To be fair, the probe is accurate enough for a more flat bed, and the z-offset gets to do it’s job. I have very good first layers and adhesion and quality.
For the prosess, just don’t turn down the heat between meshes. There really isn’t a problem to handle it at 65c without getting burnt.
You could even just do one bedmesh, see what the deviation is, and shim accordingly. On sheet of standard printer paper is a little less than 0.1mm and the entire session shouldn’t take more than a few minutes
To be fair, it’s been 5 months since I started trying to shim the bed and gave up on it because I ran bed meshes back to back without any changes to the shims or bed mounting screw tension and got significantly different results so I wasn’t confident that the iterative shim and bed mesh process would converge on a solution, but in retrospect, that may have been more an issue of creeping bed mesh due to thermal expansion than inaccuracies in the prox sensor that was measuring height above the bed to make the bed meshes. I’ve since read on this forum that the bed is still changing shape 30 minutes after it’s heated, which would be much less of a problem with a thicker annealed cast bed. However, it’s probably true that the Eddy probe is at least 10X more accurate than the prox sensor so the trial and error shimming process will converge in fewer trials with the more accurate data, and each bed mesh scan will take less than 1/10th the time of the prox sensor so the shimming process will reach a solution much faster.
I need to fix the bed warping problem AND the slow inaccurate bed mesh problem to have the It Just Works first layer, so the SV08 can reach its full potential. Regardless of how much of my first layer problems are the sensor and how much is the taco bed, I plan to implement both fixes, and the proper order is the Eddy sensor first and then the bed shimming.
I can probably get good enough results with an hour or two of bed shimming, but if not, I’ll get a proper Voron bed for the SV08 ($$$). Even if I need to spend as much money fixing the first layer as I spent on the SV08, it’ll still be half the cost of a Voron 2.4, and 10-20% of the time for a Voron 2.4, so that’s the optimist view.
I’ve lived with the SV08 first layer issues for half a year. I’m beyond half measures. I want the fix to be as inexpensive and easy as possible, but I’m committed to dropping a nuclear bomb on it if needed to get a perfect first layer every time.
I guess it will depend abit on how the bed is warped. But mine had a substantial dip in the middle, and yes you can fix it with strips of abs or something on each low point, but I found I actually could just do a large area in one go and not tighten the bed screws more than I needed.
The problem with my SV08 as I see it is that there is no support for the bed under the bed assembly. By just giving it something to rest on will improve it substantially.
I think the gantry leveling is a issue. My printer seems to like to leave the back right hand corner low by a about a layer. If I let it warm up before leveling, it seems to work better and the mesh map looks much better.
Because the Quad gantry level process is just touching the corners of the bed to establish “level” before the bed mesh can establish “flat.” If the bed is not yet done deforming due to temperature… then the bed will be in a different position when it does the QGL and the Bed Mesh. The most reliable process is: Heat Soak-> QGL → Bed Mesh.
The built in macro will always do QGL at 65 degrees, even if you are printing at 95 degrees. I think demon macros fixes this, don’t quote me.
If you take of the metal print sheet, and look on each side of the hotbed you will see 4 holes on each side. That’s the bolts holding the bed assembly to the frame.
If you undo them, you can lift it up in the back. Just be careful as there are wires coming up from the electronics bay. But the underside of the bed assembly has no support, and the metal plate in the hotbed is not strong enough to support its own weight I guess. But putting in something to give the bed something to rest on will improve your leveling.
There are guides to drill out the build plate mounting screws under the magnetic sheet and use shims to even it out. But that is alot more work. But would be the more accurate way to improve the bed. But I would still recommend the “quick” way first to get a better starting point, and knowing that the sopport for the bed is at rest first.
Quick disclaimer for anyone doing it tho, it’s at your own risk and all that.
I don’t know about Demon’s macros (personally, I looked at them, stole a few things to put in my own macros, but while they were good, there was enough futzing around with them to make everything just right that I didn’t bother.) But you minimally have to home Z after QGL before doing anything else, because unless your gantry adjustments were perfectly symmetric around the origin, your zero is now wrong. Worst case it could be wrong by a few mm, and you crash into the bed while meshing.
Personally, my print start macro does this (I’m using eddy-ng, so the cleaning part is key).
Home XYZ.
Heat bed to print target temperature with nozzle at center, z=5 (because I want my eddy probe to be at >50c where I calibrated it). If target <90c, just do a 1 minute soak. If >90c, 10 minute soak. Part fan is on at 100% to circulate air and help heat up my enclosure - I can usually get to 40c at the level of the top of the wall panels by the end of my soak.
At 1 minute left in the soak start heating the nozzle to 150c and turn off the part fan.
Once soak is complete and nozzle is at 150c, QGL (my QGL and bed mesh macros are modified so they can be executed at any temperature - this is because I have an ERCF with blobifier, and I need a good QGL so I can park my nozzle at X330 Y360 for purging into nice blobs)
Home Z.
Heat nozzle to 200C for cleaning.
Once 200C, reset to 150C while going back to do the nozzle clean (I want it hot to remove stuff, but give it a chance to cool before the next few steps.
Go back to center and home Z again
Do eddy-ng touchoff to get 0 set exactly (the nozzle is back to 150c when it taps the plate)
Problem is the only thing to “Rest” on is a poorly supported plastic base. Any significant force will (over a week or 2) deform the plastic and stop pushing up on the bed.
I installed the Eddy USB on my SV08, after some errors at the beginning, I installed a ferrite cylinder on both toolhead and Eddy cables where they attach to the gantry, updated Klipper for ALL the MCUs, and now the errors are gone. I also installed Eddy-NG for the tap function and the bed scanning, man, it never fails, perfect first layers every time, and super accurate bed leveling.
One thing I must mention: in order to fix the “taco bed” I had to search for the hidden bolts, unscrew them, and manually tighten them at 65°c until the bed was flat, now it results warped at room temperature, but after 5 minutes heat soak (I usually heat the bed while I’m setting up and slicing my parts), it is super flat 0.12 max difference, 0.028 average, and the probe fits perfectly under the stock shroud.
I don’t know exactly what issues you had and your current setup, but if you want I can help you out
This is great to hear. I’ve been considering Klicky , although Eddy would be my preferred option, for quite a while now, but I keep coming back to the conclusion that however accurate probing is it really counts for nothing if the bed flatness deviates so much! I’ve recently built a Prusa Mk4s which has quickly become my workhorse, so I’m in a position now to really spend some time digging down into getting this SV08 right now. So, I for one would be very interested in more information about these ‘hidden’ bolts, before moving on to probing options.
Excellent work. Well done you!
Best,
T
This was good to hear, I have a new M12 inductive probe coming and will try that for better accuracy before I attempt to upgrade to Mainline Klipper and Eddy NG. I did end up heating the bed to 120°C for probably 30+ minutes and then retightening the bolts I got a little less taco warping.
Problem is some of us are just a little too picky and want a fix, not a temporary band-aid. The concepts of creep, thermal coefficients, elasticity etc. are not of much concern to those that don’t care about consistent results or took physics in high school.
One of my buddies keeps telling me I need to buy a Bambu. If only they were open source.
I need to move up to a printer that will print PEEK. Think the SV08 is going to be parts soon.
Edit:
You don’t need the pattern to find the mounting screws. They are centered down the middle and 19mm (center to center) to the out side of the hold down bolts on the right and left.
Thank you for that encouragement. I’m still waiting for Sovol to sell an Eddy upgrade for the SV08, but I hope to soon be following in your footsteps. The perfect first layer on the Zero has spoiled me and I need that on the SV08.
I started adjusting the SV08 print bed mounting screws trying to get it to warp flatter when heated, but the bed mesh with the original inductive probe was so slow and so inaccurate that I decided to fix the probe first and once I have fast and accurate bed meshes, then do whatever needs to be done to fix the warped bed - hopefully just some shimming.
first layer, so the SV08 can reach its full potential. Regardless of how much of my first layer problems are the sensor and how much is the taco bed, I plan to implement both fixes, and the proper order is the Eddy sensor first and then the bed shimming.