I am wondering what are the down sides to just masking a laser to get the beam diameter you require? For example if you have a collimated eliptical beam of say 2mm width and 4mm height, but you required a circular beam of 1.5mm, is it acceptable to just place an 1.5mm aperture? I am thinking that it is not an efficent use of the laser (as a lot of the power of the laser would be lost at the aperture), possible diffraction at the aperture, and loss of the lasers "Gaussian profile". Are there any I am missing, that would need to be taken into account on practical laser systems.
Depending on wavelength, just with an aperture diffraction will kick in few cm after the aperture, and you'll get rings opening up with further propagation. For some applications that could be ok, especially if you use the beam right after the aperture.
Practically, the way I would change the beam shape from elliptical 2x4mm to circular 1.5mm is putting in place a spatial filter: a 2/1.5 telescope with a pinhole inside. Basically you need two lenses (positive focal f1 and f2) placed at a distance f1+f2. The beam enters and exits collimated. Choose f1 and f2 in order to decrease the 2mm of the ellipse to 1.5mm. Otherwise simply use f1=f2, it may be ok. Between them, at f1 from the first lens and f2 from the second, you place the pinhole (not trivial to align). Check the theory, but test some pinhole diameters (probably around 100 micron). Other than the right circular shape, you get a cleaner transverse profile (less high frequency spatial noise).
Another way could be to make again a telescope (no pinhole) but with two cylindrical lenses, in order to focus only the direction where the ellipse is 4mm down to 2mm, to get a circular 2mm diameter (if that is close enough to the 1.5mm required).
I'm not an optics guy, but I can think of a few issues you'd want to think about until a pro comes a long and explains it all to us:
The beam will diffract from the new aperture affecting both collimation and coherence length. Only a concern for very small beam diameters or long optical paths; and something that any other method for reducing the beam diameter at more than a point will also have to contend with.
You'll want to control any reflections from the collimator.
You should a Back-of-the-Envelope (or at least back-of-your-eyelids) calculation of how much energy will be absorbed on your collimator. Is that a problem? Do you need to cool the collimator? This may involve a tradeoff with the reflection issue.