Strange pattern - for the left 4 x positions values rise first and then go down again. This shifts to higher y values with increasing x and last x row just increases. So bed shape is clearly not flat.
The right fron es especially hard 11:16:05.590: G33 X290.00 Y10.00 Z-0.465 11:16:05.577: G33 X220.00 Y10.00 Z-0.247
Same on left front
11:16:05.542: G33 X10.00 Y10.00 Z0.195
11:16:05.544: G33 X10.00 Y70.00 Z0.252
These are values with exceptional high values. This also means you will not follow the curves very good at that area. Firmware assumes linear change between positions and when you get suddenly bigger changes there might be some very local change so you don't know where the change starts. It will go up a bit earlier.
The reduced z values get added to z position so positive values go away from bed while negative go closer to bed.
Since increasing z speed does not help - what is your z jerk? Try increasing it a bit as well.
My bed is definitely NOT flat lol. I can home in the center of my bed and all 4 corners are lower. I tried manually leveling the bed by starting in one corner then going around the edges then I did a 9x9 grid on G32 and G33.
I tried upping the Z jerk from 0.5 to 0.8 and it did nothing. At this point I'm wondering if the calculations needed for Z correction are too much for my 8 bit board. I'm going to try to disable the display to see if that frees up some memory or processor to handle it a little better. I'm waiting on filament for a project so I have plenty of time to try stuff lol. The main issue is I am working on a 0.1mm layer height model and the first layer has to be on the bed and not a raft or the pieces will not attach together flush. I've tried a raft on multi part models before and it just doesn't look right. If I weren't thinking about replacing it with an Ender 5+ then I'd get a glass bed and 110v bed heater.
But all that aside, I've been doing a lot of diagnostics on why the front left is always further away from the bed when printing even though I raise that corner up before calibration. I tried reversing the bed probing so it starts on the opposite corner (back right) and ends on the front left and it still did the exact same thing, so it's not the BLTouch being affected the more it probes. But it's nice to know the BLTouch does the same thing no matter what order it probes the bed.
When loaded into Excel and viewing a graph (I wish I could upload an image), it shows the area that has issues printing is actually higher than the back right where it prints too close. It almost looks as if the lower the area is the more it compensates for the curve.
Here is the tab delimited numbers if you want to easily load my G32 into Excel and look at the surface graph. The top row is the front and bottom row is the back and I made them negatives so it shows accurately in surface graph. Looking at it in Excel shows how truly warped my bed is lol.
I built a core xy with a 32bit board and when I do G33 it does the same thing but a lot smoother. I copied the jerk and acceleration from that one to this printer and it did not help at all. I bumped my Z jerk to 2.0 and it still didn't help at all. If I try to print it ends up skipping from all the sudden stop/starts. Looks like manually editing the map after doing a G32 S0 then a G33 will be the only way to fix this.
I am currently trying to bend my bed back to an almost flat shape to reduce the correction needed. I also have a Buildtak Flexplate and am wondering how flat they are.
One problem is when beds are fixed in 4 points. Mechanically that is statically undefined meaning the 4th point z is given by an equation already. If z screw is at different height it will bend the bed looking a bit similar to yours.
Think about the slicer option to have first layer 0.2 or 0.3mm. I do this frequently as it reduces first level problems massively. Also making extrusion wider for first layer helps making it stick.
Reducing it in 10mm slices takes a bit of cpu time but on 32 bit it should be no problem. What are your accelerations? What happens when we correct z is that all axis need to satisfy jerk, acceleration and speed constraints. The lowest one defines the speed. But that is not really jerky or with pauses it would just get slower.
Jerk was up to 1.2 from 0.3 on the Z axis and 8 on X/Y and acceleration is up to 300 from 150 on the Z axis. Acceleration was always 300 on X/Y. No matter what I did the X/Y would always completely stop then instantly start again at height 0.2mm. It got better the higher up it got. I tried to print something and the first three layers shifted horribly from all the stop/start.
The bed was warped worse than I thought and I used my knee to try to bend it more straight while it was heated up. I got it from 0.6 mm variation in the worst area to 0.2mm (maybe 0.25 in some areas). G32 S0 followed by G33 works a little better now even though the front left is still not printing as close as the back right (maybe 0.1mm variation if that). I may try to make due with this for now until I decide if I'm going to buy a glass bed or not. The biggest issue is I just got a Buildtak Flexplate and it is attached to my warped bed. I may be able to heat it up quite a bit and get it off, but I don't want to ruin it.
My other option is to never print anything without a raft on this printer which would mean I need to move my Palette 2S Pro to another printer.
I think at this point the cost to get this printer printing perfect outweighs the benefit. Either I need a 32 bit board to handle the z correction or I need a glass bed and that may cost me my $110 Flexplate system also.
You have helped me get it printing 90% better than it was before I posted, including getting the offsets working which will also help on my other printer. Thank you very much for all your help!
I wonder you say that it gets better the higher you come. There are only 2 areas - the one where you correct distortion and reduce long moves to shorter and the space above.
If it is not dependent on motion settings but on cpu power it should be the same everywhere. With increasing Z the correction gets less and if that improves it looks like it depends on motion setting. On the other side the Z jerk/acceleratio/speed you told you have are more then plenty for what you have. Do you also have a jerk of 20 or higher for xy motion. Acceleration 300 for xy is quite slow. Can you increase to 1000?
What are your steps per mm for all axes? Full stops are never to be expected, but if your cpu load gets 100% parts get delayed so that is the last thing I can think of. And high steps per mm combined with high speeds can always reach that. Especially if you have double and quad stepping disabled the limit is somewhere at 10000Hz.
Step/mm are 80/80/400. I upped acceleration to 1000 on x/y and max jerk to 20. It seemed to help a little so I upped acceleration to 2000 and jerk to 25 but it did not seem to help any more.
My 32 bit board has all acceleration at 300, x/y jerk at 15 and z jerk at 1.225. I can barely see/hear it slowing down every 10mm unlike the printer I am having issues with. I really think a lot of it is the bed. If it had to do less correction I think it would not be noticeable.
But with all your help and me bending my bed in some to undo a lot of the warp, I managed G32 S0 then G33 and then start a print with a 0.1mm layer height and no raft! There was one tiny place at the top left where I made a last second adjustment that I realized I should not have, but you'll never see it when it's all assembled.
A couple of more small adjustments and I think this printer will be ok for now.
The most important part is I can now do better on my core xy with the knowledge I've gained from this printer. The core xy has been going for about a year on and off and could definitely use some improvements.
Comments
The right fron es especially hard
11:16:05.590: G33 X290.00 Y10.00 Z-0.465
11:16:05.577: G33 X220.00 Y10.00 Z-0.247
Same on left front
These are values with exceptional high values. This also means you will not follow the curves very good at that area. Firmware assumes linear change between positions and when you get suddenly bigger changes there might be some very local change so you don't know where the change starts. It will go up a bit earlier.
The reduced z values get added to z position so positive values go away from bed while negative go closer to bed.
Since increasing z speed does not help - what is your z jerk? Try increasing it a bit as well.
I tried upping the Z jerk from 0.5 to 0.8 and it did nothing. At this point I'm wondering if the calculations needed for Z correction are too much for my 8 bit board. I'm going to try to disable the display to see if that frees up some memory or processor to handle it a little better. I'm waiting on filament for a project so I have plenty of time to try stuff lol. The main issue is I am working on a 0.1mm layer height model and the first layer has to be on the bed and not a raft or the pieces will not attach together flush. I've tried a raft on multi part models before and it just doesn't look right. If I weren't thinking about replacing it with an Ender 5+ then I'd get a glass bed and 110v bed heater.
But all that aside, I've been doing a lot of diagnostics on why the front left is always further away from the bed when printing even though I raise that corner up before calibration. I tried reversing the bed probing so it starts on the opposite corner (back right) and ends on the front left and it still did the exact same thing, so it's not the BLTouch being affected the more it probes. But it's nice to know the BLTouch does the same thing no matter what order it probes the bed.
When loaded into Excel and viewing a graph (I wish I could upload an image), it shows the area that has issues printing is actually higher than the back right where it prints too close. It almost looks as if the lower the area is the more it compensates for the curve.
Here is the tab delimited numbers if you want to easily load my G32 into Excel and look at the surface graph. The top row is the front and bottom row is the back and I made them negatives so it shows accurately in surface graph. Looking at it in Excel shows how truly warped my bed is lol.
Here is my G32 S0 and G33 on the 9x9 grid:
I am currently trying to bend my bed back to an almost flat shape to reduce the correction needed. I also have a Buildtak Flexplate and am wondering how flat they are.
Think about the slicer option to have first layer 0.2 or 0.3mm. I do this frequently as it reduces first level problems massively. Also making extrusion wider for first layer helps making it stick.
Reducing it in 10mm slices takes a bit of cpu time but on 32 bit it should be no problem. What are your accelerations? What happens when we correct z is that all axis need to satisfy jerk, acceleration and speed constraints. The lowest one defines the speed. But that is not really jerky or with pauses it would just get slower.
The bed was warped worse than I thought and I used my knee to try to bend it more straight while it was heated up. I got it from 0.6 mm variation in the worst area to 0.2mm (maybe 0.25 in some areas). G32 S0 followed by G33 works a little better now even though the front left is still not printing as close as the back right (maybe 0.1mm variation if that). I may try to make due with this for now until I decide if I'm going to buy a glass bed or not. The biggest issue is I just got a Buildtak Flexplate and it is attached to my warped bed. I may be able to heat it up quite a bit and get it off, but I don't want to ruin it.
My other option is to never print anything without a raft on this printer which would mean I need to move my Palette 2S Pro to another printer.
I think at this point the cost to get this printer printing perfect outweighs the benefit. Either I need a 32 bit board to handle the z correction or I need a glass bed and that may cost me my $110 Flexplate system also.
You have helped me get it printing 90% better than it was before I posted, including getting the offsets working which will also help on my other printer. Thank you very much for all your help!
If it is not dependent on motion settings but on cpu power it should be the same everywhere. With increasing Z the correction gets less and if that improves it looks like it depends on motion setting. On the other side the Z jerk/acceleratio/speed you told you have are more then plenty for what you have. Do you also have a jerk of 20 or higher for xy motion. Acceleration 300 for xy is quite slow. Can you increase to 1000?
What are your steps per mm for all axes? Full stops are never to be expected, but if your cpu load gets 100% parts get delayed so that is the last thing I can think of. And high steps per mm combined with high speeds can always reach that. Especially if you have double and quad stepping disabled the limit is somewhere at 10000Hz.
My 32 bit board has all acceleration at 300, x/y jerk at 15 and z jerk at 1.225. I can barely see/hear it slowing down every 10mm unlike the printer I am having issues with. I really think a lot of it is the bed. If it had to do less correction I think it would not be noticeable.
But with all your help and me bending my bed in some to undo a lot of the warp, I managed G32 S0 then G33 and then start a print with a 0.1mm layer height and no raft! There was one tiny place at the top left where I made a last second adjustment that I realized I should not have, but you'll never see it when it's all assembled.
A couple of more small adjustments and I think this printer will be ok for now.
The most important part is I can now do better on my core xy with the knowledge I've gained from this printer. The core xy has been going for about a year on and off and could definitely use some improvements.
Thanks again for everything!