It has been very interesting reading about diffraction and electron beams but it is quite clear that we often give the instrument credit, "the instrument does it automatically", when credit is not due.  Almost every "automatic" is a designers estimate or better still a calculation, it is a setting!

After 40 years with TEM I have to say that the only area that the instrument could set in the SA mode could be the approximate position of the first image plane. We normally move to the SA mode to free up the diffraction lens (1st intermediate) to enable it to be focussed on the diffraction (intermediate) aperture independent of the magnification system. We would then focus the image inside the aperture with the objective lens to ensure that the selected area IS the area in the diffraction pattern. Without this procedure electron rotation and misalignments could give you a diffraction pattern from an area not in the centre of the screen!   Users are correct in that if they set the eucentric point or better still, adjust the specimen height to focus at the same objective lens current value, that done the first image plane will be at the same point within the diffraction (intermediate) lens.

So how could the Philips work? I would guess they set the diffraction lens at approximately the first image plane. Then as they do not use it within their lens scheme for the SA magnifications it is fixed at this focus, good enough I think?

A second point, parallel beams are very important if you are chasing the best image quality, biology or materials, most times we need high coherence. There are misunderstandings on how to obtain a parallel beam or high coherence, setting the final condenser underfocus is incorrect. The procedure for high coherence would be to use the smallest spot size you can tolerate (this probably means you must up the emission current). Once in this condition overfocus the final condenser (clockwise from crossover) the spot, what ever size it is now, becomes your virtual source. The further overfocus you go the greater the beam coherence and that is what we are after. You will deduce the smaller the condenser aperture the sharper the spot and the smaller the spot the greater the coherence for a given degree of overfocus.

Work with a design team and they expect everyone to overfocus, they expect everyone to use small condenser apertures and they expect everyone to use small spot sizes. They do not expect everyone to use too low an emission current (almost everyone does, we have talked before about filament life being the most important feature of many laboratories) because this makes the task too difficult!

There must be people more knowledgeable than I who will cut out all the guesswork?