Diagonal 172.17, delta radius 85.74

This is what 10 point 7 factor probing is giving me. I have tried larger and smaller H factors and it is about the same.

I did have the same once.
Reseted my parameters in config.g. Shutdown everything and tried again.
But I'm using 10 points 6 factors only now.

Best I can get is: Diagonal 160.12, delta radius 82.58 with a deviation of 0.045 to 0.050

Been round and round the houses with the H factors. See H factors below:

G30 P4 X65 Y-37.5 H-0.0 Z-99999 ; Z tower
G4 P300
G30 P5 X0 Y-60 H-0.41 Z-99999 ; between Z and X towers
G4 P300
G30 P8 X32.5 Y-18.75 H-0.22 Z-99999 ; half way to Z tower
G4 P300
G30 P9 X0 Y0 Z-99999 H-0.4 S7 ; centre

The offshoot of all of that effort is I changed the G31 Z-0.3 in sys/bed.g to -0.4 to match the H factor above and added H-0.4 to the starting script where it probes the centre of the bed. Now my first layer is not squidged out all over the place. Not crashing into the part every layer either. Looks promising! Still on L160, R82.4 which is where I was before this mobius loop of 10 point probing.

Edited 1 time(s). Last edit at 10/16/2015 02:37PM by Theolodian.

My recommendation :

1. Use 6 factor calibration, not 7 factor. 7 factor calibration only works well if you are able to probe well oitside the normal print area opposite the towers. Otherwise there is insufficient data from probing, because elsewhere on the bed, a change in the configured rod length can be compensated by a change in the delta radius.

2. Set the G31 Z parameter so that the trigger height is correct at the centre of the bed. Then measure the trigger heights at the other bed probe points, and use the H parameter to provide the required corrections for those points in the bed.g file.

---

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 am back to just using 4-point probing now. I went round in one big circle and ended up where I started, but with a better understanding. The change to my starting script (G30 H-0.25) made all the difference in the world. I was wondering about that before but hadn't tried it.

Yes I should use something to directly measure the trigger heights...

To measure trigger height:

1. Place the nozzle over the XY point where you want to measure the trigger height.

2. Lower the nozzle until it just grips a sheet of paper, or whatever other mechanism you prefer to use to determine Z=0.

3. Send G92 Z0.

4. Raise the nozzle about 5mm.

5. Send G30 S-1.

6. Read off the Z height from the web interface, and that is the trigger height at that XY position.

---

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].

Take a step back, because sometimes I think people get obsessed with reducing the calibration deviation to zero.

My normal calibration is using the 6/10 routine.

My fisher tends to be in the 0.04-0.05 region, and I hit that every time in a single pass and I get great prints with excellent dimensional accuracy, they are significantly better than when it was in the 0.12-0.15 bracket but they are not discernibly better than when it was in the 0.07- 0.08 region.

This tells me that the law of diminishing returns is very much in effect.

I've also seen individuals with a deviation of zero and get crap prints.... why, simply because its possible to cheat the calibration by misusing the H parameters.

Remember, the whole point of the printer is to give you great prints, not to give you a zero deviation factor, and the true test of calibration is the printing itself, so what are your prints like?, what problems are you trying to correct?

---

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

My prints were failing, my deviation wouldn't go below 0.125 with a 6/10 calibration, my extruder was stripping the filament.

As stated before the whole exercise was a learning experience about H factors but made no direct difference otherwise. Adding the H factor to my starting script is what made the difference. All the calibration in the world won't help if your starting script then throws it out the window.

Quote
Theolodian
My prints were failing, my deviation wouldn't go below 0.125 with a 6/10 calibration, my extruder was stripping the filament.

As stated before the whole exercise was a learning experience about H factors but made no direct difference otherwise. Adding the H factor to my starting script is what made the difference. All the calibration in the world won't help if your starting script then throws it out the window.

OK, that's different from not being able to get better than 0.04, 0.125 is to high, but if you getting bad prints at 0.04 then your problem is more than a bad deviation factor.

The important thing to understand about H parameters is that they are required because it takes 3x as much force to displace the bed switch at the centre of the plate as it does when directly over one of the springs, this force has a bending effect on various parts of the machine which causes inconsistency in the probing accuracy, the process DC details above measures this inconsistency and the H factor compensates for it, consequently you should notice that your H factors increase as you are further away from the spings and closer to the centre.

0.05 is a good target, because the step resolution of the Fisher console is 0.05, and its actually fairly hard to measure H factors with greater than 0.05 accuracy.

---

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

I did learn that the H factors should be larger at the edge of the bed between the towers than at the centre. Makes sense considering the lever arm.

Yes, I am happy with a deviation of 0.05 which is where I am now.

Edit: As inferred from the title of this thread my main concern was getting L and R values that were closer to reality. I suspect that the H-0.25 I am now using in my starting script is within 50 microns of ideal to cover the centre probing point including the mushiness of the Buildtak. The little Make robot prints well at this setting. The other 150 microns is probably the warp in my acrylic bed. This suggests that for larger parts I will need to up the H factor in my starting script by 50 microns per 25mm radius to cover the warp in my bed.

Edited 1 time(s). Last edit at 10/17/2015 04:17AM by Theolodian.

Sorry to jump in on your thread but my son just got his first 3D printer and we have made a couple of prints which are "ok".

I think we need to calibrate the printer however.

I'm a software engineer and fairly technical but I'm at a bit of a loss as to where do I start to understand what it is you are all talking about? lol

Quote

This is what 10 point 7 factor probing is giving me. I have tried larger and smaller H factors and it is about the same. confused smiley

Is there a good reference page on-line to start with? What are H factors? What is the 6/10 routine? etc. etc...

TIA

Very simply....

There are a number of bed probing routines, the most basic is the 4/1 which probes 4 points and adjusts one factor (I think Height)

The rest probe more points and adjust more factors, the 'best' one to use is the 10/6, there is also a 10/7 but this isnt so good to use.

The fisher probing works by displacement of the bed which causes a switch to break, but its an imperfect system because as I stated earlier it takes 3x as much force to break the switch at the centre as it does above one of the springs, this force causes defection in the arms, belts etc and means that the probing accuracy changes across the plate, to compensate for this we use H factors on the G30 command, which effectively measures the inaccuracy and helps the compensation routine.

In detail..... let me look up the link.....

here....

[reprap.org]

if you read up a few messages DC gives the basic process for setting the H factors

Edited 5 time(s). Last edit at 10/17/2015 07:25AM by bgkdavis.

---

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

I really wouldn't worry about H factors except for the starting script in your slicing profile. The normal 'Home All' and 'Auto Bed Compensation' routines on the RRP SD card image are fine, you don't need to worry about 7 or 10 point calibration really. Run Home All, then Auto Bed compensation 3 times, then copy output of M665 and M666 to your sys/config.g per normal. Then adjust your start height in the starting script of your slicer profile. In my starting gcode script I set what was just G30 to G30 H-0.4 and then worked down until the first layer was solid instead of full of gaps. I am very happy at G30 H-0.25 but your results may vary.

Do try to bend your acrylic bed as flat as you can by hand and then check again regularly. Also be sure to calibrate your steps/mm on the extruder. For the Beta most of us ended up with M92 E137 in our sys/config.g whereas RRP started with a much higher number. On the eSun filament start with 200C extrusion temperature, anything else temperature and filament diameter may vary.

Thanks peeps. Helpful stuff!

OK, made a bit more progress. I was doing the bed leveling with the nozzle cold because that is what BuildTak advise. However, my starting script was probing the centre with the nozzle hot. Print failed with H-0.25 when I changed this probing to be before heating the nozzle. H-0.3 prints but probably still a bit low. May end up closer to H-0.4 after all.

EDIT: and make sure that the nozzle is clean before each print or it really throws it off.

Edited 1 time(s). Last edit at 10/17/2015 12:47PM by Theolodian.

Quote
Theolodian
EDIT: and make sure that the nozzle is clean before each print or it really throws it off.

I agree, it makes a big difference !