1. Got a question or need help troubleshooting? Post to the troubleshooting forum or Search the forums!

Latest Update and G36

Discussion in 'Software' started by SteveDjarrell, Mar 14, 2018.

  1. SteveDjarrell

    SteveDjarrell Member

    Joined:
    Jan 9, 2015
    Messages:
    98
    Likes Received:
    38
    Hi all, I had issues with the last update. I got to a point I can use the machine but I really want to like the Bi-Linear Bed Leveling. So I set my Z offset, went and set the probe offset, entered it M265 Z-.3 then ran the M500. I check on the Eprom and it has the correct values. So when the Bed leveling routine runs It appears the Center 3 on the Y axis are slightly off. If I print everything looks on the outer edges, however the center appears to print much higher. I was dumbfounded for about 5 hrs but I am wondering if the G36 is my issue? It runs G35 which adjusts for Ambient light. The room is darker and in the center I am just wondering if the probe is triggering sooner? Will try later today but think I can replace the G36 with just a G29?
     
  2. WheresWaldo

    WheresWaldo Volunteer ( ͠° ͟ʖ ͡°)
    Staff Member

    Joined:
    Feb 18, 2015
    Messages:
    5,806
    Likes Received:
    3,521
    I have my own opinion about G35/G36 which I have mentioned on other threads so I won't repeat them here. If you are going to play with firmware at the very least try out the 1.1.8-dev branch as Robo has made all the same changes to that version as they did to 1.1.6-dev.

    By the way there is no M265 so did you use M565 which has been deprecated and does not work in versions of Marlin newer than 1.1.0? Probe offset uses M851 and Home offset uses M206.
     
    #2 WheresWaldo, Mar 14, 2018
    Last edited: Mar 14, 2018
    supercazzola and mark tomlinson like this.
  3. SteveDjarrell

    SteveDjarrell Member

    Joined:
    Jan 9, 2015
    Messages:
    98
    Likes Received:
    38
    Will definitely give the 1.1.8-dev branch a look. I was however talking about simply in my startup Gcode. Remove the G36 and put in a G29. Also Sorry I need more coffee today. I did mean M206, not M265... Not sure where I pulled that out.
     
  4. Grim

    Grim New Member

    Joined:
    Nov 9, 2017
    Messages:
    11
    Likes Received:
    2
    Try changing the probe points to four instead of three in Marlin. That seems to work well for the people who are having the problem you are describing. It changes up the probe points and defines the bed a little more.

    G36 is doing the extra probe on the bed at the beginning of the print. It's setting the M851 offset based upon the trigger distance of the IR sensor. If you want to see what it's actually doing, set a zoffset with M206 then do a G28 G29 and then do a G1 Z0 (or whatever you feel comfortable going to) you'll notice that the printer will air print. Then run G28, G35, and G29 then go to Z0 you'll see the nozzle actually reach the bed. Robo didn't want to set a static M851 because the correct M851 offset can change with the time of day due to the IR sensor being sensitive to the changing light. G35 just automates the M851 offset collection process to set your nozzle to where it was placed during the zoffset wizard.

    If you don't want to use G36, you just have to set an M851 offset before each print.
     
  5. WheresWaldo

    WheresWaldo Volunteer ( ͠° ͟ʖ ͡°)
    Staff Member

    Joined:
    Feb 18, 2015
    Messages:
    5,806
    Likes Received:
    3,521
    @Grim, your description is not entirely correct. While Robo wants you to think that the IR sensor is so sensitive to daylight or room lighting that the Probe Offset needs adjusting every time, that is simply not the case, You can prove this to yourself by issuing a M48 Probe Accuracy Test repeating it over the course of say 20 or 30 iterations, You will find that even though the lighting won't change in the few minutes you perform the test, there will be a variation in the probe reading. For the most part the probe reads close to the same almost every time, regardless the time of day. There is some sensitivity to light outside the IR range since the sensors are not really a notch filter. But you can probe first thing in the morning and probe again late in the afternoon and there may be no change in those particular probe measurement. Besides this, you can have a value of 0 in M851 if you manually adjust the home offset M206 and it won't print in the air.

    You can love it or hate it but the G35/G36 combo was code that need not have been included in their firmware, my personal opinion. It would have been nicer had they spent time fine-tuning and adding more features to RoboOS or at the very least cleaning up the mess of an image on the µSD card, again personal opinion.
     
    supercazzola likes this.
  6. Grim

    Grim New Member

    Joined:
    Nov 9, 2017
    Messages:
    11
    Likes Received:
    2
    @WheresWaldo I disagree. One really cool problem G36 solves is people changing the color of their bed. You mentioned doing 20 or 30 iterations of the probe accuracy test. Try doing the same test with a blue bed, or a white bed. Instead of running M48, run a G30 (http://marlinfw.org/docs/gcode/G030.html) . This command probes the bed at a given point, then outputs the z position the probe triggered at. The accuracy is the same, however the actual trigger distance is not the same between the beds. So if you change your bed color suddenly, your offset does not remain accurate. It may only be off by a little bit, but it is still enough to ruin your new bed or it prevents the nozzle from connecting with the bed. G36 Measures that bed color change by measuring the trigger distance of the probe, then updates the M851 offset and maintains a great first layer.

    TL;DR There is nothing wrong with the IR sensors accuracy, but there is a discrepancy with the trigger distance based off of available light and the color of the bed.
     
  7. WheresWaldo

    WheresWaldo Volunteer ( ͠° ͟ʖ ͡°)
    Staff Member

    Joined:
    Feb 18, 2015
    Messages:
    5,806
    Likes Received:
    3,521
    Not many people have multiple beds, nor does Robo have stock for any significant number of users to purchase an extra bed.

    Changing bed colors is a rather extreme example.

    You missed my whole point, If you believe the probe always reads with 100% accuracy then G36 would make a little sense. But you can prove to yourself that there is a range of readings that happens even within the timeframe of seconds. In our case Accuracy = Repeatability.

    I had a much longer technically detailed message written but decided that simple was better.
     
  8. Grim

    Grim New Member

    Joined:
    Nov 9, 2017
    Messages:
    11
    Likes Received:
    2
    People change from the black buildtak bed, to painters tape, to offbrand buildtak, to anything an everything. By "New Bed" I did not mean an entirely new bed, just a new surface. It is definitely not an extreme example.
     
  9. supercazzola

    supercazzola Active Member

    Joined:
    Jun 1, 2017
    Messages:
    369
    Likes Received:
    97
    You guys are lucky. Mine crashes the bed into the head. Now I have lots of little ding marks on about 4 of the 9 probe points


    Sent from my iPhone using Tapatalk Pro
     
  10. Jeff Lastofka

    Jeff Lastofka Member

    Joined:
    Feb 21, 2016
    Messages:
    62
    Likes Received:
    25
    I was getting those "Probe Failures" and dings, too. I've discovered that if I whisk off the sensors with a soft paintbrush before every print, I don't get the crashes. I'm guessing something's getting on the sensors from static cling...

    I'd like feedback from someone else who tries this brush sensor cleaning idea to report back on success/failure. It sure seems to be necessary and working for me.
     
  11. supercazzola

    supercazzola Active Member

    Joined:
    Jun 1, 2017
    Messages:
    369
    Likes Received:
    97
    I have been using compressed air between every print and still had the same problem. I truly believe mine has to do with something related to light intensity and the surface of the Buildtak



    Sent from my iPhone using Tapatalk Pro
     
  12. WheresWaldo

    WheresWaldo Volunteer ( ͠° ͟ʖ ͡°)
    Staff Member

    Joined:
    Feb 18, 2015
    Messages:
    5,806
    Likes Received:
    3,521
    If you have to interact with the sensor before every print, it kind of defeats the desired intent for Robo to make this an appliance. Just an opinion.
     
    Geof likes this.
  13. Geof

    Geof Volunteer Moderator
    Staff Member

    Joined:
    Nov 9, 2015
    Messages:
    6,561
    Likes Received:
    2,264
    I havent witnessed having to interact with the IR on either the C2 or the R2...I will say if you run a bit hot you get the same white residue (get it on all machines) and if that gets on the IR you really should wipe it off (like I said- no issue yet).

    Now I can routinely crash the bed into the nozzle, but typically its some wierd junk in the octoprint upload- I cannot repeat it using S3D and a USB stick.
     
    mark tomlinson likes this.
  14. Jeff Lastofka

    Jeff Lastofka Member

    Joined:
    Feb 21, 2016
    Messages:
    62
    Likes Received:
    25
    I've only had the printer a couple weeks and have only run 15 print jobs, all PLA. I was having a lot of Probe Failure crashes until I started brushing off the sensors and that seems to have fixed it. I'll need to keep this up a while to be sure. Perhaps my sensors are weak or mis-aimed and therefore more sensitive than normal. Perhaps I should request a new sensor assembly. I'm waiting a bit to be sure this is the problem. I'm near Las Vegas and humidifying the house to around 45% so static doesn't seem to be a real problem, although there's a tendency for black plastic surfaces around here to show a light white frost of some kind of micro-dust. Especially if I dust using a Swiffer cloth (uses static charge) instead of a damp generic cloth.
    If you think it's a lighting variation problem, how about starting the prints in a darkened room, at least while the bed probing sequence is running? That would be a good test for a while.
     
  15. Geof

    Geof Volunteer Moderator
    Staff Member

    Joined:
    Nov 9, 2015
    Messages:
    6,561
    Likes Received:
    2,264
    @WheresWaldo can comment on lights on/off/variations in the lighting and the G36. I can say I tested it and got the worst first layer I've ever seen. I will not use that again. It will be a G28 and G29. I care not if the lights are on or off :D
     
  16. WheresWaldo

    WheresWaldo Volunteer ( ͠° ͟ʖ ͡°)
    Staff Member

    Joined:
    Feb 18, 2015
    Messages:
    5,806
    Likes Received:
    3,521
    There is no more variation between a room with lights on or off than there is between normal probes, at least in my experience. That is one reason why I still believe that all the time put into the code for 'ambient' light variations was a complete waste of time. The code assumes that the probe is 100% accurate 100% of the time and it simply isn't. Stupid, stupid, stupid.

    One day I might open up and tell people how I really feel about this code.
     
    supercazzola likes this.
  17. mark tomlinson

    mark tomlinson ༼ つ ◕_ ◕ ༽つ
    Staff Member

    Joined:
    Feb 21, 2013
    Messages:
    21,138
    Likes Received:
    6,997
    I would not expect ambient lighting in the norm to affect UV light sensors regardless. Actual sunlight, yes, possibly, indoor lighting? Nah.
     
    Geof likes this.
  18. supercazzola

    supercazzola Active Member

    Joined:
    Jun 1, 2017
    Messages:
    369
    Likes Received:
    97
    I wonder if there is some UV component to the white LEDs I have in my case (kilrah’s mod)? I know that white LEDs are really not white. They are broad spectrum. I think mine are cool white, which are really blue/UV diodes with yellow phosphor...

    Maybe there is some component of the LEDs causing issues? Maybe...


    Sent from my iPad using Tapatalk Pro
     
  19. WheresWaldo

    WheresWaldo Volunteer ( ͠° ͟ʖ ͡°)
    Staff Member

    Joined:
    Feb 18, 2015
    Messages:
    5,806
    Likes Received:
    3,521
    @supercazzola if that is the case, you would expect it to still be consistent if the LEDs are on every time you level, problem solved. Still no need for G35/G36.
     
    supercazzola likes this.
  20. JeffreyB

    JeffreyB Member

    Joined:
    Mar 25, 2016
    Messages:
    68
    Likes Received:
    29
    I thought the sensor was IR not UV??
     

Share This Page