[time-nuts] Re: ammonia, cesium, masers, etc.

[ed breya]

That's a fascinating article.

Regarding ammonia

Anhydrous liquid ammonia is a common commodity material, but probably 
comes in many grades, depending on use. I'd break it into three basic 
ranges. For something like this, you would want to look at reagent/lab 
grade, to get it as clean and dry as possible to start with, with 
minimal grief. Refrigeration grade may be good enough too, if you put 
impurity traps and dryers in your system. The crappiest versions are 
probably agricultural - as crude and cheap as possible. Don't forget, 
you'd also need a bought or leased tank, and proper manifold/regulator 
set to actually use it.

I don't recall how compatible ammonia is with copper or aluminum, but I 
know it's good with steel, and most likely all stainless steels. Keeping 
it absolutely dry would help to avoid any materials grief, but even so, 
it's a good idea to check out the chemistry before committing to a design.

Regarding microwave gear

Like Jim mentioned, the 22-24 GHz microwave stuff is in the right 
ballpark. I scrounged up some 1980s 22 and 38 GHz point to point 
transceiver head-ends over the years, and saved the goodies for possible 
re-use. In front of me right now are a couple of 22 GHz plumbing 
sections. At first glance, it appears they use Gunn oscillator modules 
as LOs, although they could be DROs, or a combination - hard to tell 
from outside. Then there's a bunch of waveguide BPF sections, with lots 
of tuning screws, and some directional couplers, altogether doing 
filtering and diplexing. I think these sorts of parts from this era 
would be great for this application. With all the fancy filters, it 
could be that besides TX/RX separation, the LOs may operate at lower 
frequencies, and the filters just let the right harmonics through. Can't 
tell yet, without more investigation.

I have some newer - probably 1990s - pieces in 22 and 38 GHz, where the 
guts are much more highly integrated into big modules that do 
everything, and just have the waveguide directional couplers sticking 
out, to connect an external diplexer. I've looked inside some of these 
over the years, and recall that none use a fundamental LO - they were 
all synthesized LOs in the maybe 6-8 GHz range, then amplified and 
multiplied up.

Regarding waveguide

It seems that they had the right kind of waveguide stock in the 1950s, 
so why not now? I know you can get flexible guide section, which is 
corrugated, but I don't know if there's such a thing as a smooth, 
semi-rigid form that can be bent or rolled up - kind of equivalent to 
semi-rigid coax. Maybe there is, or maybe not. Another option, for 
anyone with NC machines, could be to machine the internal structure into 
a block, say in a spiral or serpentine form, then cap it off from the 
side. It would have to be sealed though, somehow, to be good RF-wise and 
ammonia-wise.

Ed