Adjusting first layer

[abecker]

I have a Rostock v2 with the HE280 upgrade and after running the calibration the first layer is so squished nothing is extruding. Is there a way to adjust the calibrated height the autocalibration sets?

[mhackney]

Yes there is. The Z max length is saved in EEPROM and can be manually tweaked (green arrow) OR there is also a Z probe height (red arrow) in EEPROM that can compensate automatically if the situation is consistent. I found I got better results if I set mine at -.2.

[abecker]

I think I figured out the issue. The Simplify3D profile I was using had a z-offset set to -0.15mm, and was causing the nozzle to be too close to the bed. Testing it now without the z-offset.

[mhackney]

One of the many reasons I dislike compensating for properly setting the Z height in the slicer. It almost always comes back to bite you.

[abecker]

After setting it to 0 it is working much better. Just have a spot between the x and z axis that is too close to the bed.

[mhackney]

Unfortunately, a growing number of us are experiencing this and I'm actively experimenting and researching it. Most of us are seeing the same result you are with the low spot between X-Z but there was a report today I believe from someone seeing it between Y-Z. I need to verify that.

[morgandc]

As you can see I am running thin on the XZ during this temp test. I was changing the temp on the printer while it printed.

[abecker]

The good thing is, at least for what I am printing, it hasn't had a negative effect.

[morgandc]

After cleaning the surface, recalibrating and running the new calibration ring.

[mhackney]

What were your temps? Did you print my gcode file or slice it yourself?

So you do see thinning in about the same place as I do and thicker opposite.

I am wondering of the bed mounts might be the culprit but I haven't been able to check as I'm doing a looooong print now.

Did you auto cal cold? It would be worth calibrating again and giving it another attempt.

[mhackney]

Here's some info I posted earlier on my build thread that morgandc is referring to: http://forum.seemecnc.com/viewtopic.php ... =75#p97356

It has the calibration ring STL and gcode and some thoughts and photos.

[mhackney]

@ morgandc - if you could, after calibration look at the tower offsets in EEPROM and make note of them. I want to see if yours is as consistent as mine from run to run.

[mhackney]

Ok, looking at morgandc's photo I have a theory we can test. If probing at the Y tower resulted in a little deflection (even a tiny deflection of .05mm could cause an issue) then that tower would be artificially too long. This would result in the apparent bed tilt as shown. Any flex or difference in clamping pressure at Y could result in this. I think it's time to scrutinize the mounting system.

[morgandc]

Used your gcode, but it is similar results to the circles above that I sliced.
I started the first ring at 220/60 and upped the temp based on watching it to my standard 220 so the layer "meshed" rather than separate strings.
calibrated cold.

This is a hot bed calibration, and 210/60.

[morgandc]

As requested

[mhackney]

It's the tower endstop offsets that I wanted to see. You should dial in the offsets to get all 3 below 4 steps. I posted how to do this in my build thread.

Your X is 164 steps!!!! That is over 20mm. That can cause calibration issues.

[morgandc]

Tower X endstop offset [steps] 164
Tower Y endstop offset [steps] 6
Tower Z endstop offset [steps] 0

I am assuming this is the endstop screw adjustment, I will work on that this evening. Not sure how it could be over 20mm off...

[mhackney]

Yes it is. The procedure is in my V3 build thread.

[morgandc]

I was waiting for my Raspberry Pi to show up and install octoprint so I didn't have to install mattercontrol on my wifes laptop to look at firmware, but I caved last night to get the Tower offsets, so I will calibrate tonight the offsets and check for repeatability. The Raspi should be here any day now...

[mhackney]

It is so nice doing everything over wifi! OctoPrint on an RPi 3 is pretty good and quite functional.

[mhackney]

I just did a test and unscrewed the stop screw on my X to mimic your 164 step offset. Recalibrated and the thinning/thickening on X to Y-Z presents itself. Then manually adjust so the offsets are all near 0, recalibrate and I'm printing very flat again! This gives us something to look in to.

[geneb]

This is like Witchcraft 101...

g.

[mhackney]

How so?

[morgandc]

I never would have guessed that the autocalibrate had "limits". I will test tonight.

[mhackney]

Yes it does and that's true with CNC machines too. Auto correction can actually correct for bad mechanics, backlash, etc. That will be the next wave of delta firmware. I know David Crocker is thinking about it. More than a year ago I developed a gcode post processor for KISS that did auto correction that used Barycentric Interpolation. This required probing the bed at a set of points that defined the triangles. It worked phenomenally well. At the time I did not have probing so I had to create the height map manually - one of the reasons I chose Barycentric Interpolation is that it didn't require as many probe points. Today, with a probe, I would just do a 10mm grid. The idea would be to create the height map and then compensate for Z in firmware as the gcode is interpreted. When I did my tests I could literally tilt my bed 1cm at the high edge and the interpolation would follow the slant.

Auto calibration (aka tuning) - used to find the calibration parameters for a "reasonably good" mechanical build
Auto compensation - corrects for issues, wear, backlash, etc