Dear Stefan, LF Group,

I spent some time reading various texts and data books, and doing some 
calculations. The complete argument is too long and boring to type into an 
e-mail, but very approximately it seems to be the case that for a ferrite 
rod antenna and an air-cored loop antenna to have similar signal/noise 
performance, the rod length needs to be similar to the loop diameter, which 
seems intuitively reasonable. This means that the air-cored loop is better 
for larger antennas (a 1m long ferrite rod is very heavy and expensive), but 
the ferrite rod is better for smaller sizes (a 30cm long ferrite rod is 
quite reasonable weight/cost, and less bulky than a 30cm diameter loop).

In that case, "is it possible to make a 30cm long ferrite rod antenna and 
preamp with a noise level below the 136k band noise floor?" is the question 
to decide if the ferrite rod is worthwhile for /P reception from a low-noise 
location. Obviously there are many variables, but one can attempt at least a 
rough calculation.

Assume Stefan assembles his rods into 2 bundles of 7; this would be roughly 
equivalent to a single solid rod 28cm long x 21mm diameter. Assuming a high 
permeability ferrite, this l/d ratio will multiply the flux through the 
winding by a factor of about 70 compared to the same winding without the 
core (called "mu_core" in Watt's "VLF Engineering"), so the rod antenna will 
be equivalent to an air cored loop with an area of 0.025m^2. Assuming a 
noise floor of 0.06uV/m per sqrt(Hz), a single-turn winding with this area 
would have an induced EMF of 4.2pV/sqrt(Hz). With a low-noise preamp, assume 
the internal noise level is all due to the resistive losses of the antenna, 
which depends on the Q. Q of about 250 should be achievable; the inductance 
of a single turn winding depends on another permeability parameter, mu_rod, 
which depends on the rod geometry and the permeability of the ferrite; for 
this rod about 100 from Philips' ferrite data book. L works out to about 
0.16uH, and the loss resistance 540 micro-ohms. The noise voltage density is 
sqrt(4kTR), 3pV/sqrt(Hz). So the internal noise is below the band noise by 
3pV/4.2pV = 3dB. Hooray!

So it could actually be feasible. In order to make it work, it will be 
important to achieve a high Q. Obviously, a single turn winding with 
picovolt output levels is not very practical. I would aim for a parallel 
tuning capacitance of e.g. 5nF, so it can be tuned across the 136k band 
using a 500p variable. This would require L of 270uH, requiring about 41 
turns of thick wire, preferably Litz or multiple strands of thin wire. The 
parallel impedance at resonance with Q = 250 would then be 58kohms. 
Connecting the tuned winding directly to a "miniwhip" type FET input preamp 
should work well and add negligible amounts of preamp noise. The increased 
number of turns and the high Q resonant winding winding will increase the 
4.2pV/sqrt(Hz)noise floor at the preamp input by a factor of (41 x 250), so 
43nV/sqrt Hz. In a 300Hz CW bandwidth, this would be about 0.75uV of noise, 
so with a reasonably sensitive RX, no further gain would be needed 
(worthwhile checking if it IS reasonably sensitive though...).

I stress that these are all very rough calculations - you will have to 
actually try it out to find what the real values are. But they should be a 
reasonable "first guess", and it seems to show that the expeiment is worth 
trying.

Cheers, Jim Moritz
73 de M0BMU