Using your computer to ambush unsuspecting NDBs...
(Using Spectran)

Listening for NDBs:

One of the sub categories of DXing (that is, listening for distant radio stations) involves listening for NDBs - Non Directional Beacons (see sidebar.)  One of the objects is to try to pull as many of these NDBs out of the noise as you possibly can - hoping to hear new and ever more distant ones.

The big question:  Why?

That is a good question.  Typically, these NDBs have nothing in particular to say - except their callsign over and over again.  However, there is a certain fascination to be had with trying to hear an ever more distant signal - not to mention the vague mysteries of low-frequency propagation - especially during the winter months when the atmospheric noise is at its seasonal low.

Listening for NDBs has some more practical aspects as well:  Working to pull a signal out of the "muck" hones ones skills in being able to copy signals under the worst conditions - a skill that can come in handy when trying to communicate on the amateur radio bands when conditions are poor, when there is interference, or when the signals are just weak such as during EME operations.
What are NDBs?

NDB is an acronym that stands for Non Directional Beacon

What does that mean?

NDBs are intended to be used by pilots as an aid in flying toward a point or as an aid to mariners in navigation.  The Non Directional part comes about because these beacons are not intended to "beam" their signal in any particular direction unlike several aero-navigational systems.  NDBs typically operate in the 190-535 KHz region - a frequency range that gives them good groundwave coverage during the daytime - but it also allows for often excellent propagation during the night via skywave. 

NDBs identify themselves with a series of 1-3 letters (usually) that they send in "Morse" Code at a low enough rate that they may be identified by someone who does not even know the code.  The pilot (or captain) using an NDB will tune to the frequency of the beacon near their intended point (either their destination, or a point along their planned course) and using an RDF (Radio Direction Finder) know what their heading is with respect to the bearing of the beacon.

Techniques for listening for NDBs:

No doubt that every NDB DXer has his/her own listening techniques, but it all boils down to one main thing:  Patience.  If you expect to take a quick spin across the dial and hear rare and distant NDBs, you will likely be very disappointed.

One of the very best sources for techniques on listening for these beacons is the series of articles that appeared on the LWCA Lowdown beginning in the late 1980's entitled On the Art of NDB DXing by Sheldon Remington.  Fortunately, this series of articles has been made available online here on the LWCA website.  Although some of the specific information is somewhat outdated, the general techniques are more enduring.

It can be safely assumed that almost everyone doing NDB DXing these days uses a synthesized receiver - or at least one with a digital readout with resolution of at least 100Hz.  One of the more common techniques when doing a "survey" is to start at one end of the band (or even in the middle) and move up or down in systematic 1 KHz steps, pausing to see what can be heard at each step, doing some fine-tuning as necessary.

There are several problems that arise in NDB DXing making it more difficult to pull out some of the weaker beacons.  These include:

A screen-capture of Spectran showing four NDBs
A Screen Capture of the Spectran program with the receiver tuned to 311 KHz USB.  There are four beacons that are easily visible here.  Around 400 Hz, the lower and upper beacons are 9Y (Pincher Creek, AB) and HY (Hay River, NT) respectively. Around 1 KHz, the MVI (Monte Vista, CO) and BFE (Brownfield, TX) are the top-most and the next one down respectively.
(Click on the image for a larger version with labels)

Again, NDB DXing takes patience, concentration, and skill.  As technology improves, however, it is certainly within our rights to use it to pull out those beacons that you may have previously missed.  The most helpful accessory to a computer in NDB DXing (besides a program to log reception of beacons) may be a sound card:  Programs are available to not only provide good audio filtering, but to visually represent what is being received as well.

One of the best-known of these is Spectran.  Although this program is undergoing continuous evolution, it already not only provides a "waterfall" display that graphically shows the audio spectrum in three dimensions (frequency, time, and amplitude) but it can do real-time audio filtering as well in Versions 4 and above.  While this program is suited for extremely slow-speed copy of CW (such as QRSS) it can provide an aid to NDB DXing.

The image shows the result of listening on 311 KHz using USB:  I prefer to use USB as the RF frequency of a signal displayed is simply the addition of the dial frequency on the receiver and the audio frequency on the screen.  To understand this display requires a little bit of knowledge on the nature of NDBs.  Typically, Canadian NDBs have their idents modulated at 400 Hz and U.S. NDBs typically modulate their idents at about 1020 Hz (there are exceptions, of course.)  NDBs are modulated using AM (but often with just a single sideband) which means that they will also have a continuous carrier.

When a beacon is zero-beat one will see the modulation sideband received (in this case, the upper sideband) and modulation of signals with the displayed passband - at 400 Hz and 1020 Hz, as in the image.  Fortunately (for us NDB DXers) not all beacons end up precisely on their designated frequencies.  This error arrives from two sources:  The carrier frequency itself may be off causing its sidebands to be off-frequency as well and/or the modulating tone may be off-frequency, causing it to appear at a distance different from the usual 400 or 1020 Hz away from the carrier.  (Note that some NDBs actually consist of two transmitters:  One providing a continuous carrier, and another with a CW-keyed offset frequency providing the "sideband" carrying the identifier.)

The settings for Spectran for this image were a 5.5 KHz sampling rate and a 2.7 Hz resolution:  This provided a display bandwidth from about 100 Hz to 1.8 KHz, showing everything over a range greater than a 1 KHz "slice" of spectrum.  While this resolution is far too narrow to show the individual dits and dahs of the beacons, one can easily ascertain the existence of beacons.  What does this display tell us?

Why not use Spectran to display the dots and dashes directly?

This is a good question. 

In the case of the example above, the frequency resolution was 2.7 Hz.  In simple terms, this means that something something that goes from "on" to "off" faster than 2.7 times per second will "blur" together - as well as start to occupy more "width" on the display.  Since the beacon's ident is at, say, about 5 WPM or so, we'd need to be able to display changes that occur (an off-to-on) at faster than 5-6 times per second (Hz.)  Even this rather slow response tends to cause the lines to "run" together - so you'd need a display bandwidth of closer to 10 or 15 Hz. 

This can cause a problem:  At this bandwidth, adjacent "traces" (from beacons on adjacent frequencies) can start running together and it may be difficult to distinguish two signals that were close to each other in frequency. 

What you end up with - if you do it right - is a visual representation of dots and dashes on the screen.  While this method works, it is usually more effective to simply listen to the tone with your ear than trying to interpret a display on a screen:  It turns out that one's ears are better-suited to copying code than their eyes. 

Why run the narrow bandwidth then?  As it turns out, that although you won't be able to determine the ident with the narrow bandwidth, you can tell that the ID is there.  It also tells you where to set the audio filter's passband limits.

Using Spectran's audio filters to "get" NDBs:

There is really no substitute for good IF filtering.  A narrow filter can help prevent a much stronger NDB on an adjacent frequency from throttling back the AGC, for example.  However, you aren't likely to find an IF filter much narrower than 250 Hz or so - with 500 Hz being more common.  Unless you are a truly avid NDB chaser (or an amateur radio operator that does a lot of CW operation) you aren't likely out to invest the $60 or more (sometimes much more) that it may take to get another filter.  An audio filter can certainly fill a niche - especially one that is narrower than available IF filters.

There are very specific limitations to audio filters, however:  If a nearby signal is causing the AGC to "throttle back" the desired, signal, an audio filter won't necessarily help you.  Turning off the AGC may help, but whether you have an AGC or not you are still faced with the same ratio between the weak and the strong signal.  Switching to a narrower filter and/or adjusting the radio's PBT (PassBand Tuning) or IF Shift and appropriate selection of the receiver's sideband selection (to prevent audio "images" from appearing in the audio) may allow you to put that offending signal outside your existing filters.

Looking at the above image of the four NDBs, listening with the "naked ear" may allow one to pick out one, maybe even two of the signals.  As you can see, there is really nothing in the audio other than the signals near 400 Hz and 1000 Hz.  The trained ear is perfectly capable of picking out the stronger of each of the two signals at each of these two different frequencies - assuming that there are different signal strengths.  However, what if each "pair" of beacons were equal strengths, what would you do?  Several things come to mind:

The last one is where Spectran can come to the rescue in being able to set up a bandpass filter of arbitrary frequency and width.  Referring to the image, you'll notice that above the trace for the highest-frequency beacon in the 1000 Hz group along the bottom of the spectrum analyzer plot (just above the "frequency" ruler) you'll see a blue "bracket".  This "bracket" shows the frequencies that will be passed by audio filter.  Looking carefully, you'll notice that the bottom end of the filter is set between the top signal (MVI-331) and the next one down (BFE-331.)  This effectively allows only the MVI signal through.

To be able to receive the signal from BFE I placed the filter's limits just below the line for MVI and just above that line just below BFE's signal (keep in mind that I didn't know which beacon I was "seeing" until after I successfully copied its ident.)  It is also worth pointing out that at the precise moment that this image was captured, BFE's signal was not copiable - I had to wait for a few minutes before it slowly faded back in before I could copy it.

Setting up audio filters in Spectran:

Setting up the audio filter in Spectran is fairly easy - if you know how:

Your computer, sound card, and Spectran.

Spectran works with the vast majority of full duplex sound cards (Almost all sound cards made in the past 2-3 years are full-duplex.).  If you don't have a full-duplex sound card, you may be able to display the waterfall, but you will not be able to use built-in audio filters. 

For displaying the waterfall even a fairly fast '486-class machine (with a half-duplex sound card) is adequate:  My ancient 75 MHz '486 laptop will run the older Spectran Beta 3 reasonably well. 

For audio filtering, you will need a more "substantial" machine:  I would recommend at least a 133 MHz pentium-class - and much faster if you need to operate at a higher sample rate. 

I used to use a 166 MHz Cyrix machine for my monitoring, running the sample rate at 5.5 KHz (which is adequate for any audio frequency up to just above 2 KHz) and the processor loading is acceptable. 

If, on this same machine, I attempted to operate at a 22.05 or 44.1 KHz sample rate (to run the necessary 11 or 22 Hz bandwidth to actually "see" the dits and dahs) the machine begins to fall apart:  The processor simply doesn't have enough "horsepower" to update the display consistently and do audio filtering - I will sometimes use my 450 MHz machine for that. 

Another thing to watch out for is the accuracy of the sound card.  I have a very good quality sound card in my 450 MHz machine, but in old 166 MHz "Ham Radio" computer, I only spent about $8 (new) on its sound card.  This cheap sound card has (maybe) a 55 db S/N ratio and its sample rates - derived from the computer's bus clock (it doesn't even have its own on-board crystal) are, in some cases, over 0.5% off-frequency.  This can result in 5 Hz or so error at 1 KHz. The "good" sound card has no measureable error at audio frequencies, being as accurate has its on-board crystal oscillators. 

When setting filters there are several things to remember:

Other filter modes:

I would be remiss if I failed to mention that Spectran has three another filter modes available:  The Denoiser, and the Humid filter, and the CW Peak filter.

The Denoiser:

This filter is a typical noise-reduction algorithm found in many audio DSP implementations.  It works by mathematically eliminating those spectral components that are random, revealing (hopefully) those components that aren't just randon noise.


The HUMID algorithm (short for HUM Instant Destroyer) is a comb filter.that is designed specifically to remove noise that is a result of AC powerline interference - specifically AC hum.  Note that this is primarily intended to remove the AC hum that would be received by a VLF antenna system - that is, one that is designed to directly intercept RF in the frequency range capability of the computer's sound card (that is:  Plugging your VLF antenna directly into the sound card's input!)  Note that direct VLF reception is best done with a specialized antenna system and that simply plugging your HF antenna into your sound card's input is not recommended.  If you want to try your hand at receiving VLF/ULF signals,

The CW Peak Filter:


Return the KA7OEI main page, or look at these other pages:

The Radio Waves below 22 KHz site. - This site has a lot of articles about VLF/ULF topics - a must-read!

Online NDB (and related) Databases:

AirNav's Navaid Information - This provides a database that is searchable by frequency, identifier, or name.  This database encompasses North American and at least some Central-American beacons.

Canadian Nondirectional Beacons - A searchable database offered by Pierre, KA2QPG

William Hepburn's TV & Radio DX Information Centre - This has information on some LF transmissions as well as HF, VHF, TV and other radio.

Do not miss Sheldon Remington's series of articles: On The Art of NDB DXing from the LWCA website.

DSP Software for digging out the weak/buried signals (These are definitely worth getting!):

Spectran Beta 4 and ARGO - Spectran is the successor to the well-known Hamview program.  This windows version can use any standard (full-duplex) sound card. Like "Spectrogram" this program produces a graphical "waterfall" display and has a real-time audio bandpass filter.  This program is more specifically suited for amateur/radio use as it's display algorithm can more distinctly show "buried" signals than Spectrogram.  The ARGO program is specifically tailored for QRSS - extremely slow speed CW.  It's frequency range is limited to audio, however. (By I2PHD and IK2CZL. Freeware - non-commercial use.)  You can download both of these from the Weak Signals web site.

ON7YD Slow-CW Software - This software (called QRS) is designed to send both CW and DFCW (Dual Frequency CW.)  For more information about this program, go to the description at the LWCA web site.  (Freeware for non-commercial use.)

ONEasygram - This software is uses the "core" .DLL of Spectrogram (above) and provides a user-interface that is better-suited for QRSS use.   (By OK1FIG.  Freeware for non-commercial use.)

Bill DeCarle's Page (VE2IQ) - This is where you can get programs such as CRUNCH, AFRICA, and COHERENT.  Please take a look at the "Readme" files included with the programs before you run them.

KK7KA's WOLF Page - This is where you can get the WOLF program as well as some tips for using it.  Also, look at  Lyle Koehler's Wolf For Dummies page.

Spectrogram - This is a great program that may be used for analyzing audio and VLF/ELF/ULF signals.  (For radio-related applications, refer to the Radio Waves below 22 KHz site.)  This program produces a graphical "waterfall" display showing time, frequency, and amplitude at frequencies from DC through 22 KHz.  (By Richard Horne.  Older versions were shareware, but the current "full" version allows for a 10-day evaluation before requiring a purchase.)

Any comments or questions?  Send an email!

This page copyright 1999-2006 and maintained by Clint Turner, KA7OEI
This page last updated on 20060213

Since 12/2010: