IR probe offset fluctuates

Ive got a kossel mini with a duet running firmware: 1.10+4 (2016-03-30) and a DC42 IR probe.
Recently Ive had a few problems calibrating, the issue is that the Z probe height offset seems to vary.
So far i've:
Heated the bed and extruder to PLA temperatures.
Homed the printer
Brought the nozzle down to the print bed using a piece of paper to determine where Z=0.08mm and set Z to that height
Lifted the nozzle 5mm
Run G30 s-1
Read off the head position as Z=1.09 and updated the relevant section in the configuration file.
Run G32

Then when I bring the nozzle back down to the bed, I can take it all the way to Z=0 and there still be a slight gap between the nozzle and the bed.

Interestingly I noticed when bringing the nozzle down to the bed the IR probe is triggering at Z=1.30
If I repeat the above process again running G30 s-1 now gives me a Z height of 1.27mm.

Has anyone had any experience of this or is there something i'm overlooking?

The level of ambient light isn't changing and the z probe board is firmly attached to the Effector assembly.

Thanks
Stuart

What is your bed surface?

I've run into issue with mine when it was a piece of glass ontop of the mk2a heatbed. I found putting a black piece of paper between the glass and pcb solved my issue. On my other machines I have painted one side of the PEI sheet flat black prior to gluing it to the glass. I think next time I take the glass plate off, I'm just going to paint one side flat black. It seems to help with the IR probe triggering which has let me mount it higher up so it doesn't catch on prints.

I'm printing on glass (with a coating of 'prittstick' glue), with a sheet of black paper ontop of the aluminium plate.
I've cleaned the glass so that the glue couldn't be having an affect on the IR probe but the same thing happens.

You may be probing too fast to get a reproducible trigger height. Try reducing the probing speed, using the F parameter in the M558 command.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Can anyone clarify how the IR sensor works and its sensitivity.
I assume its triggering based on the amplitude of the IR frequency it is filtered for. So I would assume that a heated bed will affect the response based on its temperature - it radiating a broad spectrum of frequencies.

Quote
MCcarman
Can anyone clarify how the IR sensor works and its sensitivity.
I assume its triggering based on the amplitude of the IR frequency it is filtered for. So I would assume that a heated bed will affect the response based on its temperature - it radiating a broad spectrum of frequencies.

It measures the relative amplitude of the two IR beams in order to be insensitive to background IR. In any case, the amount of IR radiated by a hot bed at the wavelength of interest (~900nm) is tiny.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Ive spent the last day or so tinkering and tried a few more things.

I cleaned the glass bed, redefined Z=0,
Remeasured the IR probe offset using G30 S-1 (which had changed again to 0.69mm)
Updated the offset in the config file.
Ran several instances of G32, followed by M665 and M666, I ran G32 at about half the speed it was running previously.
Updated the M665 and M666 lines in the configuration file to match those of the latest probing.

And I thought that had solved it as I was able to get some nice fairly small prints.

However, I came back to It today, with a larger print which has quite a large area in contact with the bed.
This is where the problems returned, the nozzle was too far away from the bed and even though the plastic was ending up on the bed in the right place, over long extrusion distances it was curling up rather drastically so quickly.

Ive included a screenshot of the M665 and 666 results I got from running G32 today, should the variations in numbers be so large on the bed radius?

Could the IR probe be defective?

Your X and Y towers position offsets in particular are varying far too much. I suspect you have either too much friction in your delta mechanism, insufficient motor current, play in the joints, loose belts or a loose pulley, or possibly the IR sensor is not mounted securely.

One way of testing probing reproducibility is to copy bed.g to another file but change S6 on the final G30 command to S-1. When you run that file, it will probe the bed but instead of altering the calibration, it will just print the height errors. So you can run it multiple times to see how consistent the height errors are.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].