Print bed levelling

My Fisher works okay finally after replacing the nozzle, the only issue I have at the moment seems to be my bed levelling/auto bed compensation.

At one side of the bed the nozzle will be touching the bed whilst printing, but then at the other side it will be too high and the PLA won't stick to the bed.

using 7 point 6 factor bed compensation I get ~0.15 deviation which I know isn't amazing, has anybody got any tips on how to improve this? and maybe a good technique to make sure the bed is level?

1. Have you measured the probe trigger height at each probe point, and used the H parameter on the G30 commands in bed.g to correct for the variations?

2. On delta printers generally, the following geometrical errors will prevent you from getting a very low deviation,; but I don't know which ones the Fisher design is most susceptible to:

- Diagonal rods in a pair (measured between bearing centres) not quite the same length

- Diagonal rod bearing spacing not quite equal at top and bottom of the diagonal rods

- Bearing pairs on carriages not quite parallel to bearing pairs on effector (e.g. because the carriage is rotated about the Z axis)

- Towers not perpendicular to the bed

---

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].

Bed levelling on the Fisher is more complex than for others machines because the force to activate the probe varies significatively with the position of the hotend, as it pushes from one to three springs. So the machine deformation varies for each probe point. The offset variation between points may vary from 0.2 to 0.07 depending the spring force, the belt tension and the quality of assembly.
Check the offset with the 'paper sheet' method.

You shall have the maximum belt tension which does not give excessive friction.

One very important point on the Fisher is that a lot of beds are not flat, and this is not symetric (say, it is not a perfect bowl). That fools a lot the calibration routine. It is very important to have a flat bed and you may replace yours if it is not flat. Flat Acrylic, PVC or wood plate will be ok. DONT use aluminium on a not heated bed (I tested).

By construction, the angles of the Fisher are near perfect, and if you use 6 factors calibration, that may improve the calibration, but your dimensional accuracy may have problems as the inaccuraccies of your build troubles the calibration routine.
So, if this is for artistic stuff, do a 6 factors calibration, but for mechanical parts, do a 4 factors.
You will have better practical results if you have 3 or 4 probe point more than the factor, say 8 points for a factor 4 and 10 points for a factor 6.

Also, I probe with a hotend at 150°C, because low temperature probing don't work well, and high temperature (~200°C) probing tend to make holes in the Buildtak.

You also need high temp, thick, wide and slow first layer. I was using 0.35mm thickness, 0.8mm wide and 18mm/s for the first layer. Temperature of the first layer may be 10 to 20°C more than other layers. The adhesion of the buildtak is related to the fact it is heated during the deposition of the first layer. That is why aluminium don't work with buildtak, as it dissipate the heat.
Also, don't 'crush' the first layer on the buildtak.

Edited 2 time(s). Last edit at 04/03/2016 08:55PM by PRZ.

---

Pierre

- Safety [reprap.org]
- Embedded help system for Duet and RepRap Firmware [forums.reprap.org]
- Enclosed delta printers Lily [rouzeau.net] and Lily Big [rouzeau.net]
- OpenScad delta printer simulator [github.com]
- 3D printing on my site [www.rouzeau.net]

I'd also check that your three screws on your built plate are all the same height, stick your calipers over to be sure. Also check the contact points are tight as I found they can come loose and then they drift upwards a little bit.

I'd also recommend you invest £4 in a glass bed, my local glass company cut me some circles in 2mm and 4mm float and only charged me £4 each. I then use double sides carpet tape to stick them to my acrylic plates. I then use insette hair spray rather than tape or build tak.

---

DC42 Kossel 330mm x 2meters
My Thingiverse Creations

Quote
DADIY
I'd also check that your three screws on your built plate are all the same height, stick your calipers over to be sure. Also check the contact points are tight as I found they can come loose and then they drift upwards a little bit.

I'd also recommend you invest £4 in a glass bed, my local glass company cut me some circles in 2mm and 4mm float and only charged me £4 each. I then use double sides carpet tape to stick them to my acrylic plates. I then use insette hair spray rather than tape or build tak.

That's a great idea, I will get onto the glass company ASAP! Also never thought to measure the height of the screw heads, obviously if one is higher than the others then the bed will be lower in that location, I'm guessing the acrylic spacers aren't that accurate.

Quote
dc42
1. Have you measured the probe trigger height at each probe point, and used the H parameter on the G30 commands in bed.g to correct for the variations?

Thanks for the help, but how would I measure the trigger height? Sorry I'm still a noobie here.

Quote
PRZ
Check the offset with the 'paper sheet' method.

I've never heard of this before, what is the paper sheet method?

First have a look on the wiki reprap firmware macros page :
[reprap.org]
use Z 0.05 steps or lower (see above page)

Extract of a (yet) unpublished commisssioning page:

Generally (this is not the case for all systems), the sensing need more force in the center of the bed, so a larger offset, as while sensing in the center, you have to actuate three sensors/springs while only one sensor is needed near the columns.

[edit] You may use M122 instead of creating bed_report macros.
Bed, bed_report and P0 to P6 macros shall have same coordinates and be adjusted while you displace probing points, which may notably be needed if your bed surface is weared (as could occurred easily with BuildTak) or for a different usable diameter.
To test the offset, you shall :
Do at least 3 calibrations till you get consistent results.
Execute M665 and note the reported parameters to modify your configuration file.
Please note that if you use the homing macro without end switches, the height in M665 will not be used and the height shall be modified in the homedelta.g

Move to the calibration points with the P0 to P6 macros and lower the hotend with the web interface buttons atop a paper sheet till the paper sheet began to have difficulties to slide. There shall be no plastic 'blob' on the hotend nozzle. Alternatively, you could do that with hotend at 150°C, but beware your fingers.
Repeat for all the probe points, note the values on a paper.
You shall then modify the H parameters in the bed.g file and reload this file on the SD card. I find the FTP the easiest method for that.
Values found shall be substracted to the already defined offset, meaning that if you find +0,1 and the offset is already -0,22, you shall do -0,22-(+0,1) = -0,32
If using a ‘BuildTak’ surface, you shall not ‘crush’ the filament on the surface and the resulting value while paper began not to slide shall be within – 0.1~0,12 mm , so you shall substract 0,12 to lift up the hotend, say if offset is -0,25 while paper is sliding at exactly 0, you shall set -0,36.
After adjustment of H parameters in bed.g, modify the bed_report.g with same values, then reload bed_report.g on the SD card and run the macro.
If your paper report is neat, but the results of the calibration and bed_reports shows differences, that means that your bed is not flat. What count is the paper measure, not the calibration result, so do not cheat by modifying the H parameter to obtain a very low calibration value.

Note that all macros could be run after being loaded on the SD card by FTP or web interface, without rebooting or restarting.

Edited 3 time(s). Last edit at 04/06/2016 01:10PM by PRZ.

Quote
PRZ
You shall then modify the H parameters in the bed.g file and reload this file on the SD card. I find the FTP the easiest method for that.

Which H Parameters would that be? I'm a little confused. Surely this should be a Z parameter I'm adjusting?

Have a look on the RepRap wiki
RepRap_Firmware_macros#bed.g [reprap.org]

and G-code#G30:_Single_Z-Probe [reprap.org]

---

Pierre

- Safety [reprap.org]
- Embedded help system for Duet and RepRap Firmware [forums.reprap.org]
- Enclosed delta printers Lily [rouzeau.net] and Lily Big [rouzeau.net]
- OpenScad delta printer simulator [github.com]
- 3D printing on my site [www.rouzeau.net]

Quote
ByteSlinger
Which H Parameters would that be? I'm a little confused. Surely this should be a Z parameter I'm adjusting?

the H parameter is not on the default probe line, it needs adding

the best way I find to do this is to have the bed.g program open and use the web interface at the same time

first make sure the nozzle is hot, a cold nozzle can have some left over plastic on it that will deform and break off during the calibration process

now copy the first probe line of your bed.g to the Gcode box in the web interface, if there is already a H parameter on that line delete it, also dont copy the S parameter
click the go button
whilst its probing delete the P parameter from the line
when the first probe has stopped click go button again
when the probe stops it wont automatically rise, jog up 0.5mm
now using the 0.05mm button jog down until the paper just gets trapped (btw I don't use paper I use a 0.2mm feeler gauge, gives far better results)
now look at the Z value and subtract the thickness of paper or feeler gauge, use this for the H parameter on that line, but remember to change the sign, ie if you use a 0,2mm feeler Z position is 0.3 use H-0.1 if the Z position is -0.05 then use H0.25
repeat for all points

The goal is to get a deviation as low as possible without fudging the numbers, fishers are just about usable at about 0.1 deviation and print really well at 0,04 deviation, trying to get closer to 0 is an exercise in diminishing returns.

Once you have your deviation remember it and keep an eye on it, each time you home check the deviation, if it starts to creep up, or worst jumps up, then something is wrong (probably a loose ball arm etc.

---

RepRapPro Mendel 3 Tricolour
RepRapPro Fisher
-Carbon Arms
-Easy adjust Carriage+effector
-axis stiffness mods
HE3D -600 delta
-Duet 0.8.5
-PanelDue
-DC42 Height probe
-RobotDigg metal components
Simplyfy3D
RS Design Spark CAD