped sensor should drop back as manufacturing adjusts to the new requirement.
The downside will be the fact that the end user will bear the cost of any non warranty replacement and PRUSA will be on the hook for the Warranty replacement.
I come to this conclusion without a horse in this race yet (as my MK3 kit is sitting in Customs at JFK as I type this) but the fact we have a user running 9 printers in a farm that has no problems with the release firmware, as well as all the others out there that if they have no problems will be silent as far as this thread is concerned, makes this my best guess as to what is going on.
Filament sensor engine 2.0
This release brings a major update to the filament sensor. Part of the software responsible for the analysis of the filament movement was rewritten to increase the precision and reliability of the sensor. With the new firmware, the filament runout should be always detected in time and the number of false detections drops significantly. Previously MK3 and MK2.5 shared similar values for the evaluation and it was sometimes causing issues on the MK2.5, this is now fixed.
Engine 2.0 brings also new feature for the filament evaluation. During the filament load sequence, the filament is scanned using the IR sensor and evaluated. As soon as the filament is loaded to the extruder the printer knows, whether it can reliably control the filament flow during the print or not. In case the sensor can't read the filament movement properly, the user is informed on the printer's screen with an option to turn off the sensor.
Poor_filament_detection.jpg
Poor_filament_detection.jpg (109.66 KiB) Viewed 4160 times
Filament sensor precision is greatly influenced by the cleanliness of the sensor itself. The easiest way is to use a compressed air. There is a dedicated hole in the x-carriage exactly for this purpose. See the picture below to compare the clean and dirty sensor.
cleanVSdirty_sensor.jpg
cleanVSdirty_sensor.jpg (319.35 KiB) Viewed 4160 times
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
I have been reading through this thread and looking at it from the viewpoint my pre-retired life as an engineer.
This issue with the filament sensor and the 3.4.0 firmware smells to me like a QC problem with the sensor; in short the there is too much variation in the output from the sensor that pass QC checking from their manufacture and the old firmware was taking that variation in sensor output into the way it dealt with it... The solution would seem to be much tighter QC on the sensor starting with testing every shipping unit to be sure it met the needed tight tolerance. Will this increase the unit cost of the sensor? in the short term yes, but given that the 3.4.0 firmware seems to require tighter output from the sensor to support the MMU2.0 this might be the best way forward and in a relatively short time frame the cost of the shipped sensor should drop back as manufacturing adjusts to the new requirement.
The downside will be the fact that the end user will bear the cost of any non warranty replacement and PRUSA will be on the hook for the Warranty replacement.
I come to this conclusion without a horse in this race yet (as my MK3 kit is sitting in Customs at JFK as I type this) but the fact we have a user running 9 printers in a farm that has no problems with the release firmware, as well as all the others out there that if they have no problems will be silent as far as this thread is concerned, makes this my best guess as to what is going on.
Filament sensor engine 2.0
This release brings a major update to the filament sensor. Part of the software responsible for the analysis of the filament movement was rewritten to increase the precision and reliability of the sensor. With the new firmware, the filament runout should be always detected in time and the number of false detections drops significantly. Previously MK3 and MK2.5 shared similar values for the evaluation and it was sometimes causing issues on the MK2.5, this is now fixed.
Engine 2.0 brings also new feature for the filament evaluation. During the filament load sequence, the filament is scanned using the IR sensor and evaluated. As soon as the filament is loaded to the extruder the printer knows, whether it can reliably control the filament flow during the print or not. In case the sensor can't read the filament movement properly, the user is informed on the printer's screen with an option to turn off the sensor.
Poor_filament_detection.jpg
Poor_filament_detection.jpg (109.66 KiB) Viewed 4160 times
Filament sensor precision is greatly influenced by the cleanliness of the sensor itself. The easiest way is to use a compressed air. There is a dedicated hole in the x-carriage exactly for this purpose. See the picture below to compare the clean and dirty sensor.
cleanVSdirty_sensor.jpg
cleanVSdirty_sensor.jpg (319.35 KiB) Viewed 4160 times
I'm a 20 year electrical/RF engineer and I agree with your assessment, however I believe optical sensing at it's heart is fundamentally flawed. I found the thresholding & sampling required to accurately trip a sensor (even eliminating ambient light) is very difficult to do reliably without having to tweak the A/D threshold for every filament color. I think this is evident in Prusa's pivot to an inductive sensor in the MM 2.0 with a ball bearing displaced by the filament.
With 10's of thousands of printers out in the wild it would be impossible for Prusa to admit the design flaw and issue a warranty repair without bankrupting his company, so here we are with a firmware band-aid that appears to have some warts.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
Hey guys, just my 2 cents.
I run 3.4.0 since release and i got absolutely no problems with it. Filamentsensor is always on. Mostly small jobs, but also some prints >6h with no problems too. Everything is printed directly from SD.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
Hi all,
I like the new FW for the most part but the Filament Sensor IS a problem for me. (never was before)
Sometimes the print even stopped TWICE at one (quite small) layer.
I'm also pretty sure i hear the extruder skip occasionally which i almost never had before, even loosend the
screws, lubed the gears but it did not help.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
With 10's of thousands of printers out in the wild it would be impossible for Prusa to admit the design flaw and issue a warranty repair without bankrupting his company, so here we are with a firmware band-aid that appears to have some warts.
What people like me need is a solution that works, ultimately. I get a sickening sinking feeling every time I read a reddit post about how someone got Prusa to send them a bunch of replacement parts, only to discover later that it was their own error and nothing was needing to be replaced. That doesn't even consider all the dishonest folks who learn about Prusa's warranty hand-outs and socially engineer the support staff to send free items. My one single chat with support afforded me more than one opportunity to do so *EASILY* (which I did not).
The tragedy here is that Prusa chose to use this feature to move boxes off the shelf. I've heard that the new MMU sensor adds a step where it watches a steel ball being affected by filament, rather than attempting to sense the filament itself.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
I had the same issue with the 3.4.0 firmware as others .
I have started a litophane that should last for 10 hours printing. Last night, after 7 hours of printing , the printer stoped like it was a problem with the filament. There was no problem but I took out the filament , place it again and from the menu tried to restart the print. Something on the z axis were wrong so the nozzle was pushed deep inside the print, as can be seen in the print, and after the nozzle get out from the print it tried to print again but was stopped by the edge of the print. I had to stop the print. So with the new firmware I had two issues :
1. False indication of fillament fault. After reinserting the filament I got a warning that the signal from the sensor was week and if I want to disconnect the sensor. I never had a false fault form the filament sensor before , with older firmware .
2. When restarting the z axis value was wrong so the print was destroyed
A picture of destroyed print is on the Prusa user group on Facebook :
https://www.facebook.com/photo.php?fbid=2460956190598287&set=gm.1039022909615523&type=3&ifg=1
An advice from more experienced users ? Should I put an older firmware ?
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
An advice from more experienced users ? Should I put an older firmware ?
Pre-releases are not intended for actual use. They are to test changes. The github discussions around these last two pre-releases indicate that Prusa acknowledges that these firmwares are messed up and are presently narrowing down on these new troubles some of us are experiencing. SO basically if you want to be a part of that process keep right on using pre-releases but do make sure to report (or confirm other reports) your troubles at GITHUB. Here appears to be more of a community discussion forum and hangout.
Otherwise revert to the most recent final release 331 and enjoy a working printer!
EDIT: I see that Prusa has actually called this 34 junk 'final'. That was a mistake, although putting the cart before the horse appears to be the nature of this bleeding edge tech.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
I also loaded 3.4.0 a day or so ago. Everything appeared to be fine and I had no issues printing.
This morning I tried to change filament from ColorFABB black to white. It unloaded the black no problem but will not auto load the white. This has worked flawlessly with this filament until now. I presume the sensor is not now seeing the filament 🙁
I guess I will be reverting to 3.3.1 this evening after work......
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
As you discovered 3.4.0 is not labeled as testing or beta, is labeled as final. Maybe as in apple IOS version the x.x.0 is still for wide area testing over customers 😉
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
I'm happily back on 3.3.1 after around a week on 3.4.0 and am printing without issue again. Out of curiosity, a question for those that are using 3.4.0 and liking it... What improvement over 3.3.1 are you seeing? Maybe I'm a little coarse in my knowledge but other than being able to turn the sound off I saw no real improvements while using it.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
Maybe it is placebo but I think the stealth mode is a lot louder (and faster movements, especially Y-Axis which is loud) than on 3.3.1... Did anyone else notice this?
According to the change logs:
Just reverted back to 3.3.1, damn the stealth mode is actually usable again... I am allowed to print during the night again 😀 😀 :D.
The only thing I am missing are the sound options 😆
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
I've heard that the new MMU sensor adds a step where it watches a steel ball being affected by filament, rather than attempting to sense the filament itself.
The MMU is a completely different sensor. It has a steel ball that's lifted by the filament and sensed by an inductive probe.
It's not as dependent on the specific filament as the optical sensor, but it's not perfect either. It can only sense the presence of filament and not whether the filament is moving, so it can't sense jams. It also relies on gravity to drop the ball when the filament is gone, so debris in the filament path can cause incorrect sensing, and you can't mount the MMU in some other orientation (i.e. on the side of the printer). It also won't work in zero-G, in case that's important to you.
IMHO, the best filament sensor I've seen is the scroll wheel on the Palette+. It senses motion of any sort of filament and is extremely accurate. I don't know about cost, however.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
IMHO, the best filament sensor I've seen is the scroll wheel on the Palette+. It senses motion of any sort of filament and is extremely accurate. I don't know about cost, however.
Interesting!
Personally I would tend to suspect that a simple mechanical 'ticker' type sensor attached to the passive bondtech gear somehow might be workable, activated only when that gear gets pushed out from there being filament present. I am well familiar with the potential failure of that trap door that holds the passive gear, as I spent at least a week fighting the settings of my MK3 all the while not realizing the passive gear holding shaft had taking advantage of the free space it had to slide out of its hole and release one side of the gear, requiring a hot glue kludging.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
I am rolling back to 3.3.1. The filament sensor is no better - it just now tells me when it sucks instead of just sucking.
A HUGE issue I am having is when I turn on the filament sensor after the system turned it off (when I said ok) due to the sensor reading being low. The nozzle moves to the center of the print surface and drives down a few dozen mm towards it. If the head is low to start with, it plunges it full speed into the print surface and then tries to drive THROUGH the board. I have to emergency reset to stop it.
That has happened 3 times to me now, and I consider that a major problem. Back to 3.3.1 I go.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
If you pause a print the hot end temp never come back, bed temp does hot does not, also hot end temp does not come back up if print recovers from power outage.
Pete.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
Built MK3 kit that arrived with 3.3.1. Selftest completed ok. Finished Calibration Wizard. Flashed 3.4.0. Was repositioning PINDA, so assumed many errors from that. Running selftest Y-Axis length error. Confirmed no obstructions. Adjusted one loud bearing. Same error. Flashed back to 3.3.1. Self test completes successfully. Moving on to get clean first layer print. Something not quite right about 3.4.0?
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
Something not quite right about 3.4.0?
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
Reverted to 3.3.1 last night and my sensor issues went away. I can now work with the ColorFABB white again!
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
Something not quite right about 3.4.0?
issues.jpg
<shrug>
Any yahoo can open an issue on Github. A lot of the "open" issues are duplicates, feature requests, not firmware related, actually fixed in some other issue, or edge cases.
The real problem is that nobody's cleaning up and prioritizing the open issues, which makes it hard to see what actually needs work. But such is the curse of open source development: everyone wants to code, and nobody wants to work on project management.
Re: Firmware 3.4.0 for Original Prusa i3 MK3 (final release)
Just thought I;d throw it out that I am also a user having issues with 3.4.0 that didn't exist in prior FW. There are really 3 things that I have noticed but I have only been using it for a few days (~10 small prints).
1. Print stops, printer alarm sounds and the screen says "Please press the knob to unload filament".
2. Print stops, printer alarm sounds and screen says something like "Press knob to heat nozzle". After I do that and the nozzle is heated I get the "press knob to unload filament" again. Once I do that, I then reinsert the filament and it returns to the print where it stopped but just sits there and cooks filament and ruins the print.
3. I have noticed the SD card menu is sluggish. Not a huge problem but there is definitely lag when scrolling and also after I push the knob.
I will go back one FW version for the time being.
Otherwise, I like the IDEA of a lot of these features such as the volume adjust and Time to End display.