Large all-metal Kossel Delta
-
Well I never thought I'd see this:
16:15:37Stopped at height -0.150 mm
16:15:31G32
Calibrated 8 factors using 10 points, deviation before 2.630 after 0.001
16:15:00G28
16:07:55G10 P0 S130
16:07:52M140 S85
16:07:49Connection established!
16:07:45Page Load complete!That's only probing the central 100mm of my kossel XL bed, with my piezo probe using an experimental build surface on glass, but deviation 0.001 - happy with that would be one way of putting it.
-
(thumbs up emoji)!
-
@DjDemonD Can you try 20-30 random points instead of 10?
-
I can but is there a quick way to do this? or are you defining the 30 random points manually? If so I really don't have the time.
-
I'm using simple PHP script to generate random points inside bed. If you are familiar with php I can post script, if not - I can post bed probe g-code file for your bed dimensions.
-
Sure Ill try it, if you can post a bed.g file. I am using a 100mm diameter (I have a 300mm diameter bed but I often only probe the inner 100mm for printing small objects. I can't quite get down to the 0.001 levels with the 300mm diameter.
-
Please add Sx (x = number of factors) to latest line you want to use:
For 100mm bed diameter
For 150mm bed diameter
For 200mm bed diameter
For 300mm bed diameter -
Okay thanks I'll run some of these later. What sort of results have you seen so far?
-
You are using 10-point 8-factor calibration.
I just think that the smaller the number of points for calibration the easier it is to get perfect results. For example, using 3-point 8-factor calibration you will get perfect results with any sensor.
That's why I asked about 20-30 points.
-
100mm Run 1
18:56:18 G32
Calibrated 8 factors using 64 points, deviation before 0.089 after 0.011Run 2
18:58:43 G32
Calibrated 8 factors using 64 points, deviation before 0.026 after 0.007Run 3
19:00:51 G32
Calibrated 8 factors using 64 points, deviation before 0.010 after 0.009So its an order of magnitude less precise when probing 64 points rather than 10. However its still achieving at best 0.007mm deviation, so its still a very accurate machine (should be cost me enough money, and even more time) plus the piezo probe is very accurate and repeatable.
Using AndOrNot experimental build surface, e3dv6 and Precision Piezo 27mm piezo sensor in Lykle/Zesty design assembly.
-
Thank you, excellent results!
It seems for me that with such accurate build M665/M666 command output is never changing between calibrations.
-
Thank you. Yes it barely changes, I still run a calibration before each print and it usually trims the values slightly, but if this were my micro delta (https://www.thingiverse.com/thing:1874766) which I calibrate manually using Escher 3D least squares, I would calibrate it automatically once and then only return when I'd changed something, so I have never bothered to fit a z-probe to it. But on the 3 times larger kossel XL machine the effect of thermal expansion on the frame and bed might be marginal from one day to the next, but its just significant enough to be worth calibrating it frequently. Its also a good basic check if anything is amiss.
It would be quite nice to be able to code into the bed.g file , if deviation >=0.03 then halt. Or something like that.
This machine has been a nice test bed for the piezo sensor - I can rely on the printer to be accurate therefore I can evaluate one sensor design/type against another. I built a test rig but it was, as expected with printed parts and smooth rods, not terribly accurate. I was using it just to check a module I had built worked before shipping, but it broke down, and I'm so familiar with them now a few taps on the bench, and a shake to check its not too sensitive, is enough to see if they are working or not.
I was thinking about raising the hot end on the kossel XL as it currently hangs down below the effector, which might push things even closer to perfect, but its working so well I might refuse this one, for fear of ruining it, just chasing those last few thousandths of a millimetre, the path to insanity (or physics).
-
Here is this machine now with prototype enclosure, smart effector and filament monitor. Still to fit accessory chamber heater, and chamber cooling fan.
The extra heater monitor channels provided on the expansion connector are great and I can monitor temperature at the bottom and top of the chamber, to control heaters/fans to maintain the ideal temperature. Water cooling for hot end being considered.
