After we had some experience with the topic, I decided to write a short note. After several uses of the optical tube, we identified several problems:
1. The quality of the optical system is far from ideal. The light source is not collimated and is not aligned with the lens. All such problems lead to the fact that the outline of the insert is not very clear.
2. Since we buy cheap Chinese diamond inserts, we are dealing with cheap Chinese diamond inserts
Their radial runout goes beyond all the boundaries of good and evil, and we preliminary sort them according to this indicator in order to install inserts with + - the same radial runout into the cutter body. But sorting occurs using an indicator (it rests on the ground carbide part of the insert) - therefore, the radial runout of the corners of the insert turns out to be very large, 10-20 µm.
Due to the fact that the optical tube is round, the camera is round and everything is round, the position of the X and Y axes on the monitor can be any relative to the machine table. After a while, I understand that it was necessary to move the table with the optical tube in a direction perpendicular to the optical axis - and gradually make a line on the screen parallel to this movement. This would be a line parallel to the end of the cutter. It's a pity that I only thought of this now
Usually the process of setting the teeth occurs without my presence, my technologist is guided by my instructions, and at this time I am still sleeping)))) Probably, I could come up with something on the spot.
3. Continuation of point 1 - the quality of the optics is far from ideal, the camera is far from ideal, and as a result of this there is additional blurring of the image.
In general, the last few times we aligned the teeth using the contact method. Below is a photo. The measuring sensor is a linear sensor with a division value of 0.1 µm (3.9e-6 inches). On its usual tip, which screws into the M3 thread (I dropped anaerobic glue onto the thread), I attached a plastic nozzle with a very tight fit. On the plastic nozzle, a round boss with a diameter of about 10 mm (0.4 inches) was milled and two holes were drilled into which self-tapping screws were screwed.
The sensor is fixed into a regular magnetic stand for the indicator. Its vertical position is set by eye
After that, we hang two nuts weighing about 25g each on the self-tapping screws - so that the sensor is in the lowest possible position. Also, using one of the screws, the tip is twisted in one direction using an elastic band. There is a mistake in the photo, the elastic should be on a different screw. This is necessary to sample rotational play in the sensor. As a result, sensor play in all directions has been eliminated. After this, the top plane of the plastic boss is machined with a cutter with a diameter of 12mm (0.5 inch) with a cutting depth of about 0.05mm (0.002 inch), of course in several passes.
As a result, we have a sensor with a plastic plane strictly perpendicular to the spindle rotation axis. That's all! We have a boss with a diameter of 10 mm, into which we can safely rest the diamond insert. Since the plane is perpendicular to the axis of rotation of the spindle, a small (within a few tenths of a millimeter) fluctuation in the position of the insert angle, both along the radius of the cutter and along the circumference, does not matter.
After the introduction of this method, aligning plates began to take literally several hours - versus a whole day with an optical tube. At the same time, the alignment accuracy increased by 2-3 times. Of course, we act in several iterations, consistently reducing the end runout. The result is about 0.7-1 microns.