Some
History and Some Thanks. A number of years ago, I heard about
these curious phenomena called 'whistlers' on a shortwave science
program. I purchased a NASA Inspire VLF receiver kit, but never got
much out of it aside from some hiss - which I'm sure was entirely my
fault in construction. I put it on the shelf and gave up, figuring
that it was just too hard to duplicate what I had heard on the program
that day. Quite some time later, I stumbled across Renato Romero's web site, and it
re-ignited my interest. I purchased a copy of his book, and after a
read and some research on Yahoo's VLF_Group, I decided to take the
plunge and build a receiving station of my own.
This page catalogs my efforts to get my original system put together,
with some modest success. I document some of my failures, along with
bits of advice, to help others not make the same mistakes, provide
some humor, and to relieve anyone of the notion that I might actually
know what I'm doing...
I owe many thanks to Renato for writing his
book in the first place, Paul Nicholson for his patient
guidance, encouragement, and help as I experimented my way into
operation, and to all of the members of Yahoo's VLF_Group for
providing years worth of good reading, analysis, and ideas. It
is very, very appreciated.
The
Marlton Environment and Early Reception Attempts. The part of
southern New Jersey where I lived at the time consists of farmland,
dotted with housing tracts, industrial parks and small lakes. It was
criss-crossed by a number of high voltage power lines - one being
about a mile from where I lived. The soil was sandy, with pockets of
very dense clay. I lived on top of one of those clay pockets (in fact,
Marlton gets its name from 'marl', which is a sedimentary mix of clay
and lime.) In my old neighborhood, all of the homes
were fed power via underground cables, with transformers housed in
small metal cabinets at ground level every thousand feet or so. I had
one right in front of my home.
Did I also mention I lived about 11 miles from a city with 1.5 million
people?
Not exactly the best place for VLF reception.
I decided to start with Renato's Floating Solar Receiver. I
thought I would be clever and start with something that would be
somewhat 'independent' of local ground (or so I thought) and nicely
side-step all of the classic problems with VLF reception. Renato's
design is good - but my implementation
was bad. My very first receiver, frankly, was an awful, noisy mess.
After more self-induced frustration than was necessary, I scrapped my
version of the floating receiver and started over.
Which brings us to our first topic, and my first mistake...
Receiver
Location. It may seem like an obvious statement, but taking
the time and carefully selecting a location for your VLF receiver is
important. When I get excited about something, I tend to rush. Once I
had built my 'cantenna' and original receiver, I wheeled it around in
the same general open space in my backyard, and was thrilled to hear
some sferics buried under hum and interference. I picked an open spot
and set up shop. My initial choice of location actually ended up being
one of the worst on my property. My final location, about 20 feet
away, had almost 20dB less hum interference.
The AC mains hum, it's harmonics, and interference from other local
sources can set up standing patterns with 'local' maxima and minima.
Through careful study and observation, you can select a location that
minimizes (relatively speaking) that interference.
Take the time to build or borrow a VLF receiver, and survey your area
to select the best spot. You will save yourself a good deal of
frustration later.
Some tips:
- Build a mobile receiver (RX), constructed in a case that lets
you mount a whip, and a ground spike (like a tent stake or long shaft
screwdriver.) It makes taking fixed observations much easier.
- If you have a netbook or laptop, don't select candidate spots
by ear only. While it may sound like you have found a local minima
just by walking around, there are other sources of interference that
are at the edge of your hearing that will only show up on a
spectrogram. Wolf's Spectrumlab
running on a netbook is excellent for this purpose.
- Isolate the rx from your laptop with a 1:1 audio transformer,
and several meters of wire (at a minimum). This should
keep a lot of the noise generated by the laptop from bleeding into the
signal.
- Take samples for each location at multiple times of day (and
night.) Hum varies quite a bit, depending on the weather, neighbors
and a dozen other factors. A spot that sounds good now, may end up
being awful the rest of the time. As an example, here is how the AC
mains 3rd harmonic varied with time at my location:
One Week of 180Hz. Click for larger image.
- Take detailed notes. Save spectrum samples with time and
location information. You will be glad you did when you go back to
analyze your options or get a second opinion.
- If you are setting up an e-field receiver like this one, avoid the
obvious, like nearby transformers, power lines and your home. Also
avoid: trees, metal fences, chicken wire around the garden, clothes
lines with metal cores, drainage grates, metal lawn furniture,
sheds, outdoor cabling and lighting fixtures. Anything and everything
metal. All of these enhanced mains interference.
RX-6
Design and Operation. RX-6 gets its name from being the 6th
major revision of what amounts to an e-field probe feeding a very low
noise op-amp. In reality, there were many more than 6 variations, as I
made changes in design and layout, and observed the effects.
RX-6 Schematic. Click for larger image.
Circuit operation (as I see it, anyway): R1 acts as a bias R for
the input of U1, an AD823. R2 and C1 form a low pass filter with a
cutoff of approximately 630kHz. R2 and C1 also slow incoming
transients (to a degree) to allow switching diodes D1 and D2 to clamp
excessive input to either the V+ or V- rails. NE2 rounds out the input
protection and provides a flash over point between the whip antenna
and system ground. U1 is a very low noise op amp, configured as an
adjustable gain non-inverting amplifier. First stage gain levels are
set between 1 and 100 via RV1 and R3. C2 forces U1 to also act as a
high pass filter, with a cutoff frequency around 100Hz. C3 provides
high frequency roll-off to help with broadcast band (AM) interference.
R4 and C4 form and additional low pass filter with a cutoff of
approximately 160kHz. R5 acts as a load for the AD823 to help with
oscillation, as the input to the line driver stage (U2, OP27) doesn't
present much of load. U2 is a low noise op amp with a low input
impedance configured as a non-inverting amplifier with a gain of 2, as
well as a second stage of high pass filtering with a cut off frequency
of about 100Hz. U2 also acts as a line driver stage for the receiver,
and is coupled into a 1:1 matching transformer via C7 and R9.
The unused section of U1 is configured as a voltage follower,
with its input pinned to the virtual ground between V+ and V-. This
seems to reduce the current consumption of the AD823, and also
prevents the unused section from oscillating.
The power supply (shown in the larger graphic) provides the split
power supply necessary for the receiver. The current draw on the
virtual ground rail (SUP_GND) is minimal, so a simple passive voltage
divider will work. L1 and R7 help to decouple the 4 foot cable run
from the supply battery on the ground. Despite the good power supply
noise rejection of the AD823, leaving them out increases interference
in the receiver.
The gain was set at 80 (x40 first stage, x2 second stage) in my
implementation. The overall receiver as drawn is capable of a gain of
200.
The input low pass R2/C1, roll-off cap C3, and low pass R4/C4
were all necessary to prevent a local (< 10 miles) 50kW AM station,
WCAU 1210, from injecting all kinds of intermod into the VLF signal.
High pass filtering in both stages was necessary to prevent clipping
from the local hum, and to reduce the work for digital processing
later.
The AD823 has a drive capability of 15ma, and is rated for a 500pF
load. Run into a long run of twisted pair cabling with its high
capacitive load, the signal became quite distorted. If you examine the
parameters for common coax or twisted pair cable, you will see that
the total capacitive load of even modest cable runs in in the 1000s of
pf. The OP27 is somewhat beefier than the AD823, able to source around
20ma and drive capacitive loads up to 2000pf. It is better suited for
driving a longer signal line.
If your environment does not have the intermod problems, you may be
able to eliminate some of the low pass filtering, which I did in later
versions of this receiver. Likewise, if your situation doesn't
need it (shorter cable runs, better cable, etc.) you might also be
able to drop the second stage entirely.
Design tip: If you are building an op amp based receiver like this
one, be sure to check the voltage and current noise parameters of the
op amp you intend to use. Using these you can infer the 'natural'
source impedance that the op amp expects. In an e-field probe, low
input current noise and high impedance are important.
Receiver
Performance. The noise floor of RX6 is below the VLF spectrum
noise floor, allowing relatively clean reception. Overall, the dynamic
range is also quite good.
Noise Floor Comparison. Click for larger image.
In the chart above, the red line represents the noise floor of the PC
sound card used to interface with the RX-6. The green line shows the
noise floor of the RX6 itself, with no antenna connected and a small
capacitive shunt across the antenna terminals. The blue line shows a
snapshot of the local VLF spectrum from 0-24khz. As you can see, the
noise floor of the receiver is below that of the VLF spectrum we're
trying to receive.
The upward swing of the RX-6 noise curve below 5kHz may be able to be
improved by raising the value of the bias resistor R1. It's on the
list to try, anyway.
The slight upward tilt of the VLF spectrum above 23kHz in this graph
was an artifact of the sound card I was using (Sound Blaster!), and
not from the receiver itself.
Average current consumption for RX-6 is relatively low, around
12-15ma, with upward spikes during sferics and other activity.
Getting
The Received Signal Somewhere Useful. Getting the VLF signal
from an RX-6 to the PCs I'm using for processing the signal proved to
be one of the single biggest challenges I faced. While it sounds
relatively straight forward, like stringing some coax or twisted pair
from point A to point B, there are a number of issues that adversely
effect the quality of the signal and the performance of the receiver.
Tip No. 1: A
difference in ground potential between your receiver and where you
process the signal. Try this experiment: Place a ground rod in the
spot you've selected for your receiver, and add a run of hookup wire.
Use this to measure the difference in potential between your domestic
ground (which feeds your PC) and the earth. If you see a 10mV or 100mV
difference, you'll probably be fine with transformer isolation.
However, in my case, there was over 1V difference, in the form of a
nasty, spiky triangular waveform. After some investigation (turning
off and disconnecting different loads in my home, checking ground
resistance for my breaker panel, etc.) I came to a preliminary theory
that the differential voltage probably came from the placement and
grounding of the in-ground transformers in our old neighborhood,
causing ground currents. The best above ground e-field "hum minimum"
on my property was on the general axis between two of the transformers
- which places the ground rod along a ground current path.
Tip No. 2: Long
cables, even on/in the ground, can act as a very nice antenna for
interference to enter your receiver. Be careful where you run the
signal line for your receiver - avoid running parallel to power lines
(inside and outside), near signal lines for TV antennas/satellite
dishes, in-ground accent lighting, etc. Unless trees or other
obstacles prevent it, shorter direct runs are best. I have had good
success burying my signal cable 1-2 inches underground.
Tip No. 3: Even if
you utilize 1:1 transformers on your signal line to provide 'galvanic
isolation', you still aren't truly 'isolated'. Why? Every transformer
has a (usually) small but finite capacitance between the primary and
secondary windings. That capacitance acts as a path for AC signals. In
this case, it can act as a path for that nasty 1V sawtooth. I ended up
using a series of 3 transformers in my signal line, and I am
contemplating adding a fourth. Fortunately, adding the
transformers didn't introduce very much attenuation for the desired
signal.
Signal Cable Configuration. Click for larger image.
Tip No. 4: You
may have to ground and/or shield different parts of your cable run
differently. Somewhat counter-intuitively, grounding one side of the
first segment of my signal line helped to reduce hum by another 3 to
5dB. Go figure. Not a recommended way to do things, as it somewhat
defeats the isolation you're trying to achieve in the first place.
Tip No. 5: Large
capacitive loads frighten most op amps. If you take a look at the
pF/foot(or meter) specifications for common coax or twisted pair
cables, you will soon discover it doesn't take a very long cable to
produce a large capacitive load. As I stated in my analysis the RX-6,
some op amps are better suited for driving these types of cable.
However, you can help the situation by careful consideration of the
cable you have on hand or are about to purchase. Try to minimize the C
per unit length.
One of my current experiments tries to side-step some of these
factors by sending the VLF signal optically via low cost fiber optic
cable. The challenge I'm currently working on involves flattening the
response curve as much as possible from 1-100kHz. Check back for
updates.
Powering
The
Receiver For Extended Observation. Connecting a mains powered
DC supply to the receiver defeats the galvanic isolation you just
worked so hard to achieve. My initial attempts at resolving this and
powering my receiver used a solar panel with a basic current limiter
and a sealed 12V gel-cell battery. The first few iterations of my
receiver seemed to do fine with this arrangement. However, as the
noise floor lowered and the system improved, I ran into a problem:
hiss from the Sun. I removed the current limiter, double checked the
panel to make sure that it was only a series of Si solar cells (and no
active circuitry), and added filtering - but I was still unable to
eliminate the hiss completely. The solar panel was fairly close to the
receiver and antenna, so it could also have been capacitively coupled.
Either way, I ended up dropping the solar panel altogether to
eliminate the noise. I was later able to work around this problem by
moving the panels a sufficiently large distance away from the receiver
when I built my Wifi
VLF Receiver.
Currently, the RX-6 is powered via a 115AH deep cycle marine battery.
A charge lasts for months, and the batteries have two important
properties: cheap and simple.
Another option to consider is high-frequency (100kHz or higher) AC fed
via twisted pair and isolation transformers.
Construction
Tips. If you build one, I recommend trying the RX-6 out on a
breadboard. The RX-6 is not a high frequency design, so in general
layout is not that important - to a point. The gain of the receiver is
not insignificant, and coupled with the high impedance input stage it
can tend to oscillate if you don't take some care in construction and
signal routing. The first few versions were on a small breadboard with
a very messy layout (a result of many revisions and experiments.)
Cleaning up the layout and spacing things on a larger breadboard has
completely eliminated the early oscillation problems. My long term
goal is to make a PC board for this receiver, which should be
relatively painless with something like KiCad
or gEDA.
The receiver body itself was housed in a Carlon outdoor plastic
junction box purchased at a local home improvement center, with a weep
hole and some dessicant.
Be sure to put drip loops in your wiring before it enters the box!
The whip antenna is approximately 8 feet tall, and made from 3/4" PVC
pipe. Internally, it is 18ga hookup wire glued to the top cap, run
through the length, and brought out through a small hole in the bottom
cap. The body is also filled with clean, dry sand. This arrangement
provides some weight to the antenna, and seems to have eliminated the
majority of noise from wind and rain.
It may also be necessary to 'sound proof' the receiver enclosure
itself. For a while I could clearly hear individual raindrops hitting
the enclosure. Adding some light padding around the receiver in the
enclosure solved the problem.
The enclosure and whip are attached to wooden stake with some 'give'
to it, and the stake itself is secured to a large steel spike in the
ground. Experience has shown that a somewhat flexible mounting method
and whip tend to be less
noisy than a firmly mounted one in strong wind. You want something
that flops very gently from side to side in a strong wind - not
something so flexible it flops around wildly, or so stiff it sings
like a guitar string.
Processing
The Signal. Once you get the signal indoors, there are quite
a number things you can to do with it. Just to name a few:
To accomplish any of these, however, you will need to be able process
and store the signal in various ways. The two primary software
packages available for doing this (that I'm aware of, anyway) are
fortunately both very good, and free:
Windows: DL4YHF's
Audio Spectrum Analyser. If you're a ham, you're probably
already familiar with Wolf's software in other contexts.
I utilize vlfrx-tools for initial processing and storage, as well as
streaming and post processing (SID detection, etc.) I also use
Spectrumlab on the same signal, forwarded from vlfrx-tools, for it's
enhanced real-time display.
For information on my vlfrx-tools setup, please see this
page.
Results.
Overall, I was pretty pleased with the results. The receiver was quite
reliable, and sounded quite nice. I frequently listen
to it in the background as I do other work. Having completed the
receiver in the early winter, I didn't expect to hear much - but I did
expect to hear at least something.
Quite some time went by, and while the background noise and spectrum
looked and sounded like it should, I didn't hear any whistlers. No
chorus. I heard tweaks, but that was about it. Then, early one January
morning, I got 4 surprises - one of which you can see here:
Spectrum of a whistler. Click for larger image.
Whistlers! (Finally!) Since that time, I have received more,
along with a nice 6 hour run of chorus on 2012-02-15 (see below.) It
was certainly a start.
Here are some samples. You may need to raise the volume a bit to hear
the whistlers.
A whistler
received at 2012-02-15 06:38:04 UTC.
A whistler
received at 2012-02-05 22:18:07 UTC.
A 3.5 minute
compilation of whistlers and chorus, received early in the
morning 2012-02-15. I've cranked the gain quite a bit, so that you can
more easily hear the chorus starting about half way through the clip.
I hope this page encourages you to take the plunge. You'll be glad you
did.
The content presented here is the original work of Mike Smith unless
otherwise shown. Please contact me
for comments or errors.
This site was built using Emacs, GTML and CSS.
