Best surface for IR probe accuracy

[dc42]

@leblond said in Best surface for IR probe accuracy:

RepRapFirmware height map file v2 generated at 2018-10-21 22:56, mean error 0.222, deviation 0.150
xmin,xmax,ymin,ymax,radius,xspacing,yspacing,xnum,ynum
20.00,200.00,10.00,190.00,-1.00,56.00,56.00,4,4
0.040, 0.240, 0.070, -0.070
0.207, 0.390, 0.295, 0.072
0.370, 0.468, 0.368, 0.135
0.153, 0.343, 0.357, 0.112

I think the IR sensor is giving reasonably good readings, although it would be easier to tell if you reduced the mesh spacing to 28mm or lower.

[Leblond]

@ dc42 do you talk about mesh spacing?
             I had to make a 3x3 grid

1 to find the center of the bed

2 if I did a spacing of 20mm I had changes from -2 to +2 +/-

since I have the sensor, the only printing base that has improved and 'that arrived yesterday of the printbite, but I continue to have these changes, tonight when I come back from work I try to recalibrate it hot and post' results, if you need 'of data then PLEASE speak to you normally, unfortunately I am not an engineer with a hat like you, but a humble worker, but I'm getting better ... thanks for the understanding and good work.

[Leblond]

Good evening to everyone
Well as promised I tried to calibrate the print plate to 110 and these three images show the result.
image 1: car bed not calibrated
image 2: calibrated car bed
image 3: mesh

[Edkirk]

I have been using the heated beds from E3D, the older big box bed and recently their 120 volt high temperature bed. I painted the underside of the glass with white high temperature spray paint from McMaster Carr. initially to increase the visibility of the print over the black heated bed surface. This has been ideal with the IR Probe, I left the outer 10mm unpainted and the LED turns off when the sensor passes over the unpainted portion above the black. So the white paint does improve the sensor accuracy. Works great.

[Edkirk]

I forgot to mention the nozzle crashed into the glass when I installed the new unpainted bed.

[dc42]

@leblond said in Best surface for IR probe accuracy:

@ dc42 do you talk about mesh spacing?

Yes, I meant use a mesh with smaller spacing/more points, e.g. 7x7 or 9x9.

[dc42]

@edkirk said in Best surface for IR probe accuracy:

I painted the underside of the glass with white high temperature spray paint from McMaster Carr.

If you are using our mini differential IR probe, that's completely the wrong thing to. The correct thing (as described in the fitting instructions) is to paint the back surface matt black. This is to make the reflection from the front surface of the glass stronger than the reflection from the back.

If you are using some other type of IR sensor then this probably doesn't apply and I can't help.

[Edkirk]

I am using your mini IR probe, absolutely the best. I had painted the underside of my first E3D glass white because I didn’t like the black, couldn’t see how the first layer was going down. At that time I was using a touch prob, it never worked very well. The high temperature white spray paint is very flat or dull, goes on rough. I cure the paint by laying it on the heated bed paint side up for several hours at the highest temperature I am comfortable with. then rub the dust off with a cloth between coats. My comment is just to provide feedback on an accidental discovery that works great.

[Leblond]

Forgive me, if I have not made myself alive before.
@dc42 I'll settle you as soon as I finish my experiment and post my data.
As you well remember "let me give you some .." you asked me about the mesh, now since the aluminum base for all the efforts you make .. it expands too much in height since it is stopped in the 4 corners for leveling , so I thought to use a glass plate ultra-base and put it on the sheet of prinbite, I will try 'to calibrate cold for, and then hot starting from 100 degrees celsius, and then get to 120, according to you it will work '?, now ban to ciancie and below with the work ... a later

[Leblond]

Here I am

@dc42 What do you say? the first photo is with G32, the second one with G29 mesh 20x20

[dc42]

The only reason the first one looks better is that it has only 5 points. It looks to me that the bed is flat, so I suggest you use either no bed compensation at all, or a 2x2 mesh, or a 3x3 mesh.

[Leblond]

@dc42 Good afternoon, do you mean 20x20 and 30x30 mm? why a 2x2 or 3x3 shirt gives me error, or am I doing wrong?

[SupraGuy]

I constantly need to remind myself that the display for the mesh grid really exaggerates roughness. There is a grand total of 0.095mm total difference between the highest and lowest points in that mesh. This is pretty much invisible to the naked eye. (It appears that there is a slight tilt with the far left corner being higher than the near right corner overall.)

Chances are you'd need a machinist's straight edge to see these imperfections. There may be some measurement error in them as well, but I don't think that any kind of probe is going to be perfect. A total error of 0.1mm is enough to cause you some issues, but using a 3X3 mesh with a surface like this is almost certain to yield good results.

If I had results like this, I'd probably turn off mesh compensation, and just not bother with it. Actually, I think that my last set of results wasn't too dissimilar, and when I was having trouble with my Z probe, I did have it turned off, and I managed some reasonably large prints with a very good first layer.

[Leblond]

@SupraGuy The problem is that I've always had the printer this way and I've also had a fight with geeetech, but water passed ...
I'm printing a vase, and the first layer is practically perfect, I'm printing ABS at 220 degrees celsius and 120 layer dish 0.2.
You know after so many sleepless nights with this stamps I'm having excellent results, sial with the prinbite print plane "It had to be opaque, instead it is semi opaque almost shiny" both with the mini ir of @dc42, so that works so I leave stay the printer, in the meantime I'm organizing to build one from scratch.

[SupraGuy]

@leblond said in Best surface for IR probe accuracy:

@dc42 Good afternoon, do you mean 20x20 and 30x30 mm? why a 2x2 or 3x3 shirt gives me error, or am I doing wrong?

I think what he's talking about is to change the mesh grid definition.

@leblond said

M557 X20:200 Y10:190 S56 ; Define mesh grid

(That doesn't look like it should produce that 10 x 10 mesh though.) the mesh that would produce that is 180mm x 180mm at 20mm spacing, so
M557 X20:200 Y10:190 S20 ; Define mesh grid

I think that he was talking about changing the spacing, so that your grid doesn't have quite so many points in it.

M557 X20:200 Y10:190 S90 ; Define mesh grid
This would make a 3X3 grid, still covering the 180mm x 180mm space, but with only 9 data points, instead of 100. The resultant map would look a lot like your 5 point grid, but with data points at the halfway mark along each edge.

M557 X20:200 Y10:190 S180 ; Define mesh grid
This would only probe the 4 corners, which would give you a resultant plane.

Using fewer data points, you could have the probe repeat the process with less inconvenience, and probing would take less time.

[dc42]

@leblond said in Best surface for IR probe accuracy:

@dc42 Good afternoon, do you mean 20x20 and 30x30 mm? why a 2x2 or 3x3 shirt gives me error, or am I doing wrong?

As @SupraGuy says, I was referring to the number of probe points, not the spacing. I'm sorry that I didn't make that clear.

[Leblond]

Ok ok, now I understand, as soon as possible I will do it and post data, thanks for the answers