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

Trying to resurrect my R2

Discussion in 'Mods and Upgrades' started by Joseph Stevens, Jul 6, 2021.

  1. mark tomlinson

    mark tomlinson ༼ つ ◕_ ◕ ༽つ
    Staff Member

    Joined:
    Feb 21, 2013
    Messages:
    23,912
    Likes Received:
    7,338
    Well I did use a Noctua fan :) Those are really quiet.
     
    Jerome Helbert likes this.
  2. Jerome Helbert

    Joined:
    Aug 19, 2017
    Messages:
    90
    Likes Received:
    33
    Yeah, one downside of the fully 24V system is it’s hard to find super quiet fans.
     
  3. Jerome Helbert

    Joined:
    Aug 19, 2017
    Messages:
    90
    Likes Received:
    33
    Back from vacation, and have only had a bit of time to tinker with this. I finalized my Precision Piezo Orion mount (replaces the aluminum hotend clamp on the stock R2) and printed that and a lower mount for the mosquito (the Orion PCB sandwiches between these two pieces) and got everything swapped over.

    The Orion definitely takes some tuning, between adjusting the sensitivity (potentiometer on the board itself), adjust the preload on the screw that hole it all together (there are some community designs that use o-rings to make this easier, I am going to try that next after I get an order of some in), and the homing speed, I have gotten it to home and generate a bed mesh pretty reliably.

    I have to hold perfectly still while its running since it's currently so sensitive that bumping the desk or even a particularly loud noise can trigger it, but I am able to get repeatable results to < 5 micron accuracy.

    I want to redesign my BMG-M mount to eliminate features that were only there when using the uptown/downtown boards as well as improve cable management, but then I will upload my designs to thingiverse if anyone else is interested.

    I have already decided to replace the SKR 2 with a Bigtreetech Octopus, it's massively overkill for the steppers (supports 8, I only need 4) but the allure of 6 PWM outputs and the ability to have them either 5V, 12V, or 24V (potentially 9 if I decide to repurpose 3 of the 4 heater outputs as 24V only PWM outputs) is just too enticing. Now that I have been able to control some fans/devices I want to control everything :) The octopus has enough inputs and outputs that I should be able plenty of room to do whatever I want. It's also only a few dollars more than the SKR 2, so there isn't much to lose in using that instead.

    When I get around to swapping the mainboards and adding the exhaust ports on the bottom, I am also going to relocate the Raspberry Pi. I think I'm going to put taller standoffs in under the power supply and mount the Pi under there. I'm planning to put the inlet fan under there as well so there should be plenty of airflow for it. This will give me room to use the z-stepper brace as is.
     
    mark tomlinson likes this.
  4. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230

    All of the stuff I put on thingiverse for the Robo can not be found. They removed all of the Robo tags. Can you put it on google drive and post the links
     
  5. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230
    The biggest factor in quality may not be the electronics. I just rebuilt an R2 carriage and the quality of print went way up. The bed moved easily as I did not over-tighten the belts. The print head had some play but not very much. Thinking that quality was linked to the tighter head I went to my R2 clone test machine. The X and Y.are direct driven by 48mm long steppers so they have lots of drive and move easily. The quality on this machine was very good even though there was a fair amount of play in the print head. I decided to remove the play by using polymer bearings in place of the LM6LUU. This gave me a print head with zero play but it no longer move very freely.. The result is circles that look like squares with rounded corners. If the stepper does not have enough torque the micro-steps will eventually reach enough torque to move the stepper, so I am not getting the advantage of micro-stepping..

    I do not know how freely the head has to move to take advantage of the full micro-stepping. But the mechanical seems very important

    Update: I pulled the polymer LM6LUU's out and replaced them with THK LM6LUU's that I purchased from Partsbuilt. These are much better bearings. I am used to slop in the print head. I now have no slop the print head moves freely and circles are now circles again. I think this is a necessary upgrade from the Robo bearings.
     
    #25 Lance Weston, Aug 1, 2021
    Last edited: Aug 2, 2021
  6. Jerome Helbert

    Joined:
    Aug 19, 2017
    Messages:
    90
    Likes Received:
    33
    Oh I definitely agree, looking over your bearing block designs, it's obvious that the auto-tensioning spring on the belts is going to let the carriage... spring :p

    Right now I am working on a redesigned cooling system that uses two 4510 delta blowers (more cfm, higher static pressure, and most importantly - orders of magnitude quieter than the stock 3010 axial fans), but after that I am definitely going to be doing these upgrades. But, like I had hinted at previously, I think I am going to move towards an ultimaker style of carriage using a single 8mm rod in each axis to hold up the extruder/hotend assembly. I haven't decided what bearing/bushings to use yet thought. I'm super tempted to move to a sintered bushing or graphite infused bushing and get away from lubrication, but I'm not sure how they will work in a non-rotating assembly. At the same time I am hoping the single 8mm rods give me more room to bring the extruder and hotend closer together and in general reduce the moving mass of the printer.

    I'm almost done with v2 of my refit (v3?) and it's even better than I had hoped. The BTT Octopus offers the ability to select voltage for each fan port (5V, 12V, or 24V) so I've decided to to take @mark tomlinson advice and drop noctua fans in :) when the part cooling fans aren't in the whole thing is dead silent now (replaced the 25mm screamer hotend fan with a 4020.) When I have the kinks ironed out I will put together a post with the details.

    Edit:
    Forgot to mention that I'm using true PWM fans everywhere now as well. When we apply a PWM signal to a standard 2-wire fan it technically works, but not great and it's also hard on the fan itself. On 4-wire fans they have the same positive and ground, but they also have a dedicated 5V PWM control and Tach signals. For the moment I have set my fan ports all to 5V and hook that up to the PWM control signal for each fan. I then feed a constant voltage (12V for all the noctua fans) and the electronics inside the fan do all the speed control internally. The end result is a fan that has far better PWM control, is quieter, and will last significantly longer.

    Edit2:
    I also designed my own piezo/mosquito mount and have really been able to dial this thing in. Here are two repeatability measurements, one for the middle of the bed and one for the front edge (hardest part to probe sense the bed is easier to flex here)
    Middle of the bed:
    Code:
    // probe accuracy results: maximum -0.066250, minimum -0.068750, range 0.002500, average -0.067375, median -0.067500, standard deviation 0.000673
    Front of the bed:
    Code:
    // probe accuracy results: maximum 0.066875, minimum 0.066250, range 0.000625, average 0.066562, median 0.066562, standard deviation 0.000312
    The engineer in me knows that the printer needs nothing close to this, but damn is it cool to have sub-micron accuracy :)
     
    #26 Jerome Helbert, Aug 5, 2021
    Last edited: Aug 5, 2021
    mark tomlinson likes this.
  7. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230

    The problem with going to an 8mm shaft instead of 2 6mm is that the head must have no play If you use a quality slide bearing you can not make it fit within the 20 tooth GT2 gear diameter.. I spent a lot of time modeling this and could not get it to fit. A sleeve bearing is not tight enough, and I found that the graphite bushings have too much resistance. If you got an undersized sleeve bearing and used an adjustable reamer to bring it out to dead nuts perfect and then applied axle grease ( the slipperiest stuff I have found) you might stand a chance.
     
  8. Jerome Helbert

    Joined:
    Aug 19, 2017
    Messages:
    90
    Likes Received:
    33
    Are you talking about in the bearing block or in the carriage itself? I'm drawing a blank on where the GT2 gear diameter comes into play.
     
  9. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230
    The belt wraps around the 20 tooth gear determining how much plastic there is in the bearing block.. Increasing from a 6mm shaft to an 8mm shaft means much less plastic on the bearing block. It also means much larger bearing diameter in the print head, for which there is not enough room. The only thing that could clear is a custom sleeve bearing in the print head honed to remove any play. I just posted my latest bearing blocks in "carriage update".
     
  10. Jerome Helbert

    Joined:
    Aug 19, 2017
    Messages:
    90
    Likes Received:
    33
    Ok, I get what you're saying. For the carriage block I was going to use a variant of this design: https://www.thingiverse.com/thing:45236 but modified to accept an 8mm rod. This would use 8mm rods that are short enough to fit between the rails instead of over/under and I believe that would alleviate the belt spacing issue.

    I still think I can get away from linear bearings in the printhead, possibly using a pair of bushings on each axis at each end of the printhead that for accuracy/play would make it behave like one long bushing. If I can't make bushings work then an LM8UU should still be able to work, you just don't position them directly over each other...

    Like I said, I am still in the early stages of planning this out but I think it should be possible - whether or not it will actually be better is a whole different question... :D
     
  11. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230
    You must keep the existing 8mm rods in their current location, which precludes that thingiverse design. In my initial improvements I used bushings at each end of the print head. I use 24 hour epoxy and floated them in the assembled unit so when the epoxy dried I had well aligned bearings. It simply does not work as well as a high quality bearing. It also failed in 60% of the print heads I made.

    The designers at robo were clever in the design of the bushing blocks. The 6mm rods are held as points at each end. My current bushing blocks do that. If you try and clamp a rod through a hole of any length you will aim it, and it is impossible to aim it accurately. It is not obvious but my new bushing blocks are tapered to a small clamping spot. The clamp itself floats so as not to change the distance dimensions.

    The moment around two 6mm rods is stiffer than one 8mm rod. To improve it you would need a 10mm rod and I could not work out a single 10 mm rod either.

    I have been at this for years with dozens of designs and the problems are like an onion..I do not think you can get any better results than upgraded bearings from Partsbuilt will give you. My bushing blocks do not give better results than pristine robo bushing blocks. My blocks will just not break, so continue to give good results.

    I think you need a good 3d modeling program. Any dimensions you need I can supply you.
     
  12. Jerome Helbert

    Joined:
    Aug 19, 2017
    Messages:
    90
    Likes Received:
    33
    I'm still not sure about some of your statements... I think there is plenty of room for a single 8mm rod and bearing (I have given up the idea of using bushings and will likely stick with an LM8UU) without moving the gantry rods at all

    Also, two 6mm rods are not stiffer than a single 8mm rod. The moment of inertia of the 8mm rod is 3x larger than that of a 6mm rod, and if you assume the load if split evenly between the two 6mm rods, that means that under the same load conditions the two 6mm rods will deflect 50% more than the single 8mm rod.

    But with this design there wont be the same loads because the mass of the rods themselves goes down and the whole carriage should end up smaller and considerably lighter.

    As I said before, it might not work out. But right now I think there are enough pros to the idea to try and hash it out. Worst case I am out the cost out a couple bearings and I switch back to the 6mm rods and stock r2 carriage.
     
  13. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230

    Yup, you are right. just did calculations and it is a little more than 50%. The distance between vertical centers of the 6 mm rods is 16.8 mm. An LM8LUU bearing has an OD of 15mm. A pretty tight design problem. The GT2 gear has an OD of 16mm less the thickness of the belts, about 12mm. That leaves 2mm of plastic left with an 8mm rod to support the weight of the head, This again is pretty tight and I am not sure it will be strong enough.
     
  14. Jerome Helbert

    Joined:
    Aug 19, 2017
    Messages:
    90
    Likes Received:
    33
    I think I understand my confusion now, you are talking about the carriage for one issue, but then the bearing holders when talking about the GT2 gear.

    For the LM8UU clearance, I'm not planning to place them directly over each other, and instead shift them to the side of the crossing point of the linear rods. Here is a crude paint mockup illustrating my idea
    [​IMG]

    As for the belt clearances to the rod, if you look closely at the thingiverse model for the ultimaker I linked to, the carriage rods don't go above or below the gantry rods, they are shorter and fit inside the plane of the gantry rods. So there should be no worries about clearances there.
     
  15. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230
    The bearings should work that way, but they will be short. Ideally you want a print head with no play or it will vibrates as it is drawn over the print. The longer the high quality bearing the better.. That clamp should work with just a loss of vertical height.

    Once you have decided to go this route, and it is interesting. The 16.8mm restriction does not apply and you can space the 8mm bearings vertical distance any way you want. An LM8LUU is 45mm long and could be used one over the other.

    If you would like to collaborate on the design, I am interested. I would change the closed loop belt design to open loop. Other than that it is just a matter of drawing it up.
     
  16. Jerome Helbert

    Joined:
    Aug 19, 2017
    Messages:
    90
    Likes Received:
    33
    Yeah, I had been looking at the LM8LUU for that exact reason. I figure the print head has the space for it, so why not?

    Where would the loss in vertical height come from? If there is any, it will likely be offset by my plan to move my extruder and hotend closer together. I had actually been concerned that I would have to limit how high up I could move the hot end before I would run out of Z travel (Although with the Piezo board in there I have actually lost ~10mm of z height, so it's probably not as much of an issue.

    The bearing holders ought to be common, but the print head / carriage I'm planning will likely be pretty unique since I have already replaced the extruder, hot end, Z height sensor, and eliminated both uptown and downtown boards. But we can probably generate a model with the bearings in all the right places and then adapt it as needed for whatever combination is being run.

    I mentioned earlier that I had thought about designing something would would be compatible with the EVA carriage system, then people could use any combination of extruder and hotend they would like. This platform is geared towards MGN linear rail style printers, but I was thinking it might be possible to squeeze the X-axis linear rod bearing into the area that the platform has for the linear rail. I've not actually done the maths on it to see if it's even possible, and those style printers don't have a y-axis rail to worry about, so we'd have to make sure it wouldn't interfere. It doesn't support that Piezo sensor I just finished integrating, so I had started to back burner that idea

    I'd definitely be interested in some collaboration, unfortunately with summer ending the family is all going back to school/work and my level of free time to devote to stuff like this is going to decrease significantly. I also need to get the printer into a fully functional state and try to keep it there for the most part so it's ready to go whenever the wife needs stuff printed for her teaching. So my ability to actively work this will be limited. I was planning a slow behind-the-scenes development and probably wouldn't be able to cut in and test out any design until next summer.
     
  17. Jerome Helbert

    Joined:
    Aug 19, 2017
    Messages:
    90
    Likes Received:
    33
    Did a little digging quick, it's nice that they have all their models in onshape (here is one based around a BMG extruder) and can easily check things out. Looks like they have a 27mm deep opening and are ~35mm tall, so plenty of room to fit an LM8UU. The mount is also 44m wide, so an LM8LUU would probably fit in there as well. It looks like there is even plenty of room for the Y-Axis rod to pass next to the whole thing (With the new bushing block design the y-axis rod will now be above the x-axis rod, this is opposite of what it is today)

    The whole thing is pretty deep, but not very wide... So there might be a sacrifice of Y space in the R2

    I'm still not sure it fits my needs, but it is an interesting concept that would allow anyone with an R2 to easily (relatively) swap in any extruder/hotend/bed probe that is supported by the EVA.
     
  18. Jerome Helbert

    Joined:
    Aug 19, 2017
    Messages:
    90
    Likes Received:
    33
    Meh, I don't think this is an option for the R2, not without some significant rework... In a stock R2 the nozzle tip is ~56.5 mm from the top of the mounting/clamping plate, then there is addition distance between the plate and the axis of one of the carriage rods.

    It looks like in the EVA system the nozzle is only ~43mm from where the center of the y-axis rod might go. As is I think we would raise the bed as high as it can mechanically go, and it would still not go high enough to print on. So nevermind that whole idea.
     
  19. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230

    This can all be made to work with the R2. I use a dual gear extruder. The 8mm rods passing through the print head sit on the outside of the GT2 gears, so: above/below the the belts. This makes the print head longer with a lot of space. The bearing blocks can be modified to hit the x,y limit switches. It is fairly easy conceptually to design and print a print head with blocks to do this.. I use short wade E3D hotends. A shroud is an easy design. It would all fit on Robo aluminum plate.
     
  20. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230
    This is a quick mock up of what the rear bearing block would look like without it's cap. I left the 6mm rod holes in for a better illustration. They would not be in the final block.. This allows the belt to wrap around through the block. This is the rear block and has an extension to hit the x limit switch. It shows the 8mm rod now goes through the head lower, which means some loss of vertical capacity. Conversely the side blocks will. have the 8mm rod higher.. With the cap screwed on this will provide adequate strength. The print head becomes longer with plenty of space for the LM8LUU bearings.

    upload_2021-8-19_18-29-5.png
     
    #40 Lance Weston, Aug 19, 2021
    Last edited: Aug 19, 2021

Share This Page