Fill out above information andin all cases attach yourklippy.logfile. Pasting yourprinter.cfgis not needed Be sure to check our Knowledge Base and in particular this and this post
Describe your issue:
…PROBE_CALIBRATE is exhibiting very odd behavior. Without any changes in my X gantry or probe location I’ll get wildly different results when using the command. I had similar issues when I was first beginning to use Klipper. I’ll run PROBE_CALIBRATE and using a post it note TESTZ until I get a slight pull on the post it note. I’ll then choose ACCEPT and use SAVE_CONFIG to set my Z Offset and get something like 0.620. I’ll try and attempt a print and it will be printing in mid air WAY off of the bed where something like babystepping won’t help because it’s so far off. So I’ll clean off the filament that spaghettied onto the nozzle and run G28 and then another PROBE_CALIBRATE. This time getting a value like 1.815. I don’t know if this is just an issue that I have or if it’s an issue with the most recent version of Klipper. But I feel like it’s causing issues with my bed mesh as well as when it probes my bed I get a mesh that looks like a bowl but a straight edge shows nothing of the sort. I’ve tried probing with the heaters on, heaters off, completely cold, heat soaked for an hour, replacing the probe mount, clearing the bed mesh, creating a new bed mesh, disassembling the printer, checking with a straightedge and a light, and still get varying degrees of bad when using PROBE_CALIBRATE. Not once have I gotten a result that would get me close enough that I was able to print and just babystep to fix.
When I first started with Klipper I had the following happen.
Manually set z offset in printer.cfg to 1.300
Ran Probe Calibrate and the paper test from that ended up changing it to 0.245. Ran it again which changed it to 0.683. Neither of which were anywhere close to my manual paper test of 1.300.
So I reset it back and set it in my printer.cfg as 0.000 and removed the section at the end of the file. Ran probe calibrate again and it decided this time my z offset should be 0.604. This was still almost a full millimeter off the bed when trying to print. Ran probe calibrate again and this time it was 0.925. This time it actually stuck just a bit as it was close-ish.
But if me manually setting z offset of 1.3 and getting a perfect layer you’d think that running probe offset it would at least be close. But instead it’s all over the place. Each time when I run probe offset I’d run the nozzle down until it would snag the paper a decent amount and then backing it off just a hair to get the slight tug. Each time I had the filament retracted and the nozzle clean.
And don’t forget to store the results with SAVE_CONFIG
I started with a value of 1.25 and ended with 1.46
As the into to that section says: Don’t do that. They will adapt later.
Can you please share the values.
When you have a probe, the paper test is for calibrating the probe, not the nozzle:
It is important to understand the "paper test" even if one has an "automatic Z probe". The probe itself often needs to be calibrated to get good results. That probe calibration is done using this "paper test".
Without any changes to my X gantry or nozzle location or where the probe is located the z offset shouldn’t change much at all.
When I run probe_calibrate it sets Z offset’s that are differing by half a mm sometimes between each probe_calibrate command.
Each time I run probe_calibrate I follow the Klipper documentation on how to use TESTZ and I’m using the same post-it note/paper each time. After I accept that probe_calibrate has the right setting I use SAVE_CONFIG to save the z offset and restart Klipper.
I can run probe_calibrate and get a z offset value of 0.300 after save_config and then run probe_calibrate again and get a z offset value of 0.650 after save_config. That’s 1/3 of a mm difference between the two offsets which is nearly 2 times a typical 0.4 nozzle layer width. The difference between each probe_calibrate/save_config varies wildly in this same fashion.
This is the current probe_accuracy when right above X/Y home position:
probe accuracy results: maximum 1.622500, minimum 1.615000, range 0.007500, average 1.619000, median 1.620000, standard deviation 0.002550
With a probe_accuracy that small the difference isn’t in my probe. So it makes me feel like something is not quite right with Klipper.
I’m using the same board with the same drivers and a genuine BLTouch 3.1 on one of my printers. I cannot confirm your findings.
Maybe the vast amount macros are leading to an unwanted interaction.
I thought this as well and have since removed all of the macros but still get the same issue. I’ve since manually babystepped my nozzle to do a paper test without probe_calibrate. Once my current set of prints are done I’ll run it again with it being set perfectly and see how well it does.