TAZ 6 Z-probe inconsistent

I’ve had a Taz 6 for 3-4 years now, and I am always having to re-adjust the z axis. Alot. For instance, a couple of days ago I printed some PETG, and had it set to give me a wonderful 1st layer. Switched to some ABS after today, and had to change the z offset from 1.25 to 1.5, and it still could go a bit more, we’ll see. I will guarantee that in a day or two I will have to bring it back up again, for no apparent reason, even if i don’t switch materials. I always heat the bed plate up for 15 mins or so before i start. Seems like it will be fine for a print or two (or ten sometimes, but not usually), then it will require a big adjustment to get the layer to stick down or to keep it from smashing onto the bedplate. Makes the self-leveling feature on this machine a joke.
This seems like a crazy high amount to have to jack with every time, and I am wondering if others with this machine have this type of inconsistency?

My TAZ 6 has a Z-offset between -1.2 and -1.3 and as long as I have the same tool head installed, I don’t change the Z-offset.

Not sure why yours is positive.

Mine IS negative, I just forgot the sign. Currently at -1.50. I get the large offset differentials with no toolhead change.

Does anybody run with autolevel disabled? Is there a way to do do that? Just wondering if there would be a way to isolate the issue - if it goes away with autolevel disabled, that could help me figure out the root cause.

What version of firmware are you running?

In addition to G29 , Marlin defines M420 and M421. However, I don’t believe the bed probe data is currently stored in EEPROM. To change that would require creating (and maintaining) custom Marlin firmware. I’m also not sure how much room is left in the EEPROM.

The output of autoleveling can be viewed with either “G29 V4” or “M420 V”. As long as the data is valid (check with “M420 V”), another G29 is not needed. “M420 S1” turns bed leveling on.

I have my TAZ 6 connected to a Raspberry Pi 3B running OctoPrint which gives me the ability to capture the communications in a log file. I’ve changed my slicer start gcode to do a “G29 V4” instead of a plain “G29” so I can view the results later if necessary.

Currently running version

Seems like I’ve upgraded firmware throughout the years, and haven’t seen any change to this issue.

I have found that if the nozzle is even slightly dirty, it will push the table down and get the wrong value when leveling. I have also found if the temperature is too high of the print head during leveling, a bit of plastic will drool from the print head making the head dirty while leveling.
Watch the bed very carefully while it is leveling, if you see the bed move even a little bit on the second probe of each corner, stop the print, clean the tip of the extruder and try again.

That sounds plausible, thanks for the tip. I typically hit it with scotchbrite, but maybe not often (or thorough?) enough. I will keep a close eye on it next time.

The z-offset is equal to the distance between the top of the corner washers and the print surface. This is a physical distance built into the machine - does not change with material. Measure and set and leave alone. Mine is -1.35 mm.

Most likely, your hot end needs cleaning to get a consistent zero.