When my old employer first started in SMD our Eng lab bought a LPKF mill. It was great but noisy, messy and too slow. We ended up ditching the mill and spending more time on peer reviewing and allowing 2 protoytpe runs in our development schedule. Also got a agreement in place with PCQ for fast proto turn around.
That was our experience as a professional eng group in a large national electronics company, in a small shop, it will be difference. I would definitely buy one now if I could get the right customers.
My CNC machines were all large "full size" machines, not for PCB - though making cabinets or PCB's is really pretty much the same process, usually just with differences in power, accuracy and overall size. There are a ton of solutions to build smaller PCB machines.
I mill my own proto types. I have a sherline machine with a ball screw conversion. Using mach3, eagle, and pcb gcode plugin. I thought i could save my self some money building all my own stuff if i had it to do all over again i would have just bought a ready to go accurate machine like this http://www.flashcutcnc.com/7300_prototyping_mill.php. with a highspeed spindle and automatic tool change. the machine i have put together has a resolution of .00005 per step. It was able to do a 16 pin QFN. but you would be better off with one that is built for the job.
I think I'd have to agree with Rob - you can spend some serious times futzing with getting mach3, your CAM software and all the hardware working. For the little extra it costs for an out of the box solution is worth it if you really just want it to work and are not in it for the "challange".
I've got enough "challenges" in my life that i'll certainly be looking to buy something that is ready to run, software and hardware.
In my research in the last couple of days, i've been fortunate to find that the local polytech has a set up ( and some other rapid prototyping tools ), and that they have a program to work with local industry.. Excellent! Talked to the guy this morning, and they are really happy to help me learn some stuff about CNC milling, prior to going and buying something.. This is a perfect scenerio for me, both expertise and machinery..
I will almost certainly buy something in the near future, but i think that this is going to make the learning curve a lot easier, and hopefully a lot less risky in terms of mistakes.
I think it works just fine. I use it fairly often for prototype work.
In my mind, a prototype is something you make quickly and far from perfect. You can make very accurate boards if you want to put in a lot of time. The way that youâ€™re suppose to do this is make a first pass with a pointed bit and then use end mills in increasing sizes to remove every nook and cranny.
What I do is first always use a ground plane. I typically use a .015 or .02 inch clearance. Then I just use the pointed bit and down into the fiberglass far enough to get my clearance width. 90% is typically routing just the ground plane. The bit stays down the maximum time this way. A typical board is less then an hour, start to finish.
This way I use only one bit to route the board. I use a .032 router bit to contour drill all of the holes. Then I use a .062 router to cut the board out.
If you do it the instructed way, it could take up to 8 or so hours. Itâ€™s a very nice looking board but what a waste of time and wear on the mill!