What core material would offer the highest sensitivity in the design of an ELF/VLF handheld rod antenna?

Question

I want to design an ELF/VLF handheld rod antenna. What rod material would offer the highest sensitivity at these low, low frequencies?

Below 50 Hz.

Answer 1

As said in the comments, I'm really not an expert on magnetic materials. But at ELF frequencies (3 to 30 Hz, definition of the ITU), as others have said, ferrites are unnecessarily lossy / low in $\mu_r$.

Basically, you're looking for materials that someone would use in a grid power distribution transformer. I don't know the cost of thing laminated silicon steel sheets, stacked to form a rod. I think this should be relatively easy to get, custom-made, because if you think about power transformer cores, they are usually formed by combining an "E"-shaped part with an "|"-shape; and they exist from "sugar cube" up to "building-sized", so if you have something you still can carry, someone, somewhere produces that, or knows a junkyard with old transformers¹. Small transformers are usually fine.

^{Single layer of a transformer core's E and I parts; you stack these until the rod is square in cross-section
Source: wikimedia commons, Author Dominik49, License Creative Commons CC-by-SA 4.0}

---

So, you want to build a high-sensitivity antenna, but sensitivity is actually a property that only emerges from the whole reception system.

You'll want to amplify directly at the receive rod; while a compact coil won't be very prone to picking up low-frequency/static electric fields, any cabling between antenna and amplifier will be. Luckily, these are easier-than-audio frequencies to amplify. So, you'll want:

• grounded shielding around your windings
• a balanced amplifier, ideally an instrumentation amplifier (sounds more scary than it is, 3 opamps, which you'll usually find in a single IC) to get the differential signal from the magnetic antenna that your rod antenna is (electric fields will largely be common mode). You'll profit from using modern low-noise, low-bandwidth opamps here. If any circuit you find online recommends an opamp that contains the number "704" in its part number, close the tab and find someone who knows what they're talking about instead of copying schematics from the magazines of the early 1970s.
• an active low-pass filter. People processing low-frequency signals tended to put a lot of effort into analog filtering here, to reject high-amplitude interference, as from grid-frequency sources (50 for me as European, 60 Hz for you with a North American callsign). That is generally a good idea, but no need to go more than two stages here, one stage of low-pass filtering should totally do, because:
• an ADC driven by an appropriate buffer as early as possible will be the most sensible choice here. "Sample first, ask later" is one of the guiding principles for SDR, and for good reason: As long as you hit the dynamic range (and even a below) of the ADC, oversampling low frequencies (and I consider everything < 10 MHz low, here) is
  • practically free (even $2.00 microcontrollers sample at 1 MHz these days)
  • gives you free resolution when you consequently resample (downsample) in digital domain; one bit for every oversampling factor of 4 (and assuming you sample at 131.072 kHz with a ENOB of 6 bits, and want a band of 8 Hz, that's theoretically (6 + 131,072 Hz / (8 Hz)/4) bits, that's far more bit depth than you'll ever need, or, conversely, better SQNR than your signal could ever have; and you really don't need any more analog filtering for that than necessary to avoid saturating the ADC.
  • once digitized, for all practical purposes in this context, math done on numbers in computers is "lossless" and "noise-free", very very much unlike filtering in the analog domain
  • Designing a filter (or a set of filters) to run as software is not only much easier, you can also implement much steeper and preciser filters without tuning
  • You don't have to tune your filter to a single band to begin. You can instead get the full bandwidth from the antenna and preamplifiers, and then just inspect in as many channels as you want

And all that can absolutely be done on a laptop, with free and relatively easy-to-use software. If you shy away from building your own ADC board (and I'd recommend you shy away for starters), you'd use a sound card². Use GNU Radio, drag and drop an "Audio Source" onto your flow graph that you design in GNU Radio companion, boom, instant access to your receiver signal.

---

¹ I just looked at Youtube for transformer dismantling videos, and wanted to scream at a few people: the oil in older oil-cooled transformers is often, actually usuaklly, carcinogenic. If you deal with with the impregnated paper that you sometimes find between layers of windings in power transformers: also not the healthiest chemicals that go into making these flame-retardant. Neither is likely to be a problem if you dismantle transformers for luggable components, as these small ones typically don't have oil cooling or such flame-retardant separators.

² You'd then want to instead have a mixer, typically just a chopper, to upconvert your 3 to 30 Hz to some frequency sound cards are happy with, but luckily you have a sound card as source of local oscillator for that mixing, too.

Answer 2

For low frequencies and low field strengths (receiving only, similar to earth's magnetic field) you could also look into Mu metal or amorphous metals, which can have extremely high permeability.

I've only used their cores for suppression at VHF frequencies, where they are quite lossy, but they're also used for current transformers, Earth-Fault / GFCI sensors, where they're more compact than the powdered metal or laminated steel alternatives.

For example VacuumSchmelze makes Nickel-Iron and Cobolt alloys with permeability up to 300,000. This extensive PDF gives a lot more detail about the materials.

---

I also agree with the idea of using regular steel. Steel has an initial permeability of ~ 1000 which is far higher than any of the ferrite materials, but much lower than the more exotic materials.

Ideally grain-oriented laminated magnetic silicon steel - you could obtain this by taking apart a toroidal transformer, unwinding the core and laying it flat. You could also use a collection of wires (think of a box of TIG welding rods, but individually painted before bundling up again). Even simpler - at 50 Hz I think the loss from a plain reinforcing rod would not be too large.

Answer 3

Another thing to consider is the strong background noise at these frequencies, caused by thunderstorms. Your antenna only needs to be sensitive enough to measure this noise, more sensitivity makes no difference.

The classic ITU noise curve only goes down to 10 kHz, but you can see the trend is strongly upwards.

Another clue is the antennas already built for various purposes. This paper [pdf] gives values of ~ $100 \text{ fT}/\sqrt{\text{Hz}}$. FemtoTeslas! If your receiver bandwidth is 10 Hz, you can expect 300 fT magnetic field of background noise.

The antennas used are described in Fraser-Smith and Helliwell, 1985, which says:

The ELF antennas are specially designed and constructed loop antennas of circular cross-section with a mean radius of about 0.49 m. To control their electrical characteristics, the coils were wound in 12 segments of 97 turns, giving a total of 1164 turns. Their resistance is 75 ohms, their inductance 2.7 henries, and they weigh roughly 30 kgm (65 lb). The VLF antennas can be supported above ground by means of a mast, but to avoid wind-induced noise the ELF antennas must either be buried or carefully shielded from the wind with an appropriate structure.

Flux through a 1 square metre area antenna is just ~ 100 fWb/rootHz. Induced voltage will be about $4.44 \phi F N$ or ~ 6 nV/rootHz, this is about -164 dBm/Hz into 1000 ohms, which is not unreasonable. (You'll need a high-impedance pre-amplifier, Zant is about 1000 ohms at 50 Hz).

By wind-induced noise, the authors are referring to movement of the antennas creating currents by interaction with the earth's magnetic field. And here is your second useful datapoint - at 50 uT, the earth field is about 10^8 times stronger than the background noise. That's OK if the antenna is perfectly stationary, but if you move it even slightly you can see this will dominate the radio signal. (it will be mostly lower frequency components, but there will certainly be some 50 Hz component from basic handling vibrations). So your handheld antenna will be first of all an extremely sensitive microphone. You might need to bury it, like the Stanford project did, to eliminate this source of noise.

Answer 4

It won't be ferrite, I can tell you that much.

Once you find what cores are effective at 50 kHz, then the next question should ask about the core shape and the antenna design.

You didn't specify your frequency range. ELF and VLF have different meanings, depending on where you get the definition from.

Powdered-iron cores are what is needed. Micrometals has a mix that works in the upper audio range.

Please see this answer for Need a ferrite suppression material for choking 15 kHz to 60 kHz

In general, though, ferrite materials are not suitable for low frequency applications. Look at powdered iron materials like (Micrometals) #26 instead.

---

https://www.micrometals.com

---

Here is a relevant question with an answer from ee.SE:

https://electronics.stackexchange.com/questions/92379/50-hz-antenna-for-monitoring-mains-frequency-fluctuations

---

You may not even require a core. I have two Blitzortung lightning receivers that help feed the lightningmaps.org website. Their antennas are air core, 3 or 4 turns on a 1 meter diameter support. They feed into some kind of amplifier and then into an SDR. Lightning peaks at about 16 kHz.

A more compact version of the antenna could likely be made from many more turns with an air core.