Nozzle Calibration

[robert@serve101.org]

I am still confused about nozzle calibration, either I am messing up or there is a bug. The pictures below will show what I have found...

Strangely, parts placed that do not need rotation, place about the same with or without nozzle calibration. But if the part needs rotation to be placed, the uncalibrated placement is much more accurate. In my case, the rotation is specified in the CSV file.

----
The latest two results are below, they show the same result. What ever is happening is consistent.


The image below shows two parts placed on the PCB with nozzle calibration enabled

Screenshot 2018-11-09 18.40.48.png (772.44 KiB) Viewed 4988 times

Calibration values below
A: 0.000, X: -0.158, Y: -0.027
A: 90.000, X: -0.326, Y: -0.145

The image below shows two parts placed on the PCB with nozzle calibration disabled

Screenshot 2018-11-09 18.43.07.png (777.72 KiB) Viewed 4988 times

[robert@serve101.org]

For now, I have given up on nozzle calibration. I cannot get it to work, it actually makes placement worse.....

[JuKu]

I'm looking into this, trying to find better ways to do and test the calibration. I found one bug so far: When testing, the probe function, that is supposed to take the nozzle down using calibration, doesn't use nozzle cal until a nozzle change or a real placement operation is done. In other words, it doesn't do calibration initialization after startup like the other operations.

I'm not going to do a bug release for this only; as said, I'm working on the whole thing anyway. In the meanwhile, if you are testing the nozzle calibration, do a force status on the nozzle change page; you'll see a green "using calibration" message on the log window.

My apologies for the troubles this has caused!

[wormball]

Strangely, parts placed that do not need rotation, place about the same with or without nozzle calibration.

This is the only right behavior. If you draw rotation of the nozzle with attached part, you will see that regardless of specific settings rotation error is strictly zero if the nozzle is not rotated, and it reaches maximum at 180 degree rotation. Moreover, in the latter case the error is twice the nozzle to camera offset error, so it provides pretty convenient way to adjust nozzle to camera offset.