OK, so I finished testing the custom 1/8" collet, and I'm pretty impressed with its performance relative to the set-screw collet adapter I had. Here's a picture of the piece of wood I tested on:

- IMG_2633.jpg (163.8 KiB) Viewed 4568 times
And here's a piece of PVC sheet that I tested on as well:

- IMG_2626.jpg (102.67 KiB) Viewed 4568 times
Some summary observations:
- On the whole, the custom collet had less than half the runout of the 1/4" collet with a set-screw collet adapter. In a number cases the runout was considerably better, approaching 10x less runout.
- Oddly, with the stock 1/4" collet and adapter, my spiral up-cut bit burned the wood. You can see it in the picture if you look at tracks 2 and 5 from the upper left. After that, I did a bunch of extra tracks on the right edge of the wood to try to reproduce the burning with the custom 1/8" collet, but was unable to.
- With all the tracks cut with the stock 1/4" collet and adapter it seemed that more force was required to move the bit through the material at the same rate as the 1/8" custom collet. I assume that relates back to the burning mentioned in the previous item, but that result surprised me. Can anyone explain why worse runout would require extra force or cause wood burning? Or was it maybe not a problem with runout, but the set-screw adapter not sinking heat as well as the custom collet?
- The fact that the flat two-flute bits wouldn't slide all the way into the custom collet did result in higher run-out numbers than for the spiral bit, which slid all the way through. Nevertheless, the runout numbers were still considerably less than with the stock 1/4" collet and the set-screw adapter.
I made a spreadsheet of the measurements, and defined some formulas to determine the runout. So that none of you need to own Excel or Numbers to view the numbers, I have included it as a PDF, which should be viewable by everyone:
Let me explain the methodology used to test the collets. I'm sure the experienced machinists among us will find many faults with this methodology, but it was the best I could do in one afternoon, so I apologize in advance for any shortcomings in my data.
- Since my Shapeoko is still under construction, I hand routed the tracks with the RT0701C base, run along a straight edge.
- I routed test tracks and some holes in a piece of scrap Poplar and a sheet of PVC plastic just to test on different materials.
- I cut tracks with a two-flute straight bit and a a single flute spiral up-cut bit. Why those bits? Because that's what I had in my Inventables starter bit set.
With those bits, I cut one track for each bit with the custom 1/8" collet, and one with the stock 1/4" collet plus a set-screw collet adapter. (As I noted above, the extra tracks on the right side of the wood are where I did additional testing for burning.)
- I used a digital caliper to measure the bits at the tip, taking three measurements to determine an average.
- Then, I used the digital caliper to measure the resulting tracks. I took three measurements at various places within the middle third of the track. This seemed to be the most consistent part of the tracks, presumably because I had the router moving at a fairly constant speed with good pressure against the straight edge by that point.
- Track measurements were then averaged and subtracted from the average bit diameter values to determine an average runout value for each bit.
- Finally, I also clamped the RT0701C router to my workbench and used my XY table to drill a hole in the wood. I then compared this hole against a hole created with my drill press, which has a pretty low runout. I was pleasantly surprised how well the clamped RT0701C with the custom collet fared against the drill press.
The numeric results of the testing are shown in the spreadsheet PDF, and summarized above.
Because of the testing, I was unable to start mailing collets today. Furthermore, I need to head into the office tomorrow, so I don't think I'll be able to get the collets out in the mail Thursday... It will have to wait until Friday when I can work from home again. I know after seeing the numbers, you're probably all anxious to get your collets, so I apologize for the added delay. I'll let you know as soon as I have sent yours out to you via PM.
If you have any questions about the testing, please let me know.
Peter.
Shapeoko 2 #5059. Assembly in progress...