Homing to reduce z backlash

Watching Sprinter and Marlin home the z axis, it seems that the last step involves a slow downward move until the endstop is triggered. Since we print moving upwards, any backlash in the z axis results in the first layer being thinner than the rest of the layers since the initial z movement will be taken up by the backlash. It would be better to have the z axis moving upwards in the last homing step to take up the backlash.

Before I do anything, do I understand the firmware right? Is there a version that does this already?

Thanks.

At least with Teacup firmware you can home with only a Z Max endstop. As the Z home point is measured point minus available Z length, this requires to configure the available build volume pretty precisely.

Another option is to simply turn the build volume upside down and mount the endstop on the top side.

A third possibility is to send an appropriate G92 right after homing. Like G92 Z.05 to make the first layer 0.05 mm thicker.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

It seems more complicated to use the z max endstop to home. It seems more difficult to measure the distance between the nozzle and the bed accurately this way.

I don't know how turning the build volume upside down would work.

The widespread practice of setting the bed clearance with a sheet of paper (typically about 0.1 mm thick) acting as a feeler gauge is accounting for the backlash to some extent. Of course, if we know exactly what the backlash is, then we can pick the correct feeler gauge thickness or enter in as z offset in the slicer (slic3r, skeinforge, etc) or firmware using G92 as suggested. We can also experimentally account for the backlash by printing a 1 layer single wall object or a skirt and adjust the bed clearance until it comes out perfect. This is essentially what I do now. But I'm looking for a way to save on this trial and error method. And modifying the firmware seems to be the most correct way.

Measuring the distance is easy, home the axis, use some sort of "feeler gauge" to set the Z to some known value, use M114 to get the offset, set it in the firmware.
IMO it's a lot easier than trying to use some sort of adjustment to set the min Z end stop to 0.

This doesn't give you a measure of the backlash until you measure the thickness of the first layer on the skirt or a 1-layer print. Also, note that your offset will limit the layer thickness to be no less than the offset. You also can't measure a gap if there's not gap when the nozzle is pressed against the bed. Setting the z offset to 0 or the thickness of the sheet of paper is not really that difficult even with the standard Prusa without an adjustable endstop screw on the x end.

Edited 1 time(s). Last edit at 10/08/2012 06:05AM by brnrd.

On my printer, gravity takes up the Z backlash. I know this because I've measured the backlash with a dial gauge.
I home to Z max because it makes more sense to me, I home Z first which guarantees the operation won't hit anything.
My bed is rigidly mounted, I've set the offset maybe 3 times since I swapped to doing it this way.

Edited 1 time(s). Last edit at 10/08/2012 11:36AM by Polygonhell.

On my part I don't do automatic homing at all. Just move the head to a known position before disconnecting and send a G92 with these coordinates after a reset. Assumes/requires zero step losses across many prints, of course

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

is there a way to set z-offset with the lcd / rotary encoder menu and store it to eeprom ?
since this is a very machine specific value it would be great to have it stored inside the machine.
could not find it in the menues though.

grtx

Marlin at least uses the Z Max as the home position of max endstop, this isn't one of the variables stored in EEPROM, so there is no easy way to change it from a rotary type encoder without a firmware change (which probably isn't very hard).
The only time I change mine is when I disassemble the axis or swap to different glass, it's one number on configuration.h and uploading the firmware takes maybe 40 seconds.