I think Andrew's suggestion of using a static mixer to mix the epoxy and then adding it to the pot and mixing it with the aggregate under vacuum is a good suggestion in practice. My main concern is that most of the commerical metering pumps with manifolds and cascade mixers on the end for epoxy that I've seen are almost as expensive as a surplus vacuum pump. Grebeard's suggestion of DWV pipe for the vacuum chamber is ingenious.

See http://www.michaelengineering.com/ for example costs on static mixing epoxy dispensers.

The 1975 paper by gamski said that setting under pressure was not effective in removing air voids but that it helped maintain good aggregate distribution and produced stronger samples. It's my opinion that bagged setting under pressure will do better for keeping air out of the cast and good aggregate distribution than vibratory compaction unless thixotropes are used to help maintain the spatial aggregate distribution and they require vibration.

I'm getting the impression that a carefully poured mixture that isn't too viscous and was properly deaired will be able to be poured into a mold and work ok though vacuum bagging and pressurizing ought to work better.

The biggest problem with air that I see is that it doesn't just sit around forming voids in the matrix. The air is concentrated in pockets around the aggregate effectively reducing the mix to pure epoxy for strength and modulus.

I agree with Andrew that the commercial samples Walter got are a bit sparse. I recall walter saying that they were from a machine base which I'd guess probably didn't need the high test stuff. To me, they appear to have only a few large voids around the edges which would indicate they were made from vacuum deaired epoxy poured into a mold at atmospheric pressure.

The distribution of aggregate uniformly throughout the sample ought to be random. When it isn't: that's when there's a problem. The Machine Design article way back written by folks from accures castings indicates that accures uses accurately graded aggregates to ensure the proper size distribution.

Finally, I'll have to politely disagree with Walter about steel tube generally being bad. While light gage steel tube, the stuff you'd voluntarily cut with a hacksaw, is going to be tough sledding for machine building, heavy wall stuff or I beam is likely infinitely easier to get to perform well than E/G. Most of this could be put together by bolted joints with epoxy grout in them so they stay square (this is the E/G thread).

I'd guess that with an oxy-acetylene welding cutting setup, an angle grinder, a hack saw, a drill with lots of sharp metal drilling bits, a big square, epoxy grout, and some stuff like 80/20 channel for the precision parts one could build a very nice machine. I'd also go out on a limb and state that such a brute force approach might be less difficult and less expensive than E/G. The secret is that none of the metalwork is precision work: it's just holding the precision parts up. I'd also suspect that the tools to make the metal one will ultimately be cheaper and more generally applicable than the tools for E/G.

My neighbor is a composite materials expert for Boeing on the international space station. He said to me the other day that if composites had been invented first that everybody would be rushing to aluminum as the miracle material for most applications. He recommended aluminum filled epoxy as used by industry for "soft" dies in forming operations for what we are doing.

In short, I like E/G for the reasons that Walter does: it's theoretically inexpensive and easily formed with wood molds into complex shapes that home shop types couldn't form without extensive metal cutting tools. On the downside however, it is apt to require relatively bizarre equipment like vacuum pumps and because E/G isn't particularly strong, it will require heavily engineered parts to avoid metal reinforcement.

If all a person wants to do is make a big CNC router, I'd think that several other cast materials besides E/G will be less work and have a more predictable good outcome. Metal dust filled epoxy and metal whisker filled epoxy like some of the stuff in the PTM-W catlog might be a lot easier if they aren't prohibitively expensive. Using E/G for a router, it will be a lot easier to reinforce with rebar where necessary and accept the theoretical few percent vibration increase than to design metal free parts. My own interest is to create an open E/G design that will work without reinforcement for machine applications but it doesn't mean I think that this is the easiest way. It is however the most interesting way and the road less traveled. It's probably also the cheapest way for a <u>mass produced</u> part.

Hope I've been playing devil's advocate here and not the devil,
Cameron