The “rules of thumb”:
c/o Ed Harfmann:
Foil coating waveguide (if making waveguide with aluminium foil) wrinkled – Is it more than 1/20th of a wavelength? 21 cm/20 = 1.05 cm. If < this then for amateur work not a concern.
Why do cables between the probe … to the LNA? Preferred none. Cables result in lose and we have an extremely small signal to begin with. That said, practical matters can (somewhat) overrule. Preferred as short of a run of cable as possible to the first LNA. After the first LNA, longer runs (with good cable) are acceptable as most people do not have their receiver (SDR) out in the weather.
(Unwritten but…) Why use white for the dish, other containers? Temperature and temperature swing.
- For the probe, reflected sunlight (for those times when the Sun passes through the beam) can melt the probe (or at least any dielectrics supporting the probe. A black or metal dish can reflect enough IR to melt a copper probe.
- For the filter, the tuning will shift with temperature as it expands and contracts. Usually the tuning is wide enough to not be a concern.
- For cable, the length will change as it expands and contracts. (See Wolfgang for some of their experiences with cables and an interferometer. The things radio astronomers have to think of…)
- For the LNA, the gain can change enough to think you have a signal… even with a dummy load instead of the probe. (e.g. a 0.01 dB change in gain is a 1K change in temperature.) Temperature stability (usually heating) is well advised. Why heating? Because generally our systems are not calibrated and we are looking for a change in signal, not an absolute value. (An absolute value is a next level game and is seriously hard work to get stable. There is a reason, beyond the big cost of the dish, that professionals have to pay a lot of money for their equipment.) If you have a stable gain, you can detect a change more easily (within bounds). Think of it this way: If you have a stable noise floor of 100K +/- 1k, you can reliably detect a smaller change in signal than a 50K noise floor +/- 50K. Cooling systems are expensive and (generally) generate a lot of RFI.