I am looking at what it would be like to do a large CO2 machine with, say, 4x8 bed and a 150W RECI. And then, for speed reasons, want to see if it's possible to do a fixed tube as opposed to putting the tube on the gantry.

This would be about 4M of max beam travel. The RECI specs I found was a 6mm beam with 3.1mrad divergence. This calculates to 18.4mm dia at 4m. I believe this will negatively affect the cutting consistency overall. In fact it's hard to get a final lens big enough to accept a 18.4mm beam (plus any alignment imperfection).

But from what I read, the way to collimate the beam is a beam expander, which fundamentally must trade off beam diameter for divergence. To halve the divergence, the initial beam diameter is doubled. Recalculating for like 1/4 the divergence, I'd have a 24mm initial beam diameter and 27.1mm at the furthest point.

Which... even if I used another beam expander at the head as beam reducer before the final , a 27.1mm+ lens is pretty huge.

Could I be missing something about how this works? All I saw was beam expanders. Something may be wrong here because I have worked extensively on adjusting a 4x2 laser with a 5mrad 4mm beam output- it does have a "collimator" (I believe it's a beam expander) and no beam reducer at the head. When I was doing beam alignment (prior to the output lens), the beam diameter seems to cut about 4mm dia holes across the bed. Calcs show an uncollimated 5mrad beam would become 13mm, which it's clearly not (but it has a collimator).

From what I understood about beam expanders, like I say, any improvement in beam divergence is supposed to be a tradeoff with beam diameter.

I do see a beam expander can be focused- which, I assume, would be that you'd focus it to converge at half the max X+Y travel distance, but I don't see how to calc for the beam diameters across the bed to predict how it would perform. But, again, the 4x2 with a collimator seems pretty consistent.