This is going to be a long post, so pls bear with me.
But before someone reaches the real questions, I will post it here: how should I approach the buying of a new CO2 laser cutter what too look for to achieve the results I need?

I have been going through a lot of thinking the last couple of months about how to setup the production for the near and mid future. In short I have been running my production site with a cnc router (about 1600*1600mm) with relatively good results, cutting kind of aluminium composite and making parts for mass production we assemble them later.
Now the new product that I launched and that looks very promising in the market is made of (frosted) acrylic (5mm thick). The main focus in the parts I have to machine is on the tolerances in the parts as they have to overlap perfectly (+/- 0.1mm). As I use frosted acrylic, the edge is not that relevant, so the router provides fair finish on the edge – except for the final skin which I have to keep for vacuum and that I have to manually remove as some parts are small (100*15mm).
Also important to mention that the parts have also pockets – so a router must be used at some point, a laser cutter cannot do the job alone.
Also, some holes should be quite precise (tolerance +/- 0.03mm) to fit metal parts – this can be though adjusted manually with a file, so good to have but not that critical.

THE PROBLEM I have now is that even though I use closed loop motors (leadshine) and set them to 0.3mm error alarm trigger, I still get parts that have0.5mm differences in size. Besides the 0.3mm error allowance I mentioned, I can “blame” the usual suspects:
- Backlash – I measured an 0.04mm on each axes, compensated in the software (UCCNC), but understand that it can still be there
- C7 grade ballscrews – on a 1000*1000mm I can still get further differences due to screw tolerances
- End bearings – this check is smth I did not do in a while – it is only now when writing this that I remember it

Other than that I use 6mm acrylic Belin (amazing!) carbide bits, strong vacuum (as long as there is left a skin), powerful leadshine motors, good spindle, rigid frame (Bosch Aluminium frame) etc. So there must be a little devil somewhere I overlook.

I know that if I put more effort in the process I can decrease the total error:
- Solve the backlash
- Run slower accelerations

I will be still left with issues like:
- Keeping occupied the router – I need extra time to fit the new product
- Hand finish the skin of acrylic and the size of each part after assembling them – very slow

With a CO2 laser I (hope I) can get the parts I need with a better tolerance. After laser cutting the contours, I use a stiff small (4030) router with dedicated nesting and I machine the acrylic parts for pockets.
Question is:
- Can I really get parts that are in the 0.1mm tolerance with a laser? ( within the same batch – same acrylic, same machine settings)
- If I buy a Chinese 1612 or so CO2 laser can I expect to achieve these results? Note (1): a non Chinese laser (so spend more than 10k) would not justify the investment; Note (2): if I use C5 ballscrews instead of timing belt ; Note (3): I do not intend to use the laser for engraving, so high speed is not relevant
I read quite a bit about what is really relevant in the CO2 laser to obtain better precision, and for example stiff frame much more important than C5 ballscrews.

I should stop here as I already made it too long. Hope it is still readable. Also hope it could have a business side of it, relevant I think to many people here.
All in all, recommendations on the laser specs, pros and cons to move from router to laser etc are more than welcome.

Thanks
Marius