God awful bed heatmap



  • So I just ran my mesh grid compensation and it is god awful and I'm unsure as to why. It's using the DC42 IR probe, but some points are showing a deviation of over .8mm over just 40mm in the X axis. I'm thinking it might be my bed surface is insufficient for triggering the probe. I am using a build surface (this one to be exact: https://www.amazon.com/Printing-Surface-Printer-Platform-Sticker/dp/B075ZMHVB6) on a piece of aluminum plate. I'm thinking maybe the white and black contrast are messing it up? Here's a picture of the heatmap I got after calibration.

    It might be sag in the X axis, seeing as it is about 320mm across, however the tool head is not that heavy and it's using 2 8mm steel rods. I was going to upgrade to some 12mm carbon later on, but it looks like I might have to do that sooner rather than later.

    0_1549733317993_Shitty bed level.jpg
    0_1549733380904_Shitty bed level 2.jpg



  • I wouldn't go the carbon route, yes carbon is stiff, but only for it's weight, you might find that it sags more, also it's not a good surface for bearings- also, some carbon fiber is woven to be strong in only one direction.

    I would go 2020 beam with a linear rail, it will be very stiff.



  • Other people have complained of the ir probe not working well on print surfaces with a black white contract like that.

    One way to test would be to cover the surface with a dark matte piece of paper and running the probe again.

    You could also test by probing on a dark section repeatedly with G30 S-1 and then comparing the results to probing on a white section.

    However, the bulge does seem to be fairly consistent with X axis droop. The best way to verify this would be to do an actual test print of a bed check model with mesh compensation active and inactive. This is exactly the type of problem that mesh compensation is good at resolving. So it's not the end of the world.

    Try printing this: 0_1549736347013_bedlevel_nozzle_0.4_200x200-0.3-0.8.stl



  • @phaedrux It seems very consistent with X and Y carriage droop. It's a corexy, so the heavier the X carriage the more it's going to droop. I think maybe I should go with carbon rods on the X and linear rails on the Y. I'd rather not do linear rails on the whole thing. I've already dumped a ton of money into this project. I'll try it with a piece of glass. It could possibly be my bed but I doubt it.



  • Are you able to put longer bearings on your axes? longer or farther spaced bearings may help droop so long you don't increase mass too much.

    Think of a heavy barbell, hold it in the center and it will droop like crazy, hold it near the ends and the droop is greatly reduced.



  • @wyvern I could. I have 2 igus busings per rod right now. Honestly, I think a lot of my problems stem from the stupid heavy X axis. That makes a ton of droop occur in the Y-axis, and the print head experiences droop from X and Y. I think if I upgrade the X axis to 12mm carbon rods I should be okay. If the issue still persists then I'll put Linear rails on my Y axis and be done with it. The X axis consists of 2x500mm 8mm steel rods, so they each weigh around 200 grams, so I'm looking at about a pound on the X carriage with just the rods. Couple that with the head and such and I would say the carriage is nearing two pounds. I'm using some masking tape on the bed right now. Hopefully that will fix some of the extreme points I saw on the bed. If not it looks like I'll be getting some borosilicate glass cut for it.



  • So I just put masking tape on the bed (couldn't find any dark paper) and this is what I'm getting now. Still not great, but a hell of a lot better than before with those crazy peaks. Looks like I'll have to find a different print surface. 1_1549744071688_Annotation 2019-02-09 152627.jpg 0_1549744071688_Annotation 2019-02-09 152610.jpg



  • Now that you have a more accurate height map saved you can take the tape off and it should stay relevant as long as your bed stays relatively stable between prints. Just load the heightmap before printing. You don't need to do a full probing run every time.



  • This inspired me to try to get my removable mechanical leveling probe idea working, just stick it under the hotend fan with a magnet and remove it when it's finished.

    I need to find where I setup bed probing, since I have always just used paper.



  • @wyvern Just follow the documentation, it's super simple. Now I'm ripping my hair out figuring out how to fix this because a linear rail won't work.



  • I have a Hypercube Evolution. Started with 10mm hollow carbon rods for X but quickly replaced them with 10 mm hollow steel rods (ie 10 x 6 tubes) - much better. I also have 12 mm Y rods and 16 mm Z rods so the only measurably deflection is in the X rods. Over a 250 mm bed there is slight deflection but it is acceptably small with no need for bed compensation (Yet ?). Your demands for precision may vary. I also increased the X rod spacing to 90 mm to help with X carriage rotational stability.



  • @garis I have 8mm rods on everything, including Z, but there are 4 rods on Z so it should be fine. Why did you move from carbon to steel? Seems like carbon is better from what I have read. I think I'm going to go with 12mm carbon tubes on the X and 10mm steel rods on the y.



  • @surgikill
    Carbon fiber has a good deal of flex and low frequency natural resonance, and the surface is not made to have bearings/bushings glide on it- it's just epoxy.

    I have worked with a lot of CF with R/c and race drone frames, even designed a couple frames myself, it's the best if you need light and ridged, but high-grade aluminum and steel are stiffer and stronger.

    A large sized steel tube would be a great compromise of stiffness and weight.



  • @surgikill I encourage you to go steel tubes. I did the stiffness comparison of carbon vs steel but the data I found or the type of carbon fibre I got from AliExpress didn't line up. End result is steel is much better. Perhaps thinking on the go hear one X rod of large diameter to reduce sag and a smaller one as the helper. Also the X carriage has a tendency to rotate around the X directions so having the nozzle as close as possible to the lower X rod is preferred.

    In respect of Y and Z axes I guess you are committed. For future reference (!!) I see no reason to try and finesse those diameters. The only dynamic mass penalty is a slight extra on the Y bearings. For Z the 4 rod solution makes no sense to me. The best solution whether it be for single or double Z screws is 2 Z rods as big as you can fit. For double Z screws the Z rods should be in line with, and outboard of, the screws with a redesigned bed bracket. This allows easy precision assembly, less "over-constrained" z bearings, and hence lower chance of bearings binding, and improved bed alignment and stiffness in the Y direction. The Hypercube Evolution I think is a useful design concept, having improved on the original. However further design enhancements from Scott's version is necessary before it matures. As people simply extend the XY dimensions without understanding the mechanics (mass, acceleration, flexibility etc) leads to unhappiness. I speak as a mechanical person and try and avoid all the electronic corrections where possible.



  • So I put a mirror on my bed (mirror side down, used the matte back piece.) and this is what I got. Definitely sag.

    1_1549772636540_Annotation 2019-02-09 231421.jpg 0_1549772636540_Annotation 2019-02-09 231402.jpg



  • @garis The 4 rods are for the 4 lead screws. There is a rod near each lead screw. Only reason I am leaning towards carbon is due to the weight flexing the Y rods. That's my largest area of deflection right now, and the 10mm rods won't stiffen it up tremendously. What's the weight on 12mm steel tubes?



  • I hope I am understanding you - You are concerned about the mass of the X assembly including X rods causing deflection of the Y rods? If that is the case just go big on the Y rods and make them solid - easy to get and cheaper than tubes but normally cost is not an issue here. There is no performance penalty for big Y rods as they are stationary, just the added mass of bigger Y bearings which isn't normally a concern. ie there is no complex engineering trade off here. Going from 10 to 12 is a big step up in stiffness and there is negligible deflection from their own mass. Or go to 14 or 16mm and then you won'y be able to measure any deflection. Only the X rods need to be optimised.



  • @garis In a CoreXY printer, the mass of the X assembly not only affects deflection, but it also affects things such as acceleration and resonance. You can watch this video here. I can't just go throwing 16mm steel rods on everything unfortunately. I also can't go too large on the Y rods due to the fact that I would have to once again redesign large portions of the printer. Also, what steel/aluminum tubes do you recommend? https://www.youtube.com/watch?v=AKTvykTPjQw



  • have a look at this: https://www.thingiverse.com/thing:2839395
    it uses mgn15h for the x axis and its very strong. however it does add quite a bit of weight.


 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.