BoXZY, a 4-micron CNC mill, 3D printer and laser engraver.

[Warhawk]

I don't see much point making it do PCB track routing. Results are never good and you don't get PTH or mask or silkscreen.
I would much rather design the PCB and wait the 2 weeks.

PCB routing is utterly stupid but etching your own PCBs is simetimes the only viable solution when you do not have a gold shitting donkey.
The keyword is turn around time and prototyping.

I can spend 80 Euros and will have the project finished in 1.5 months.
I can spend 30 Euros and be done in 3 months.
I can spend 150 Euros and be done in two weeks.

If i do it at home i will have the finished PCB in 3 weeks at 40 Euros or 6 weeks at 15 euros.

And if i do not need soldermask and double sidet i will be done in one day and 5 euros in.

Personal experience, real live numbers, no bullshit. Your reality may differ.

Aisler makes decent PCBs with 2 days turnaround. Shipped for free from Germany. Why would anyone in (not only) Germany bothered with doing PCBs at home?
Believe or not, I used to make a lot of DIY PCBs. Including the solder mask. Nowadays, It does not make any financial sense. I submit my data on Friday and get the PCB at my desk on Thursday.
6PCBs 3x2cm for 14 eur. I paid more for Steckerlfisch today. 

[Mechatrommer]

Aisler makes decent PCBs with 2 days turnaround. Shipped for free from Germany. Why would anyone in (not only) Germany bothered with doing PCBs at home?
Believe or not, I used to make a lot of DIY PCBs. Including the solder mask. Nowadays, It does not make any financial sense. I submit my data on Friday and get the PCB at my desk on Thursday.
6PCBs 3x2cm for 14 eur. I paid more for Steckerlfisch today. 

Aisler probably did not exist in 2015... and 3x2cm for EUR14 is still some price, if some mistake/correction, they are good EUR2 keychains.

[wizard69]

And that is what will cause this project to go off the rails. 

People think "XYZ motion is easy!  I can make...

Actually the motion these days is easy.   It is getting the rest of the machine to cooperate that is hard.

-PCB miller
-Pick and place
-CNC milling machine
-3D printer
-Laser engraver
-Waterjet (now)
-Router
-Engraver

etc, etc, etc.

But all those devices are just different enough such that when you try to combine them, the results suck. 

Yes, the problem here is that people don't understand the mechanical implication.   Somebody posted pictures above that highlight just how bad this machine is.

"CNC milling" with a freakin' $99 Makita palm-sized router?   No wonder they mention 0.050" depth of cut in brass (and the finish looks like shit).

I'm not event sure this machine would be good enough for front panel engraving of say 0.010".   It is surprising how well the human eye can pick up minute errors in a tool path.

As a side note I spent years of my career working on custom diamond turning machines and there we had to keep tolerances sometimes under 2 microns.   That isn't easy if the machine is running a 10 second cycle time 24 hours a day.

It is interesting because I believe the basic idea of this style machine came from some kids out of MIT.   The problem is that they apparently missed some basic machine design concepts such as the need for stiffness in any machine that has to deal with reaction forces.

If anybody is thinking about buying a small router or making one, the site: CNCZONE has a great section dedicate to building such machines.   There are ways to DIY a small router that doesn't result in complete crap.

[Striker]

Even if it was only a PCB mill I'd avoid it, let alone a combination of 5 other things. As part of a makerspace I ran an LPKF PCB mill for awhile, an expensive one-function PCB mill with all the whistles, and even that gave fairly inconsistent results that I could rarely justify using it. It costs $20 with shipping to get 10 PCBs from China in a week and a half. If you're not using unusual or expensive materials (Rogers for RF, etc) or doing strange antenna patterns that cheap fabs couldn't do, it just didn't make sense.

[Spirit532]

This is definitely a case of a mechanical engineering dropout(almost certainly due to it being too hard) picking up a couple ballscrews and linear rods, realizing they are heavier than plastic, and thinking: "Well these seem rigid enough, why can't I mill with it?".
What the dropout doesn't realize is that cutting forces are not only orders of magnitude higher than he thinks, but all components always imperfect and on top of rigidity issues you also get accuracy/backlash/tolerance issues.

The way real 4 micron machines do it is as follows:

• Get some cast iron or machine-specific epoxy granite
• Triple that amount
• No, triple it again
• Bolt that to a 20cm thick reinforced concrete pad
• Embed a bunch of channels for heat transferring fluid into the base and all large moving parts
• Attach a high heat capacity chiller-heater combo to keep the entire machine at a set point(usually 1-2C lower than ambient)
• Place very oversized, over-rated components on all linear axes(ballscrews, linear rails, all rated at several hundred kN higher than your estimated loads)
• Preload the hell out of them(impossible without rigidity)
• Place linear encoders directly on the translating frames, not at the input(which already has encoders, because you're using very well tuned AC servos)
• Spend a few months in R&D tuning the several control loops
• Design, build, and install a zero-tolerance preloaded spindle(which results in ~1-2um of tool TIR, never perfect)

Congratulations. Now your machine starts at $1 000 000 with no options like tool carousels, work piece pallet systems, etc. You'll want those, too, because you just built a machine that needs to be running 24/7/365 to pay for itself in just a few short years, and downtime means bleeding money.

I'm willing to bet $100k of their money that you could cause 100um of deflection by pressing on the collet end with your pinkie finger.

A few other problems:

• It's too slow for 3D printing. You need fast retracts and rapid moves for 3D printed parts to look clean - even for the simple reason of ooze, which is now a standard feed-forward calculation option in most slicers. Also part cooling. Not to mention printing slow means you're... printing slow, and any cheap $200 kit can print faster and cleaner.
• I don't see any protection for that 2W laser, which on top of being multimode and hard to focus, as mentioned earlier, is extremely dangerous optically. So on top of being near useless(maybe for burning letters into leather or something), a single unlucky reflection of a slightly shiny surface can result in permanent, irreparable blindness. The reason cheap CO2 laser cutters get away with large transparent windows is simply because polycarbonate is opaque to 10.6um light.

[Cerebus]

I'm willing to bet $100k of their money that you could cause 100um of deflection by pressing on the collet end with your pinkie finger.

While I agree with you about the likely amount of deflection, it is traditional to put your money where your mouth is, not someone else's. 

[Spirit532]

While I agree with you about the likely amount of deflection, it is traditional to put your money where your mouth is, not someone else's. 

By the time the campaign ends neither of us is going to have that much money

[josehenry1800]

I totally agree with you. this is not a bad.