Having an issue with variable nozzle height

My setup is as follows:

CoreXY printer.
Glass bed on top of a Mk3 PCB heater, with black card under the glass.
DC42's IR Z-probe
Duet 0.6
Latest firmware and web code (1.17a + matching web code)

When I was running under 1.14 firmware I very rarely had to change the Z offset for the IR probe - set it once and it was good for ages, even though I often move my printer across the room when I've finished with it. Now I've moved to 1.17a I find that I can't hold a consistent setting.

What I've done today is:

Clear all maps.
Heat up the bed and hot-end to typical working temps.
Home everything, then set the nozzle height at X100:Y100 so it just grips a piece of thin paper.
Tested the nozzle height a few times after re-homing everything to make sure it's consistent.
Either:
Run G32 and retested Z height.
Or:
Run G29 S0
Run G29 S1
Retested Z height.

I seem to get different results each time I try...

Now what's been happening is that I've tried to print test pieces - some come out just fine, others have the nozzle so close to the bed that the extruder skips, and others are so far above the bed that the extruded plastic barely grips the glass. This is when using G32 and/or G29 commands (tried them both together, and individually). As for testing with a piece of paper sometimes it's just right, sometimes the paper is pinned to the glass, and other times it flaps around in the gap.

Over the past couple of days I've probably reset the Z offset 5 times, and still I'm getting variable results.

The setup is much the same as previously used on old firmware - in fact, it's probably more rigid after upgrading the bearings and parts of the carriage.
Nothing is loose around the hot-end or carriages, and everything is reasonably firm when given a vigorous shake (no obvious movement or rattles).

Now I'm looking for guidance regarding what to check next...

tl/dr: I used to get consistent results (using G32) before upgrading the firmware but now I can't hold a consistent Z height.

I doubt that it has anything to do with the firmware upgrade, but you could always go back to the old firmware if you want to check. My best guesses are that your print bed is no longer stable, or your Z axis is no longer moving smoothly.

---

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].

I run into inconsistent Z heights with the ir probe every now and again. I usually clean the glass or PEI surface with the alcohol based glass cleaner, reset my Z height with the paper and everything is good after that when using G32.

Trying to use the G29 was almost impossible for printing as it constantly tried to use different starting heights every single time. Gave up with G29 and been using G32 knowing my bed is flat and level enough for the 0.15 layer to stick without being squished everywhere, but every so often I do have to clean the dust off the glass or PEI depending which surface I'm using.

Quote
dc42
I doubt that it has anything to do with the firmware upgrade, but you could always go back to the old firmware if you want to check. My best guesses are that your print bed is no longer stable, or your Z axis is no longer moving smoothly.

The Z axis was fine up until I upgraded a couple of weeks ago. I'll give it a once-over just in case, though.

I might try going back to 1.14, just to make sure that I'm not imagining things (if I can find a spare SD card, so I don't overwrite what I've got).

PDBeal: I might have another go with just G32, which is what I used to use before the upgrade. I had cleaned the glass before testing, and given it a fresh coat of hairspray (my usual technique, so no change before and after upgrade).

I've just tried another test:

1. Heated everything up to working temps
2. Homed X, Y and Z
3. Checked and adjusted nozzle height at X100 Y100 Z0 - just gripping a piece of paper.
4. Moved Z axis to max height and back to 0 several times, then rechecked afterwards - still OK.
5. Ran G32, then checked nozzle height at X100 Y100 Z0 - nozzle too high, paper slips through (probably get 2 sheets in the gap).
6. Sent M561 to clear the compensation, then rechecked nozzle height - just gripping the piece of paper again.

Next test is to reload 1.14 to see how that behaves in the same test.

Quick update: I haven't tried 1.14 yet, but I have clarified my problem - the system seems to be over-compensating after running G32. I'll start by admitting that my bed isn't level - but it's not far off. If I get the X100:Y100 nozzle height correct it will be too low at X0:Y200 (almost touching the bed) and too high at X200:Y0. After running G32 it's too high at X0:Y200 and too low (touching the bed) at X200:Y0. It seems to have reversed the error, rather than evening it out.

And, before anyone says anything, my next task is to get a manual bed-levelling system working properly. One job at a time, though...

David - any views on what I reported in my last post?

I haven't yet tried to go back to 1.14 as it's a lot easier to go forward a version than back... I am now in the process of sorting the bed levelling though, so can't do any more testing until that's finished.

Final update: I sorted the bed levelling mechanism (now 3-point suspension), levelled everything up as best I could, and tried again - and now I get a good print across the bed.

However, while this is a good result, it doesn't really answer the original problem as I suspect that my printer will now give me a decent print without invoking any bed levelling! The issue of over-correcting the errors on my old setup hasn't gone away.

Edited 1 time(s). Last edit at 01/11/2017 04:12AM by David J.

One thing that has recently come to light is that if you run G29 to probe multiple times, then you should use this sequence:

1. Send M561 to clear the bed trasform.

2. Home using G28.

3. Run G29

If you don't send M561 before homing, then the height map can affect the Z position after homing, leading to a Z offset that affects the G29 results. Likewise, if you load a height map, you should clear the bed transform and home first to establish the same Z=0 setting as when you generated the height map.

I will review the interaction of Z homing and the height map to see how this can be improved.

---

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].

Thanks David,

I will have to review the sequence of events I've set up. I'll have a go at that tomorrow.

Edited 1 time(s). Last edit at 01/11/2017 05:22PM by David J.

David,

I had a quick look at my start-up G-codes, which are as follows:

* I have a macro that runs M561 followed by G29 S0, which gets run at the start of a printing session (if I think it's appropriate).
* My startup G-code is:

G28 			; home all axes
;G32			 ; auto-levelling
G29 S1			; use the bed mesh map
G1 X10 Y10 Z20	; go to the front left side, raise & await next move

Note that G32 is commented out - I've been swapping between G29 and G32 while testing.

Now as far as I can see, if I run several print jobs (or try setting the nozzle height repeatedly), the old transform will never get cleared out and subsequent jobs may end up with the wrong Z height. Am I right in thinking that this is the problem you described? And can I assume that this problem would also occur if I use G32 instead of G29?

I have modified the startup G-code as follows:

M561			; clear current bed transform map
G28 			; home all axes
;G32			 ; auto-levelling
G29 S1			; use the bed mesh map
G1 X10 Y10 Z20	; go to the front left side, raise & await next move

I've not yet had a chance to retest, but I am hopeful...

Quote
dc42
One thing that has recently come to light is that if you run G29 to probe multiple times, then you should use this sequence:

1. Send M561 to clear the bed trasform.
2. Home using G28.
3. Run G29

If you don't send M561 before homing, then the height map can affect the Z position after homing, leading to a Z offset that affects the G29 results. Likewise, if you load a height map, you should clear the bed transform and home first to establish the same Z=0 setting as when you generated the height map.

Based on this, is it worth modifying the homez.g and homeall.g files to include the M561 before the homing routines are called?

Final update:

Adding M561 as I described above seems to have fixed my problem completely - yesterday I printed some large pieces that went to the edges of the bed and all came out perfectly.

I shall now download version 1.17c and see how I get on with that - should I try it with M561 in the startup gcode, or out?

If you are using the Z probe to do Z homing, then sending M561 before homing is no longer necessary in firmware 1.17c.

---

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].