Hi Chris,
Sorry for the out of sequence reply – I read Ted’s question first, but I 
definitely want to comment on your items: So your points in order:
>Cast iron surface plates do corrode, but you could maybe treat it with 
an organic inhibitor?
--- Although I know there are cast iron surface plates, I envisioned 
using a standard machine shop granite surface block. Also there’s a 
strong possibility that pieces of plate glass – not standard window pane 
glass that is float made – rather plate glass, some of which is even 
polished. Old science experiments often recommended pieces of plate 
glass for interferometer experiments. So they have to be reasonably flat 
and for this experiment might be more than sufficient. You’d only need a 
slab about 4” x 4”, which a glass shop would probably give you from 
their scrap bin. Additionally, both glass and granite, especially 
granite are harder than cast iron and unlike cast iron, don’t corrode or 
oxidize. Obviously the glass will gradually flow out of alignment being 
a super-cooled liquid, but that’s a discussion for another day and a few 
years down the pike! 
Some approximate numbers
Mild steel 120 HB
Cast iron 200
18-8 (304) stainless steel 1250 HB
Glass 1550 HB
Hardened tool steel 1500 - 1900 HB
So my feeling is that a granite surface plate would be best, but you’d 
be dealing with some bulk and weight that has to include a method of 
leveling to 0.0001”. So as a compromise, glass plate on a steel platform 
with leveling screws.

>The thermal expansion properties will need to be very similar, or 
changes in the wire may try to rotate the cylinder. The wire / cylinder 
circle will have a high friction tending to prevent any slippage, which 
is a +.
--- The wire length expansion falls problem falls out of the equation as 
long as the two bands are of equal length and expansion. Un-equality 
shows up as a twist in the axis of the cylinder, i.e., if the cylinder 
is along a N-S line, it starts to rotate to an E-W line. This may or may 
not result in an apparent rotation with relation to the sensor, but the 
cylinder will just roll to place it’s center of gravity down. For the 
same reason, just make sure the “tails” of the wires are of equal 
length, then the expansion just shows up as a change in tension, but no 
movement.

>  Maybe use an all SS construction with SS foil instead of wire? You can 
buy 30" rolls of 2 thou SS foil from www.ksmetals.com You could use two 
end foils and one double width central foil wrapped half way around the 
cylinder just and stuck on at the top?
--- This is probably the area of most uncertainty. The first and biggest 
pause about using foil is the problem of obtaining bands without ripples 
on the edge that I’m certain would spell disaster for this scheme. You 
can’t just cut them with a scissors – any scallop at all would be a 
monstrous stiction point. So if they were to be used, they’d probably 
have to be etched out of a larger sheet, certainly a viable and standard 
method for many products, but in small or one-off prototyping, a big 
pain requiring masks and somebody who can etch steel, whereas wire that 
has been drawn is typically very consistent over long distances. Also as 
I touched on in my reply to Ted, I think that the higher pressure of a 
line contact from the wire is more likely to ignore certain types of 
surface imperfections and contaminations.

>  Have you given any thought as to what sensor and / or feedback 
transducer could be used, please?
--- I think Randall’s SDC is a perfect candidate. Stick the movable 
plate on the top of the cylinder, like a rooster’s comb and put sensor 
plates (supported by the surface plate) on each side. Randall’s SDCs can 
sense standard pendulum movements and as I mentioned to Ted, the 
cylinder would have an inherent 2X gain over the simple pendulum.

I would like to make one more point about temperature immunity. A full 
cylinder, while having the drawback of more rotational inertia, should 
have outstanding thermal stability. Although the diameter would increase 
significantly (in terms of normal seismometer criteria) the symmetrical 
nature should only change the distance of the mass balance point, not 
cause it to cross over and in effect overturn the cylinder. This would 
certainly be true if rather than build the cylinder exactly according to 
my sketch, use aluminum bolts (assuming an aluminum cylinder) that are 
threaded into the cylinder and go across the diameter. This way 
something that is supported by those cross members will be very stable 
in its reference to the cylinder axis, even as the cylinder expands and 
contracts. So although the period may change a bit, it doesn’t flip or 
have any reason to change rest position. For sake of construction, 
assuming a 3” aluminum cylinder, use 4 pieces of a 3 ½” 6/32 or 8/32 
flat head bolts. Drill opposing clearance holes across the diameter, 
rotate 90 degrees and slightly offset along the axis and drill two more. 
Now countersink two adjacent holes (two that are 90 degrees apart). 
Repeat on the other end of the cylinder. The countersunk holes will be 
the down side of the cylinder. Start bolt into the countersunk hole, add 
two nuts, push bolt through center weight, add two more nuts, push bolt 
through opposite hole, add last nut. The first nut is tightened up to 
hold flat-head tight to cylinder. The next two hold the weight nearly 
centered, and the last two nuts clamp the free end of the bolt to the 
opposite side of the cylinder. Repeat three more times. A thermally 
expansion balanced design.

Regards,
Charles Patton
__________________________________________________________

Public Seismic Network Mailing List (PSN-L)