Palette 2 - print failed due to blocked buffer

Comments

25 comments

  • Avatar
    Redemptioner1

    Whats you pongs look like as your pings are all over the shot even before you start hitting the buffer limits

  • Avatar
    Picaboubx

    I confess I did not pay attention to the pongs. I did not think it was important.

    If I stick to what I see accustomed, they always turn around 100% 101%

  • Avatar
    Jonny Yeu

    Hi Picaboubx, thanks for this information, we're currently looking into this and will respond once we have an answer for you. Thanks for your patience!

  • Avatar
    Redemptioner1

    Yeah, most don't because Mosaic always harps on about the Ping, but if the Pong is not extremely accurate then the Ping is useless as it is relying on the Pong to measure accurately for the Ping.

    The Pong realistically should not change more than 0.05%, in fact it should not change at all as the buffer size is physically constrained to ensure this. When you start getting incorrect colours on layers and the pings are resonable enough (say under 2.5% variance) I have found it is almost always the scroll wheel failing causing the Pong to start moving around (again I stress the fact the pong should NEVER change) which results in the transisition zone on the transition block slowly being consumed till you have a failed transition and the wrong colour showing up on the model.

     

    I have chased Mosaic a lot about this but they seem to try and point to anything but the scroll wheel, I now have 2 palette 2's that can't be used because of this problem.

  • Avatar
    Jonny Yeu

    Hi Picaboubx, could you please send your print files and octoprint.log files to examine (support@mosaicmfg.com)? If you do have pongs from this print, that would be helpful as well.

  • Avatar
    Picaboubx

    Good evening,

    Unfortunately I did not get the octoprint.log or the Pongs on those failed prints. I did not think it would be helpful.

    Since my basic post, I redid 1 test that worked (but also shows that I still have not managed to calibrate the Palette 2)

    I am trying again, I would have passed the elements in case of failure.

    Frankly I start to find Palette 2 unreliable and impossible to calibrate.

    Buy in February, I still have not managed to make a perfect impression, there is always a problem that appears.

  • Avatar
    Ed Williams

    Picaboubx,

    If you are still having issues I would suggest the following

    1. Check your extruder calibration
    2. Switch to accessory mode
    3. Switch to Chroma

    These steps should be done itteratavly. Dont try them all at once. When I run into an issue like the one you are having I find it best to tun through that exact workflow, it helps remove any questions as to the status of my printer and slicer settings.

    Once you have found the issue, work back in what ever piece was removed. Good luck and let us know how you fare.

  • Avatar
    Picaboubx

    Good evening Ed Willams,

    I have already done all the checks and adjustments many times since I have this material.
    I'm on Chroma since almost the beginning of my use of Palette 2.
    If I'm not mistaken I'm already using the access mode since I'm running the printer with octoprint.
    I also calibrated my extruder via this procedure: https://mattshub.com/2017/04/19/extruder-calibration/

    I also change the PTFE bowden tube for a Capricorn for more accuracy.

    My impressions are good qualities but there are always layer errors.

  • Avatar
    Ed Williams

    Good stuff. To check what mode you are in open up Chroma, select the printer profile, and select palette. It's either in Accessory mode ( acting independent of the pallete) or connected mode (octoprint controls the whole thing). What slicer and settings are you using?

  • Avatar
    Picaboubx

    Good finally I start again a calibration attempt.

    But it still goes that it works because I have trouble understanding why I have this kind of result that degrades as and impressions

    So I start from the beginning ... for the last time
    after I throw this machine that does not work with my printer

  • Avatar
    Ed Williams

    That's a wierd one. What do your pings look like?

    When I was first running my calibration I too ran across some issues. I would recommend resting the pallete to factory and removing any printer profiles. Then start with just 2 colors (key chain). Once that looks good move to your block design but continue to use 2 colors. Once you have a good result from that bring in the 3rd. Once 3 is good add #4.

     

    Also if you are using Chroma, look at setting the variable transition. From the looks of your blocks they all look to be the same length. 

    Good luck and keep us posted.

  • Avatar
    Redemptioner1

    Well on such a small print you are getting hardly any calibration adjustment from the pings anyways as you need 5m of filament to run a ping and adjust then another 5m to apply the adjustment and read another ping. 

    The degrading transition length suggest a pong issue rather than a ping or calibration, did you capture any of the pong data on these prints?.

     

  • Avatar
    Picaboubx

    Hi Redemptioner1

    Here are 2 Pongs screenshots of my last last impressions (successful on the model above).

    But there! I have the same kind of Pongs with the failed version of this same model. So I do not see how that can come from the Pongs.
    And besides, what are the causes of bad pongs?

    Capture 1 :

     

    Capture 2 :

  • Avatar
    Redemptioner1

    @Picaboubx

    As suspected, and completely missed by Mosaic Team yet again, the issue is most definitely being caused by your Pongs and is not a calibration issue or anything you can fix running splice tuning or changing print speed. Your pongs should not change over time and should be within  0.20% at the start of a print with less than 0.05% variance throughout the print after the 2nd Pong (should be less than 0.01% actually). 

    Note your pongs continuously increase, this is because the scroll wheel is miscounting the steps (over-reading in your case). The more colours and splices you have the more this will impact your print as the bad Pongs are eating into the transition zone on the transition tower. The Pong is used to say how long 100mm is compared to the position change in scroll wheel (i.e. 20 turns of scroll wheel is 100mm), when the Pong over-reads it is saying the filament is shorter (so 20 turns is 120mm). A small print with say 150 splices and 80 layers at 0.2mm and 2 colours is not going to eat into the transition zone enough to have the wrong colour being used in the wrong place (has to chew up approx. 40mm of filament first). This is because there is generally a 40% buffer between one colour and the next (the transition zone) and only doing a single colour transition every other layer means the small reduction (say 0.5% reduction) in the transition zone does not eat up the 40% "buffer". When you change this for a 4 colour model and say print 2 of them (same model as before) then you have 6 times the number of transitions which means you start eating up the transition buffer at 6 times the rate causing an incorrect colour on a section due to the transition happening too early. This compounds overtime as the previous incorrect Pong (over-read) is used in the calculation and error adjustment for the Ping (so it calculates the ping to be too long and shortens the filament on the next 5m of splices), this of course makes the transition happen even earlier, then the next Pong is out and it compounds the next Ping correction error even more. This is demonstrated by the fact you Pings end up all over the shop and out by several percent.  Now if the Pong was under-reading then the most likely culprit would be the lack of pre-load on the idler pulley opposite the scroll wheel (allowing the filament to slip past without turning the scroll wheel), but your Pongs are growing which means it is not a mechanical issue, rather one to do with error handling in the software, a failed scroll wheel (most likely cause) or electrical interference. 

    The fact that I have 2 like this and there are several others, who also print significant amounts of filament at reasonable speeds (your prusa running at 60mm/s is not a reasonable speed, this is SLOW) with the same issue, I would wager the issue is with a cheap nasty scroll wheel being used and the scroll wheel is failing prematurely on people who "try" to use the palette more than printing the occasional 100gm print. Not surprising when you consider Mosaic used cheap and super nasty motors & hobb gears (the sound these make are simply cringe worthy, let along the ridiculous amounts of EMF they create), a cheap Arduino for the brains (64bit quad core is only double the price but 50 times the power and reliability, when the CPU only costs $15 to begin with WTF)  and super crappy mechanical switches (we all know about buffer switch issues, and don't get me started on run out sensor switches), why should they spend any more money on the part that is REQUIRED to allow the product work...........maybe so it works????

    I would also like to point out again that Mosaic Support Team does NOT seem to understand how their own product works and continually demonstrates that they are unable to do some basic problem solving and work through a process of elimination, you know, simple service desk support. Clearly, if the customer has calibrated multiple times and having issues, then the first thing to look at is PONG every time, if the Pong is not super accurate then nothing else matters. In the case being discussed here the Pong increases and therefore the problem relates to the scroll wheel over-reading which completely eliminates any "slippage" on a mechanical level. If the Pong stayed constant after the 2nd Pong, then the issue would be calibration and would have a very limited effect on the print success (would not cause colours on wrong segments as the transition zone would not collapse). The fact it keeps changing, increasing in this case, the issue relates to the scroll wheel being faulty, in saying that all these failing scroll wheels seem to fail with the scroll wheel over-reading so there maybe something in that (don't have the time to pull the failed units apart to investigate Mosaics mess atm).

    There is obviously a problem with the scroll wheel as there is more than enough people now having this issue to say confirm a pretty major warranty fault. It seems if you print more than 30hrs a week since the Palette 2 came out you have hit this problem, a few seem to have hit it almost straight away which suggests the quality has also dropped off since initial production run ([pretty common practice for Asian manufactures). 

  • Avatar
    Jonny Yeu

    Thanks for sharing this information, Picaboubx. Contrary to Redpemtioner1's suggestion, pongs do change throughout a print and there can definitely be a variance larger than 0.2%. Pongs correct for differences between the scroll wheel and outgoing motor and checks the steps during a print with the steps that are calculated when calibrating Palette. This means that if there is a slight difference in the values, it would make sense that pongs would increase, but it's not an issue and is expected behaviour. 

    We can imagine that your steps saved to your unit is a line on a graph and a second line represents the difference in steps during a print. The space between the two are where pongs calculate the difference and account for this in filament production. If the pong is increasing, it means that Palette is accounting for these differences and adjusting the amount of filament that is being pushed to your printer.

    We appreciate your patience with this and want to get your unit up and running. Would you be free for a phone/video chat sometime in the next few days? It may be easier to work through these issues live rather than through forums posts.

  • Avatar
    Picaboubx

    Hello Jonny Yeu,

    I admit to being totally lost. I only have the test to use the Palette 2 in accessory mode. That's the only thing I have not done yet.

    It will be with pleasure that I will make a video chat but I do not speak English at all which may pose a big problem on such a technical subject. (I am French)

  • Avatar
    Redemptioner1

    Show me the maths where that works with a growing Pong Johnny....

     

    If the Pong changes after the second Pong, especially if it keeps growing or shrinking, then you have a problem. The pong reading is taken into the equation when measuring the Ping, if the Pong is not accurate then the Ping won't be accurate, the problem is the adjustments are made in the Pings PID loop (shortens/extends the next filament segment) which is why the transition zone shrinks when the Pong continually grows. The problem being discussed is where the scroll wheel is not correctly reading the output of the filament correctly, the actual filament length being produced and used is accurate, this is evident from the transition zone shrinking while the Pong increases and generally the Ping spends most of its time a few percent out, eventually you run out of transition zone and the colour transition starts taking place on the print and you start to get "stray" colours in the model.

     

     

  • Avatar
    Ed Williams

    Redpemtioner1,

    I get that you are smarter than everybody else and Mosaic should fire all of the engineers and bring you on staff to fix all the issues. Until that happens though can you please stop interjecting so much hate into the conversation. There are people who are just trying to work together to help each other out and your comments don't seem to help anybody.

    I do believe you possess some sort of higher level of knowledge, but I think that you forget that Mosaic is not RepRap, or Pursa, or Marlin. They are not open source and won’t share the type of data you are constantly asking for. They are a private company and are looking to make money. What you are asking for are thing things that make them competitive and unique in this market.

    I do agree that there are issues with the 2 (mostly the switches) and I do hope they come out with a drop in replacement for those. However, this is a new market and there is still a lot to be figured out.

    I know you don't like to hear the obligatory "it works for me", but there are quite a few people who have a fully functional configuration, and haven't had to go to the extents you have gone towards.

    Thank you.

  • Avatar
    Redemptioner1

    @ ED - I can only guess you work for mosaic or you got one for free or some other connection to them. You interject with this same sort of "give mosaic a break" comment regularly. You are correct, they are a private company that promised a product that could do XZY and took my money in good faith and have only managed to deliver a product that can only do half of X and no date on when the rest of X will be provided let alone Y & Z. If you make wild claims that cannot be backed up then you get called out for this, the fact is a PID loop is no proprietary code/algorithm nor is a dual feedback loop incorporating an encoder. The maths behind this is well known and is why Johnny was called out for "his maths" when he made the statement the palette 2 can compensate for an incorrect/error reading on a single encoder output (don't see the second encoder to allow this anywhere), unless Johnny knows how to bend the laws of physics......maybe he does? 

    When you say "this is a new market and there is still a lot to be figured out", that would be fine if I hadn't paid for the "retail" product but I did, the packaging didn't say "only some features work, but we hope to make the other works in the future" now did it. We did not agree to, or sign up to anything mentioning that we would be "testers" or having pre-trial products shipped, so why should we be treated like beta testers?

    You also need to stop with the "I get that you are smarter" bit you keep spouting, the only one bringing personal bs into the argument is yourself, unless you can converse on topic why don't you please just stay out of the conversation. Your "obligatory works for me" is just rubbish, you have failed to ever provide any evidence something working whenever you mention this. I am sure you would not be happy if you bought a new car and only some of the wheels turned would you?, would you accept the car dealer then telling you only 2 of the 4 wheels are meant to turn despite the fact that physically can't be true?? Would you accept then someone telling you that they have the same car with only 2 wheels that turn and it works fine, despite never providing any evidence to support the claim???......come on, this is a forum not a bookface like fest.....

     

    So back on topic, to sum up things to date;

    - Picaboubx still has a non working Palette 2 like several of us

    - There is an obvious problem with Pongs continuously growing (how do we know it's a problem, ahh maths')

    - Transition buffer slowly shrinks till transition takes places on part

    - Calibration does not fix the issues

    - Only solution provided has been to change the slicer and re-calibrate, both have not impact

    - Mosaic support believes there is nothing wrong with the Pong and it could not cause the issue being shown

     

     

     

  • Avatar
    Jonny Yeu

    Hi Picaboubx, please do test without a Hub to see if you notice any differences, this may provide us with additional information about why you're experiencing these issues. Although there is a language barrier, we can still attempt a video chat to see if we notice anything out of the ordinary that we haven't seen in the videos that you've shared.

    Ash, Manmeet has already spoken to you about the issues that you're experiencing and has already provided suggestions on how to fix your issue. We have many users who display the same pong pattern as Picaboubx and have found success. Our in-house units display the same pattern and we complete prints regularly.

  • Avatar
    Ed Williams

    @Picaboubx,

    I Apologize for turning your post into a war of words.

    I too had an issue early on that I was able to resolve with accessory mode. Then worked connected mode back in.

  • Avatar
    Picaboubx

    Hello everyone.

    Do not fight if you like, it does not matter. Thank you all for your advice and advice.

    I have been fighting with this material since the beginning of February and I only managed 2 impressions which were 2 airbud cases. It was several weeks ago. Since I repeated several calibrations of Palette 2, improvements of my printer (more precision gained), but in the end, I still have no correct result.

    The only solution I have not tested yet is the Accessory mode. Because I find it impractical.
    I will still test it this weekend if I have enough time.

    I'll let you know with screenshots and photos or videos.

    we will see if this is a problem with my version HUB (octoprint on Raspberry 3 + Plugin Palette 2) or if it is something else.

    Jonny Yeu:
    For a possible video communication, I am available every evening from 21:00 (French time) or Weekend on Saturday and Sunday afternoon or around 21:00 (always French time).

    Thank you for your patience.

  • Avatar
    Jonny Yeu

    Thanks, Picaboubx! Let us know how everything goes in Accessory mode and we can schedule some time next week if you'd like to video chat.

    Also, a similar issue occurred in Connected mode with another user. They then tried slowing down the feed rate during splices (wrench icon --> Palette 2 --> Enable advanced options) and the issue disappeared. It may be worth trying this option if you haven't done so already.

  • Avatar
    Redemptioner1

    You are correct, Mammeet has requested an additional video, of the same detail of the videos already sent only using my finger to rotate the scroll wheel instead, which I was happy to provide I just have not had the chance to setup for this video to send through (admittedly it has not been a priority either due to the video clearly not going to display anything different using my finger to turn the scroll wheel rather than the printer). Just so we are clear, Mammeet has been sent videos from me of the issue, so I have provided the evidence as requested, he has just requested some more footage but has not provided any “suggestions on how to fix your issue” at all, other than telling me to clean the scroll wheel with a cotton bud which I already mentioned I did this (along with blowing palette and scroll wheel out with compressed air and checking to make sure there was not stray bits of filament or other “gunk” that could be causing a problem).

    Now for your comment “Our in-house units display the same pattern and we complete prints regularly”, to date you have not been able to provide any evidence to any statement you make about your “in house units”, this is why it is raised so often. If you are being honest then it should be a no brainier to provide a video clearly supporting your statement and rather easy to do considering you are constantly mentioning all these in house test units setup and running. You certainly seem to have plenty of access to video equipment (and live streaming), so I think it is only fair you provide some evidence to support your claim here (a tripod and a $40 camera will work just fine), I have and so have many others in the past. It should be pretty easy to provide a video based on you “in house display units” having the “issue” and working fine considering they must be easy to access and already setup. Simply put a video camera recording one of your Bulbasaw models scaled up a little (say 15-20cm tall) with a 0.4mm nozzle and 0.2mm layer height and have the video show the printer creating the model alongside the Pongs & Ping data coming up on printer LCD with the steady Pong growth “issue” and no transition problems (we want to see the Pong & Ping data in the recording as the print is going). Even better, live stream it for us as you will be able to show us the transition tower clearly and the fact the transition buffer is staying constant despite a growing Pong and we won’t have to be concerned about some tricky video editing like we have seen in some of the sales videos. Just make sure you pick a printer we can easily see the LCD display on.

  • Avatar
    Jonny Yeu

    Hi Ash, if Manmeet requested additional information, it's likely that these details would help him determine what the issue is.

    We're currently using our camera equipment for other projects, but will ask our team to save any prints that share the same pong patterns as above. Once available, I'd be happy to share this with you.

Please sign in to leave a comment.