so i decided to do some bed tramming recently on my H2S with dial indicator, i printed holder for it and noted measurements of 4 corners, and regulated them accordingly. And i managed to achieve decent results - on 4 corners deviation 0.01mm, but…
I measured middle of the bed, and there is -0.20mm, left and right side of the bed measures +0.10 - that is 0.30 deviation and banana shape - look at the picture
so my question - can i do anything to improve that (mostly the -0.20 middle point)? Some kind of capton tape underneath? Anyone done something similiar?
Really don’t worry about it but if you want to test, remove the screw from the middle front support bracket which is only used for transport and re-tram and re-test
Bed mesh is very good at handling numbers this small
I would re-measure it at 75C @ 15mins, not re-tram it, just for the info since you already went thru the work And retram only if you’re wildly out of spec IMO
I would suggest to confirm your measurement with a reference edge, e.g. a stiff steel ruler (shouldn’t be able to bend) and a feeler gauge. I once did a similar test as you on my old Ultimaker 2 and after lots of swearing I found that the force from the dial indicator actually did slightly bend the gantry and I hunted a ghost.
I don’t want to praise BBL for their exceptionally flat beds, the bed of my new X2D is hollow by 0,1mm from front to back. But I’m surprised about the banana shape. sheets usually hate being bent in two directions. it would have to stretch the material in some spots or compress it in others within the plane.
Maybe the bed is bent front to back and the X axis is deflected by the dial indicator depending on x position?
Just some thoughts to save you from my measuring mistake back then
By the way, I plan to flatten the build plate magnet with a diamond sharpening stone. Still unsure if I ask support if they allow that without loss of warranty on the rest of the printer. I’m sure, they will call 0,1mm within tolerance and refuse to exchange my bed.
For me it is about parts having a flat bottom. E.g. they don’t wiggle on a flat surface, or when you want to mount parts together they don’t have gaps. a difference of 0,4mm like for the OP is already very significant and will be even double when you mate parts. My 0,1mm are on the edge of significance for sure.
Question: why have you done this, you had a first layer issue?
No, i did not have any first layer problems, i just had dial indicator laying around, and i measured my bed out of curiosity first time, and it turned out, that out of the box i had 0.75mm deviation between front and back of the bed (back had -0.40 and front had +0.35).
When I upgraded my H2D to H2C my back was 1.5mm below front . I calibrated and printed and had no issue . When I realized I had not trammed after upgrade, this is when I discovered it… my part was not tall so it may have been skewed a bit but not a concern due to its small height
I have to disagree:
ABL ensures that the first layer has the correct thickness everywhere. It achieves that by following the height profile of the bed, e.g. when the bed is high in the front and the back and low in the middle by 0,3mm, then it will print lower by 0,3mm in the middle, so that the first layer has the same thickness everywhere, ensuring good bed adhesion. In the next layers, it follows the height map a little bit less every layer, so that after a bunch of layers it prints completely flat layers.
So in other words, ABL does quite the opposite of a flat base. Instead it intentionally warps the first layer when you have a warped bed. That achieves a nice base, that adheres good to the bed compared to a flat first layer.
So it is clearly desirable to not have a warped bed in the first place instead of trying to compensate for it.
The deviations in the OP are quite acceptable imho. And not every part is full bed size.
Surely it would be great to have a flat bed, but the shape changes anyway between 20C and 120C. Additionally the warping (depending upon the filament used) of the engineering parts will change that deviation again.
If you really need that high accuracy, you can start flattening it out with Kapton tape, but that is a very tedious job. And you are probably only tuning one sweet spot temperature.
When you design engineering parts for 3D printing, it’s important to consider that there might be deviations and design it in such a way that this is no issue.
If you need high precision, FDM printing on a hobby machine is probably not the way to go.
I made different experiences:
On my X1C, if the parts don’t detach from the build plate (and the build plate stays on the bed, both a big “if” without heated chamber), then even PC parts, that I have to print diagonally to fit on the plate, have been perfectly flat after cooling down, certainly much better than 0,1mm.
I haven’t investigated explicitly, if / how much the shape of the bed changes with temperature, but the parts I obtained at 120°C and 55°C bed temperature didn’t seem any different regarding flatness. But a good point to examine. Better know directly than to assume from results.
Anyway, I’m very sure that temperature causes far below 0,1mm deviation.
Regarding automatic bed levelling (should be better named automatic bed level compensation):
I’m sure that it is far better than 0,1mm. And that is necessary for a perfect first layer. Even 0,03mm have an impact on first layer quality and can cause first signs of under extrusion or slight bulging. When you think about it, 0,03mm is a whopping 15% of a 0,2mm first layer height! You certainly have an idea, what an effect a 15% error in flow rate means.
That is, why ABL probes the bed when it is heated. That allows to compensate even deviations in the range of 0,01mm.
So, if it is important for you, you can absolutely achieve 0,1mm or better flatness of the base of your parts. That certainly isn’t relevant for all or not even for many people. But for those who want or need it, there is certainly better advice than to generally say, that it is out of reach with FDM.
Maybe not on a Bambu, but on my Voron with a 8mm solid aluminum machined build plate, all my meshes between 20C and 90C differ. Not at my computer atm. so I cant share them. Even after 30m heat soak.
Materials with different expansion coefficients will cause deviations at different temperatures. That’s just simple physics. So the build plate in general is already a moving target. How much? Depends on the materials used and the construction.
When you are printing Cf or GF filaments, warping is minimal. But when you start annealing them your deviations might increase quite a bit. Non filled filaments tend to warp even more.
I didn’t give a general advice that FDM cant reach 0.1mm flatness.
But I should have been a bit more clear on accuracy. Still 0.1mm flatness over approximately 450mm (H2D) is very hard to get with FDM.
So I didn’t say it’s not possible, and everyone has to decide if their machine can meet the accuracy requirement they have.
And that might require the adding of Kapton tape to get within spec.
And surely, most materials are flexible enough that they allow a certain amount of deformation to get it to fit.
And retram only if you’re wildly out of spec IMO
. I calibrated and printed and had no issue 
