I haven't seen the article, but there was something vaguely similar in one of the IEEE magazines a few months ago.

Think of an antenna (and the fields in its immediate surrounding space) as a somewhat leaky tuned circuit. The lower the Q the more leaky it is, so energy escapes more easily. The energy which escapes is the radiation field. So for a given amount of tuned circuit energy the lower the Q the more energy which gets radiated. That in turn means that for a given amount of radiated energy the lower the Q the less energy you need in the tuned circuit.

Then remember that a tuned circuit consists of a capacitor and an inductor. These store energy in electric and magnetic fields respectively. For an ordinary capacitor almost all this energy is between the plates. Similarly for an inductor; the energy is stored mainly between the coils. An antenna is like an 'opened up' capacitor/inductor, so the energy is not concentrated but spread out around the antenna. Most of the energy simply gets converted from electric (potential) to magnetic (kinetic) energy every half-cycle - exactly like any tuned circuit. However, an antenna is a leaky tuned circuit so some of the energy leaks off to infinity.

When someone tries to sell you an antenna never believe what he says, especially if:
1. the antenna is second-hand, or
2. the antenna is new, or
3. he says it doesn't need a balun/ground/counterpoise.

Most antennas need one of the items mentioned in point 3. If not given it they will use your feed cable and shack equipment as a balun/ground/counterpoise. About the only exception is a dummy load masquerading as an antenna, which will 'work' quite happily and give low SWR with no assistance.

A crude test is to touch the cable, or move your hand along it. If this affects the results then the balun is not working too well.

As others have said, you need to get rid of all the copper near the feedpoint. This adds capacitive centre-loading so will tend to move the resonance upward in frequency and raise the impedance.

It seems to me that part of the confusion about ground radials is that people confuse them with counterpoises. A counterpoise (e.g. as seen as part of the VHF quarter-wave 'ground plane') has to be vaguely similar in length to a quarter-wave, and isolated at the far end so the near end presents a low impedance. It acts like a transmission line. Putting several counterpoises in parallel (electrically) but in different directions (spatially) reduces impedance still further.

A radial just at or above ground and isolated at the far end could act like a lossy counterpoise. Lossy because it is near ground. When the radials are much shorter than a quarter-wave and grounded at the far end then they instead act more like connections (multiple connections) to the ground. As the ground is not a good conductor a single connection will have a highish resistance, so we need lots of connections wired in parallel. This will work best when the connections are as spread out as possible. Two ground connections near each other will act like just one connection. So my conclusion (based on thought rather than experience) is that if radials have to be short (the usual amateur situation?) then have them different lengths so that the ground connections are as far spread out around the available area as is possible i.e. don't make them all as long as possible with the ends grounded at the boundary but have some shorter ones with grounds nearer the centre of the area.

Does that make sense?

Funny SWR readings usually mean a fault in the SWR meter, or user error. Some SWR meters have a switch setting for SSB, I believe. SWR is a ratio, so the meter has to divide one voltage by another (or use a twin needle meter). In either case if one of the voltages is based on a different smoothing time constant from the other then errors will arise if the RF is modulated. Sorry, that is a long-winded way to say that a capacitor in the meter could be faulty.

Insufficient grid current means insufficient negative bias, so you get more anode current. Basically, the valve is being pushed less hard into Class C. You need to investigate why the loading is not working properly.

Yes, it must run parallel. This is because the Pawsey stub acts as half of a shorted quarter-wave twin line. The other half is the final quarter-wave of the feed line.

You may be able to do some impedance transformation in the feed line too, as that will happen on the inner currents. The Pawsey stub uses outer currents.

Bear in mind that modern devices often have dual or triple-band antennas. Quite difficult to replicate these. I suppose that if you know that in a particular situation the device will only use one band then you can provide an external antenna for that. Also some use diversity reception so two antennas may be needed unless you can guarantee no local variation due to things like people or pets moving around.

The D in 2D^2/lambda should really be the size of whatever is supporting radiating currents. For an antenna in free space it is just the antenna size, but for an antenna near other conducting objects it should include those objects. For a ground-mounted vertical the ground currents contribute very little radiation in the far field but near the antenna they are likely to enlarge the distance you need to go before you get to the far field region. At the least you should use twice the antenna length so you include the antenna image.

For some very small antennas mounted at low heights it is likely that ground currents are a significant source of radiation so their extent must be taken account of.

You might get a net improvement of 3dB in S/N ratio. The signal will add linearly, but the noise will add quadratically if it is uncorrelated. This assumes that the noise comes from the preamps (e.g. VHF and up in frequency). If the noise comes mainly from the antennas (e.g. LF, HF) then there may be little or no improvement as the antennas may be picking up exactly the same external noise.

The J-pole antenna needs a balun if fed by coaxial cable, although for some reason many people seem to imagine it does not. Without a balun you will get RF current on the coax outer and that could find its way back into your shack. At VHF and higher frequencies you can't just attach a good ground to short out this current, as might work for HF.

For best results use a balun at the antenna. Exceptions would be if the preamp has a balanced input and is directly connected to the antenna, or if you decide to use a balanced feeder (unlikely at UHF).

Don't be confused by people introducing different issues like impedance matching (at UHF you want a good noise-match not impedance match), or a second balun at the bottom end (won't help and might make things worse).

That WikiHow article may have been entered into a competition for maximum number of technical errors. The author can't seem to make up his mind whether he is making a 5/8 wave or a folded dipole, and the stuff about lengthening it is pure drivel. That is the nice thing about the internet: everyone has the same democratic right to share their ignorance of any subject with the whole world.

16.5 ohms and 2200uF give a CR time constant of 36.3ms. Unlikely to be noticeable.

So you want performance equivalent to the higher end of hi-fi FM tuners, but it has to be portable and run off a couple of AA cells? Oh, and it has to tune lower in frequency than a standard tuner. Nothing too special, then.