Alright. Finally got back to address my issues on the z-axis. I believe my main issues were the router itself, I got one from the bottom of the QA heap.
I had the bungie cord with preload on there router but I was still experiencing issues. In the end I think the bungie was hurting the performance, I needed something I could more accurately reproduce over and over. More on that later.
In the end I did a couple things that really have made it rock solid. First thing I did was check to see how much I was losing to “slop” then figure out where that slop was coming from. So I used the z axis function in GC to first move the router up about 1/4" then in increments of 1/32 I nudged the router down until it actually started moving down. Then I did the reverse, nudge it until it started moving up. It ended up being an average of about 3/32 of “loss” in each change of direction. That starts stacking up and led to me never being able to get through a 1/2" peice of wood.
So, seeing this I looked for the source of the “loss”. I had 2 issues, the threaded shaft had up and down play in it and the router lock tab would actually twist before applying the force to move the router up and down.
To fix the shaft play, I removed the e-clip and took a look at the shadt and the bearing surfaces. Because the router isn’t expected to used in this kind of configuration the steel threaded shaft just rides on the surface of the magnesium router base. There is a very slight ridge that I believe was supposed to hold the threads up off the router but I can’t see this lasting long. To remedy this I took a washer with the right thickness, drilled it out to fit around the shaft and placed it between the threads and the router base. This took out all of the up down shaft play. Success!
The router lock tab was a little more difficult. It is clear that we could bore out that lock tab and place a flange nut inside the tab itself however, that removes the mechanical fuse of the orange spring tab. The main issue we have is the twist due to the small load bearing areas made of a soft metal. I already had signs of wear on my locking tab.
The solution I came up with was pretty simple and still allows me to get plenty of Z travel to get through a 3/4" sheet of plywood. For this I included some pictures to explain the modification. It keeps the orange mechanical fuse and costs about $2.
While I was doing this I spent some time with a red Scotch Brite pad to polish up the inside of the router base. I noticed a slight flaw in the casting, in the anti-rotation groove about halfway down there was a ridge that looked a but like a parting line. I could see it was catching because it was nice and shiny unlike the rest of the groove. A little work with a thin file and I was able to flatten the whole slot down. No more catching! As a finishing touch I took 3 strips of 0.010 UHMW tape and put them on the router body and wiped the inside of the router base with Slipit (great stuff BTW, I am using it on my linkage kit as well).
The final step was to apply a consistent but repeatable amount of preload on the router. I left the 2 router handle bolts on place and then used #64 rubber bands hooked to a 3 inch S ring (stole that idea from someone else. In the end 2 rubber bands per side is more than. Enough to get repeatable, smooth, no lost motion, z-axis goodness.
Sorry for the length of my post, just thought it would be good to explain all the changes and to mention that it seems the variation in the quality of the router base may have made it much more difficult for me to get repeatable results.