Quite a few PCB routers have been built that work fine with small tools like the Dremel, so my aim was to make something that would be a bit better than this, a bit quieter and hopefully have bearings that would last a bit longer, for around the same price.

I've got a couple of Dremels and the newer one seems to have around 2 thou or so of combined run out and play in the spindle, the older one is quite a bit worse. Their real problem seems to be the eccentricity in the collets, as the tool run out is significantly worse than the spindle - to get the best out of them means re-seating the collet several times to true up the tool. Despite this poor accuracy, many people have shown that they work fine for modest PCB milling, such as the homebrew stuff I'll want to do.

The very best I could hope for using ER type collets is about 1/4 thou TIR, assuming no other error sources, as that's their spec. A 1/10 thou TIR would be totally unachievable using this type of collet chuck, I'd have needed to go for something far more costly if that was my aim.

It's debatable as to whether better than a thou is needed for such an application as DIY PCB milling anyway. Minimum track width is unlikely to be less than about 15 thou and assymetric cutter tip loading isn't an issue for a single flute vee tip PCB milling tool. Given that so many people are very successfuly milling out boards with spindles that are far worse than mine is likely to be I really can't see me having a real problem, apart from the slow speed .

This spindle might be a bit too slow for PCB work with small diameter tools, as it will be limited to around 16,000 rpm. It will be ideal for the bigger mould making requirements I have, using tools up to around 6mm or so, so I may just try it for a while with that task, see how it goes, then look at making a faster spindle for PCB work if I need it.

I've found a source for an ER25 chuck with a 1/2" shaft that looks promising for a future higher speed version. Reducing the shaft size from 20mm to 1/2" gives the possibility of using bearings with a higher speed rating, which will allow me to get up to the 25,000 to 30,000 rpm (or higher) that would be best for the PCB task.

It's good fun experimenting anyway, even if I do get one or two failures along the way.

Jeremy