- User Since
- Aug 31 2016, 12:51 PM (67 w, 1 d)
Wed, Dec 13
@logan @MikeR I would like to try and adjust slice settings before adjusting the model. over extrusion will make the screw holes harder to line up as well as making the spool arm harder to put up when installed on a machine. @EricNugent can you try and tweek the slice settings on these parts to help with this issue? We can definitely adjust the model if need be but were trying to get away from fudging a parts dimensions for the sake of printing with ABS. We have a lot more options with Cura2 to help us get the parts as they are designed instead of messing with the models to get them to work with the limited slicing options in Cura.
Thu, Dec 7
@logan for sure man. yeah lets get new parts on them then
@logan The parts didn't break did they? I think we will be looking for molten/droopy parts after running in cluster, not cracked/broken like from a drop. I dont think we need to replace the parts.
Wed, Dec 6
@logan @karrad i think if testing goes well without filament then we will need to run it again with filament and see if that extra load on the motors has any adverse effects. This will add a bit of extra load to the extruder motor and x and y motors also. X because it will be pulling some filament off the reel sometimes as is makes a move toward x=0 and y because of the added weight of the part it's printing. but overall i think itd be good to test without filament before we spend the time updating parts
@logan @karrad I think we could run them without filament. Just have them running some g code to see how they hold up under normal use in the cluster but dont actually print anything. Hows that sound?
@kent okay i can get this modeled up and ill print a few out to test
Mon, Dec 4
@kent ive pushed a new x belt tensioner design (8817c8a8ad0a). This deflection still happens a bit but not as much as the old design. Let me know what you think. I will have cluster print some out for you to test
Fri, Dec 1
@kent ive got a new belt tension design that should correct this issue. Ive got a couple printing out. Also @Baldwin had another good idea for a tensioner. Once i install this design and verify it ill bring you one to try out. Im going to try making the one that @Baldwin suggested also.
@kent i agreee if we widen it it should only be to make it tighter. And thats not a bad idea either!
@kent yeah i have encountered this as well. Once your belt is tensioned though there should be no need to retension it since it cant really slip. This is more of a complication that could occur during assembly. We could try and open up the back side of the belt tensioner so that you can slip it back around the post, but then the posts that hold the tension arent going to be nearly as strong and could snap. Especially since the tension is held along layer lines. We can try a design of this style and see how it goes but i am worried about them breaking. Ill make up a few and we can give it a shot.
@kent the x belt spans about 300mm. so thats one leg of the right triangle. The deflection is about 1mm on either side of the belt so you can say the other leg of the triangle is 2mm long. This would give a hypotenuse of sqrt (300^2 +2^2) = 300.0066666 That means that there would be a loss of accuracy of 6.6666 microns over 300mm. We claim 100 microns of accuracy in x and y. This is not a big deal. Even if each side is deflected 2mm (which it's not) then thats sqrt(300^2+4^2) = 300.026666 which is a loss of 26.66 microns over 300mm. So i really dont think there should be any concern about it not being completely parallel to the axis. The only real issue then is that some people may think it looks bad. We dont really have any room on the back side of the x carriage to incorporate a clamp design like the y axis has. You just cant get a screwdriver in there. In order to squeeze more build volume out of the z we had to make the x gantry more compact. We can look into a redesign of the x belt tensioner to make it parallel but we wont get much reward from redesigning it. I dont see what you mean by side loading the bushings? which bushings?
Thu, Nov 30
After discussing with @MikeR we cant find the machine that exhibited this behavior. I believe that it had a cable chain that didn't have 18 links. Its possible an extra link was added to one that already had 18 links and that would explain why the chain hit the frame. Im closing this ticket for now. We will keep an eye out for this behavior and re open if need be
@kent we are moving to 100 steps per unit in X and Y as calculated so this is not going to be an issue anymore
@Baldwin i think that we need to have some profile testing with various fan speeds and see what the lowest speed we can run is before it impacts print quality. The slower speed should reduce the noise level, but it would be worthwhile to also get some sound level data on these printers with the fan running at various speeds and compare that to previous versions of the mini. @anolen would you be able to get some testing with various fan speeds on the aero toolheads? We can get you a fan shroud that will be on mini2 and install it on one your mini aero toolheads
@karrad I asked logan to drop test the PETg hibiscus since he has it. If it survives the drop do you want to put that printer in the cluster as is or do you want it to be rebuilt with completely up to date parts in PETg when we rebuild all of the alphas?
Hey @logan can you get this printer drop tested?
@logan i think we should keep x and y where they're at. There has only been discrete tensions for the y axis on gladiola with the "new" y belt mount/clamp style( with the teeth) and we haven't had issues with production falling in the 23-27N range. I do think the z axis needs to be a bit tighter than the x and y though. We could widen the tolerance for x and y if need be.
@logan I installed the new extruder mount on hibiscus 15 that @Steven has and it fixed the wipe location. Also it was missing the front right washer when probing and it now hits it. However it still barely touches that washer. @marcio we may need to shift the probing locations to the right a bit for a more consistent probe
Wed, Nov 29
@logan the wipe location shouldn't change. There are no changes to the sheet metal or the wiper pad location. It turns out the model for the extruder mount was placing the nozzle about 2.75mm back from where the nozzle on dodo is. This has been corrected with extruder mount v2. pushed to hibiscus. Cluster is printing some out now.
Tue, Nov 28
@Baldwin @kent We have roughly a 1kg mass for x gantry. With a drop from 1 meter (h = 1m), the potential energy is PE = m*g*h. Due to conservation of energy this is all converted to Kinetic energy right before it hits the ground. So m*g*h = (1/2)*m*v^2 This means v = sqrt(2*g*h) = sqrt(2*9.81*1) = sqrt(19.62) = 4.43 m/s . Impulse is a change in momentum so m*v_i = 0 and m*v_f = 4.43 m/s therefore the impulse is m (v_f - v_i) = 1kg ( 4.43 m/s - 0 m/s) this would give an impulse of 4.43 kgm/s or 4.43 N*s . Please correct me if I'm wrong. Also this is divided amongst the two z belts so each side should see 2.215 N*s
The gusset location on z uppers v2 will not work because they intersect the spool arm mount. These have been stopped in the cluster. After a discussion with @kent we believe this issue is not as prominent as initially thought. We believe that the gap between the z upper and the frame made it look like the part was flexing a lot. When in reality the part doesn't flex very much at all. @kent and I are closing this ticket. I am removing these v2 z uppers from the repo.
@kent cool sounds good man!
@kent I have rev'd the z upper left and right to have a gusset that should prevent this flex. They are Z upper L/R V2. Pushed to Mini Hibiscus. I will get some printing in the cluster now
@kent did you get a chance to replace the x end motor on this machine?
@karrad @kent @nickp Thanks Brent and Nick that helps a lot! The frays are mostly going to happen from the belts slipping while being clamped, not from being bent around too small of a radius. Also as @west stated both TAZ and MINI used to have their belts tightened extremely tight prior to gladiola and taz6. We only implemented belt tension meters into production when these printers hit the assembly line. I feel comfortable moving forward with the new tensioning systems at a 6mm diameter bend. Unless there are any objections I'm going to set this ticked to closed.
@karrad or was it that the belt was slipping? Ive seen y belts slip on gladiolas, but the mount and clamp were changed mid production to correct for this. Ive never seen a belt break on one of our printers
@kent the belt bends around a 6mm diameter loop on the x ends, x belt tensioner, and y belt tensioner ( this is the same diameter bend on the y belt mount which we used for all of gladiola production) while the minimum recommended sprocket diameter for a 2mm pitch GT2 is 7.13mm so we are just on edge of the minimum sprocket diameter. If this is major concern we can adjust that bend on all belt tensioning parts to meet the 7.13mm diameter. However, I dont forsee this causing any issues with our belts since we only tension them to around 30N. To my knowledge we haven't seen any issues with the integrity of the y belts on previous machines. Please correct me if I'm wrong.
Mon, Nov 27
@kent "The tooth shear strength is greater than the strength of the tensile member when at least six belt teeth are in mesh with sprocket grooves."
@kent we wouldn't need custom smooth rods for this to work mechanically, however if we choose to go to blind holes on the x end motor, we will have excess rod hanging out the end of the idler side, we could widen the idler to cover this up cosmetically, however my concern remains that we would be relying on belt tension to hold our x ganrty together, and if we went to blind holes, then the 25N from the belt would be held by 2 setscrews instead of 4. which is then roughly 12.5N on each setscrew
@kent im not at all against continuous belts, but at the time when we discussed this option, there was a 6 to 8 week lead time for the belts we would need i9n production quantities. Also are these part numbers from Misumi for small quantities (onesie twosies) or are these for production quantities? We cant just make a few printers and then be forced to wait 6-8 weeks to build hundreds/thousands more. We need to be able to get production quantities well before we release this product. If you want to order some belts and mock one up im all for it. We would all like to see continuous belts, thats why TAZ7 will get them, but given our research into this before, it seemed unlikely to pull off by the time we release MINI2
@kent another problem with using blind holes in the z lowers is there is a gap between the top of the rod and the frame, and unless you sandwich the rod between the frame and the blind holes (not advised), the rod will be free to shift up and down and will essentially be "loose".
@logan sounds good hope to see a few prints of this soon
@kent I dont think it would be advisable to just not install a fastener there as it would appear to the customer that we forgot one during assembly and would appear as an oversight on our part. It really isn't very hard to work around. I have been able to leave all connections attached to the RAMBO and remove the 4 screws and push the board to the side to get access to this screw. Have you tried doing this?
Wed, Nov 22
I inspected the binding on this printer and @kent and i believe it may be due to an issue with bushings on the x motor side. Id like to replace that part and see the one that came off that printer for further inspection
@logan looks great lets get a few printed out and give it a shot
@kent yes the set screws being blocked by the motor is a trade off for more z build height and i think its well worth it.
@kent is that not what we're doing? the new design is much easier to assemble and assembly correctly which is exactly what you're asking for. Have you even tried the new z belt clamps? Also i agree that with us doing the repairs and reworks we need to make it easy on ourselves, I'm actively trying to do just that. I haven't heard any suggestions from you on how to correct this issue just that you perceive it to be a problem.
@kent as @west mentioned, the EINSY needs the motors to be run in series so they get enough current. In parallel it acts as a current divider. Im guessing youre seeing "binding" when moving the axis up.
@kent While I agree that this complicates serviceability, this has been the case for all of MINI production, and since there is no change to the sheet metal on MINI2 there really is nothing we can do about this.. We would either have to move the RAMBO/EINSY location or move the mount location for the z lowers, both of which would require redesigns of sheet metal as well as printed parts
@kent actually since theres 4 setscrews and 2 rods with one belt tensioned at 25N they each hold about 6.25 Newtons no? Again we cant be held accountable for things like the user
over tightening things or assembling them wrong. You can over tighten the lug nuts on your car and strip out the posts just like a user could over tighten the screw in a printed part and break it. Theres a reason for documentation of torque specs and we post those specs in our OHAI kits, we cant be held accountable for the user not following our instructions.
@kent I see what you mean however it shouldn't really matter that the steps for the axes are set differently, the issue is that the XY axes are set to 100.5 to compensate for ABS expansion where as the Z is set to a calculated value, if we set the steps to what they calculate out to then we should have no issues with alignment. XY should be 100 steps per unit
@kent we're looking at getting some higher torque motors for the z axis. We discussed higher torque motors with andy from moons this morning in our conference call. Also you might check that the slice isn't causing the precision bushing to sit crooked, which would definitely make this issue worse as it would increase friction. As far as i know that is the only machine that has had any binding issues. I think this will require further testing and investigation
@kent i think we should have a 30-45 Newton tolerance for z belts with both sides being tensioned as close to each other as possible. And yes i do think that would be useful!
@kent we already are making it less likely to be assembled at an angle with the new belt clamps. Also we cant be held accountable to what the user does to their own machine. They could re assemble all kinds of things wrong, that is not our fault though we just need to make sure it is assembled correctly when it leaves the factory. They could install their heat bed upside down, they could reattach the toolhead hanging off the side with only 1 screw, they can rewire it completely wrong, but thats not our fault. The design works with proper assembly.
@kent are you using full height NEMA 17's also what motor current are you running? is it on an EINSY or a RAMBO?
@kent this is due to the fact that the x gantry guide rods are spaced closer together. This was so we could pull 180mm of build height out of the z axis without changing the size of the MINI. We could try using blind holes on the x motor side and get rid of the setscrews entirely but i am a little hesitant to get rid of the set screws altogether. the belt tension should hold the rods in the blind holes but then you are relying entirely on belt tension to hold together the x axis. This doesn't seem like a good practice
@kent as I'm sure you're aware, this is because the belt teeth are interlocking to hold tension, so we essentially have steps of tension. we have a belt tension target range which should still be easily achieved using this method.
@kent with the new style of tensioning the z belts this is not an issue. The belt tensioning collar does deflect the belt path away from the axis a tiny bit, however this collar is not needed for tensioning the z axis anymore, it can be used as a reduntant clamp or just to hide the slack. This still does deflect the belt path a small amount in the x axis but only by about 1-2mm or so. We can change the way the x belt is tensioned to get rid of this but i do not see it being an issue. Also both the MINI and TAZ6 have y axis belt paths that arent parallel to the y axis by a lot more than 1-2mm and we havent seen any issues there.
@kent we can likely add a gusset to prevent flex in this part. also increase density or shells in the slice
@kent this sounds like an assembly issue. Also there has been a revision to the z belt clamps which prevent the stress marks from tightening and should also prevent the belts from being installed at an angle
@kent which parts are you talking about? the Y idler and Y motor?
@kent yes bending a conductor does weaken the material, but it only gets bent once and only to about a 45 degree angle, and as @west said with the EINSY likely being the new board we wont even have an endstop switch there. Also we do currently bend them on the finch toolhead as well as the standard toolhead for gladiola
@kent the same could be said about the lead screw driven z axis on TAZ 1-6 as well as the MINI since the lead screw is offset from the smooth rod and theres only 1 guide rod. The x and y axes on all of our machines have the drive component positioned equally between two guide rods so there isnt a moment on the bushings. This could be corrected in the z axis by positioning the belt in the middle of the guide rods, however i havent seen any hibiscus experience z binding. Have you actually seen a z axis bind up? Also we need to have the precision bushing be rigid in the part so the part doesnt experience both lateral and radial play about the guide rods.
Tue, Nov 21
Mon, Nov 20
Pushed to Hibiscus
@logan I just installed the new revision of the clamp with longer standoffs and now there is no whitening with good bite on the belt still. Im printing another set for you to try out right now.
@logan yea i can def do that. ill make those changes and get a few more printed out to try.
i will try both a thicker clamp and a horseshoe design
I think we should both deepen the pocket for the nut a bit and the chamfer for the screw to sit flush
Thu, Nov 16
i agree with @karrad that we should avoid introducing new materials in the cluster for production. I dont see the stress marks being an issue. Another problem with pc max is it doesn't adhere to PEI very well, and often needs gluestick to adhere which we cannot use for production parts as it would require a wash after printing. We would then be dumping lots of PVA water down the drain which is not ideal
M5 X 25mm would stick out of the nut by 3-4mm. we can pocket the nut holder on the part an additional 1-2mm for better engagement of the nylon.
@MikeR after cutting away as much as possible on the x bearing holder, the belt only begins to rub when the toolhead is at the edge of the right side of the print bed. I believe this is as good as we can do, since people will rarely be printing on the edge of their beds anyways i think this belt will see minimal rubbing. I have pushed this to hibiscus. this is x bearing holder 3.5 please give it a try and let me know what you think
we could cut more material away from the bushing holder but im a bit concerned it might be too thin and break if we do that. there is a taper on that part already so the belt just barely rubs. i will model one with a deeper cutaway and try it out
Wed, Nov 15
i suspect there may be a varying number of links in the x chain. the prototypes were built using x chains for gladiolas. These chains only have 17 links, where the hibiscus should have 18 links.
this has always occurred with the spool mount arms for mini. Users likely wont be flipping their spool arm up and down very regularly, and ive never heard of it causing any issues.
this could also be adjusted by adjusting the z height for the wipe in the start g code.
also the rods are too long for blind holes. We were planning on using the same rods used in the older mini's. If we want to use blind holes then we would need to get custom rods at a shorter length.
After seeing various installation lengths, myself and @MikeR agree that 90mm is a good length
Nov 14 2017
Nov 13 2017
Nov 10 2017
there is a new fan shroud in hibiscus that fixed this interference. @logan and myself have confirmed that this correction does fix the issue
@MikeR Have you actually confirmed it tears them up faster than the regular nozzles or is this more of a perception? I havent seen this and i know @Baldwin said he hasnt seen this either. This may be the case, however these nozzles are superior to the old nozzles used. So we wont be changing nozzles. Another thing to note is we state for customers to watch their probe sequence, so a bad probe should be caught by the user and they can clean their nozzle more if need be.
Nov 9 2017
you have to slide the tensioning collar down until it is about 3-5mm above the x end. It should then clear the z pulley
how tight are you torquing these to? they should only need 2-3 in*lbs to hold the rods.
I would recommend lengthening the harness on the toolhead side of the connector. If we extend the length from the cable chain beyond 80mm, were then shortening the wires on the rambo side of the harness by that amount. The toolhead assembly has some planned changes as is so we can lengthen the wires there.