Thursday, June 26, 2008

"Digital" Mode of the Week: HF FAX

FAX stands for facsimile, or in this case radiofacsimile. Unlike the more recently developed office FAX machines used over wire telephone circuits, it is completely analog, using FM emission (F3C). It dates from work done by various inventors in the late 19th and early 20th centuries to send pictures over telegraph and later radio.

HF FAX was very widely used in the 20th century to send news photos and even entire newspapers over HF point to point circuits. Machines used huge drum scanners and similar drums to print at the other end. They were very precise, and very, very expensive. Now it can be done with simple computer sound card software, though at considerable loss of precision in this rather fussy mode.

Today's HF FAX uses a continuous FM carrier with a narrow 800-Hz deviation. Usually, anything around 1500 Hz, the lower limit, is reproduced as black, with the brightness increasing until it reaches a white point up around the high limit of 2300 Hz. By ear, fax can make a number of different sounds, but usually it makes a cyclic pulsing with a high-pitched shreik underneath.

Nowadays, instead of using true FM transmitters, the corresponding baseband audio tone is sent to a normal, single-sideband, suppressed-carrier, HF transmitter. The result sounds the same, only displaced in frequency due to the offset from the missing carrier. FAX is tuned in USB, or else it will reproduce as a negative, and the dial should read 1.9 kHz (the tone center) lower than published frequencies. Most of the time, one centers the waterfall display between two marks, and it's tuned in. Slight mistuning has no effect except to possibly wash out the blacks or whites, depending on which direction it is.

Two other parameters are essential for the reception of FAX. The first is called drum speed, though obviously nowadays transmission speed would be a better name. It's expressed in lines per minute. The receiving software must sync to this speed, or the pictures will not reproduce properly.

Far, far the most common speed is 120 LPM. It is used for nearly all weather FAX. Some more complex pictures are sent at 60 LPM, a speed which can sound like a time signal.

The second parameter is Index of Cooperation (IOC), expressed as an integer. This is a really arcane thing also dating from the use of drum scanners. It measures the resolution of the FAX. Nobody really has to know what the IOC measures, just what the number is, and it is nearly always 576. On very rare occasion, it's 288. This is way less important for computer screens than it is for rotating drums.

Those who are really curious as to just what the Index of Cooperation is will be pleased to know that it's commonly defined as the product of the total line length and the number of lines per unit length, divided by Pi. It's measuring the ratio of how much distance the scan steps for a new line to the diameter of the drum. The length of time any single FAX can last is determined proportional to the IOC. At 120/576, the standard weather format, this is 18.8 minutes, which is plenty. Going to 60/576 doubles this, along with increasing resolution, though of course the fax takes twice as long to send.

Now it's all clear, right? :-)

Most FAX software allows the setting of black and white or continuous tone (grey scale) modes. Except for satellite weather images, just about every FAX sent on HF nowadays is in black and white. I'll get some argument here, but I've always just kept it on continuous anyway. Otherwise lines get kind of jagged, and typical HF ionospheric multipath fuzz can get grungy looking in a hurry. Plus you have to change it whenever a satellite image is sent, and then change it back again.

Radiofax typically uses a standard called APT (Automatic Picture Transmission). This allows the unattended reception of faxes. Different bands and services use different versions of this, but the one we're interested in is as follows:

5 sec Start Tone (alternate black and white at 300 Hz)

Phasing (30 seconds of white with one black pulse per line)

Image (may have a white interval for sync)

5 sec Stop Tone (alternate black and white at 450 Hz)

Optional 10 sec black

Different agencies have slight variations on this, and these are usually just enough to confuse amateur software. Even so, one will notice that all the information needed to set parameters and receive faxes is here, except maybe for IOC and that's almost always 576 anyway.

Why more computer programs don't take advantage of this is beyond me. They don't however, and so your received fax is usually displaced (left edge is somewhere else in the screen) and slanted (computer is not printing lines at the exact speed of transmission). There are nearly always ways to fix these, either during or after reception.

Slant is way the more annoying of the two, because computer sound cards really aren't up to the task. They don't have the tight frequency tolerance that an expensive dedicated device would have. Typically, the proper slant compensation has to be found manually for a given station, and then it will vary for others, or even change during a long FAX.

I like FAX because it's radio with pictures. One learns how to read the weather charts pretty fast. They're not that different than the ones on the Weather Channel. Also, the satellite images are as good as the ones on your local TV, and better if you get into colorized WEFAX downlinking on higher frequencies.

Many coast guards and weather offices still send hours of these a day, and since they are being used by boats for serious navigation and safety of life at sea, they won't go away any time soon. Check the various lists and radio loggings for something audible in your area.

More from Charles Brain on the Vista Problem


Hello Vista users,

I have done a Google search and it seems that the problem is due to the fact that Microsoft have moved the sound drivers from Kernel space to user space. This has caused the hardware acceleration used by some cards to stop working. Accurate information on this problem is very difficult to come by.

It seems the reason that Microsoft did this was to protect Vista from errant sound card drivers not under their control. They were getting fed up with being blamed for crashes that weren't their fault. There are some other reasons for wanting to move the drivers to user space I have been told.

At the moment I can't see any short term fix to this problem as it appears to be outside my control. I understand that this problem is causing a lot of grief among P.C gamers too as they have lost some of their special sound effects.

I am guessing here but what I think is happening is that when I ask for a 48K sample rate the card is returning something else (possibly 44.1K) and that is why it does not work. I base this guess on the fact that I can get PC-HFDL to start on a Vista laptop and display hfdl packets on the spectrum display (indicating the sound handling is working) what it does not do is decode the actual packets (indicating a sample rate problem).

I am sorry I can't be any more helpful/hopeful than that.

- Charles

Thursday, June 19, 2008

Annual VLF Alternator Transmission Coming Up

The Alexanderson Alternator was the first continuous-wave CW transmitter. It is literally a large electric motor-generator that makes RF. A global network of these huge stations was rapidly replacing "King Spark" for maritime coastal radio, when vacuum tubes replaced both.

A working example is at SAQ, Grimeton Radio. RF power is on the order of 200 kW, but the frequency is so low (17.2 kHz, not MHz) as to make it a real catch in North America. From the web site in Sweden:


The annual transmission on "Alexanderson Day" with the Alexanderson alternator on VLF 17.2 kHz will take place Sunday 29th June 2008 at 09:00 UTC (tuning up from after 08:00 UTC) and will be repeated at 13:00 UTC (tuning up from after 12:00 UTC).

The station will be open to visitors.

Amateur Radio Station with special event call "SK6SAQ" will be QRV 09:15 -12:00 UTC on the following frequencies:

- 14.035 kHz CW
- 14.215 kHz SSB

>From 07:00 UTC also on:

- 3.755 kHz SSB.

QSL via SK6DK or direct to address below.

QSL-reports to SAQ and SK6SAQ are kindly received:

- E-mail to:
- or fax to: +46-340-674195
- or via: SM bureau

- or direct by mail to: Alexander - Grimeton Veteranradios Vaenner,

Grimeton 72

Charles Brain: Bad News for PC-ALE Under Vista

From the hflink group on Yahoo:

Hello Folks,

Well I have done yet another Google search to try and find why programs like PC-ALE won't work properly under Vista.

All I can find is that the Microsoft Vista sound development team have completely re-written the Sound subsystem, they have moved it from Kernel space to User space. Apparently the reason they did this was because Microsoft was being blamed for OS crashes by errant sound drivers that were not their fault.

The side effect of this change has been to disable hardware acceleration in a number of soundcards most notably Creative Devices ones. Creative got around the problem by using their version of ALchemy.

I have not tried this but it might help.

Unfortunately this problem with Vista goes places in Windows that I fear to tread so unless some genius comes up with a simple fix Vista support for PC-ALE will never happen. It seems like the only cards that stand a chance of working are the simplest ones with no hardware acceleration.

Looks like my next project will be Linux based!

- Charles

Wednesday, June 11, 2008

Baltops Participating Units

From German Navy:

The multinational Baltic maneuver BALTOPS 2008, a total of 47 international units from 13 different nations.


FGS U 15
24 Minen
1x PC-9 Pilatus
1x MPA











Vereinigte Staaten von Amerika
1x MPA

NATO Baltops Exercise Under Way

5102.5 lit up with US and German participating units. Norwegian navy also heard. Expect tactical comm on many European frequencies. NUKO formatting and unit calling NAA. Sounds like old times.

Friday, June 06, 2008

Digital Mode of the Week: PSK31 (Part 2 - Phase-Shift Keying)

PSK31 was developed in 1998 by Peter Martinez, a British ham with the call G3PLX. It was intended to improve on RTTY for the purpose of direct, keyboard-to-keyboard contacts and rag chews. PSK stands for Phase-Shift Keying, and 31 refers to the baud rate of 31.25. This sounds slow, but it is perfect for hand typing, and it was also easy to count down from the 8000-Hz sample rate common for sound cards of the era.

Unlike many digital modes, which continued to be somewhat esoteric for ham use, PSK31 caught on immediately. From very early on, it was extremely well suited for generating and decoding with user-friendly, sound card software on ordinary personal computers. In use, the mode proved to be very efficient in its use of power and spectrum, allowing hams to communicate on HF without the equipment taking over their lives (and pocketbooks).

Right now, one can tune to 14070 kHz any time the band is open and find multiple signals all warbling away. Often it sounds like an attack from extraterrestrial science fiction insects. Less active frequencies are 10138-40 and 7035.

While PSK31 can have different modulation schemes, the one most commonly used is Binary Phase-Shift Keying (BPSK). Data is transmitted by commanding the sound card to shift the phase of the baseband audio signal by 180 degrees, in effect inverting the polarity.

Ordinarily, this would create a tremendous key click, but the amplitude is synchronously shaped by cosine modulation at the same time. The result is a best-case bandwidth for a PROPERLY MODULATED signal that is theoretically equal to the baud rate - yes, 31.25 hertz!! That is narrower than "hard" on-off keyed CW at a comparable 50 WPM.

Of course, things in the real world are somewhat more complicated. As can be easily seen on the waterfall, many signals have varying amounts of clicking or IM audio distortion, and they are maybe 40 or 50 Hz wide. This gets worse if the operator is hitting it too hard and causing more sidebands to appear. For the most part, though, PSK31 is one of the narrowest modes ever designed for hams.

The narrow bandwidth and low data rate make PSK31 practically immune to selective ionospheric fading (that Moog whooshing sound you hear on wider modes like HFDL and STANAG 4285). It performs less well under phase and Doppler distortion such as from aurora.

In this keying system, any phase flip signals a 0 bit. If there's no shift at the right time, it's a one bit. As we've seen, the state doesn't matter, it's the change in state. Again, it can be tuned in USB or LSB, though USB is traditional.

A character space is two zeroes - 00. This is why all the characters end in 1.

The PSK31 idler consists of all zeroes, which flip the phase at the baud rate. This and the cosine shaping produce what is basically double-sideband suppressed-carrier emission (J2B). ITU refers to it as 60H0J2B, in the long international designators. This means suppressed-carrier modulation for automatic reception using a maximum channel width of 60 hertz.

The transmission starts with a distinctive railroad track pattern on the waterfall. Bursts of hand typing create something looking more like a DNA molecule. At the end of the transmission, it all collapses back into the single tone.

Receiving consists of syncing to the baud rate, which is determined from the signal, and sending the resulting data to a Viterbi decoder. There are no connections, and no error checks. Stations take turns sending, just like RTTY. Character hits print gibberish, again like RTTY. However, the copy is usually much better on HF, everything else being equal.

There's not much else to know. BPSK31 is very uncomplicated to the user. You click on a signal and copy comes out.

PSK31 immediately gave rise to a bewildering array of phase-shift keyed, character-based, half-duplex modes. If you've used MultiPSK you know what I'm talking about. Each new version has more little buttons and more funny noises than the last one. Since decode sync is determined from the signal, there's also an autodetect mode.

An early variant that should be in just about all PSK31 software is QPSK31. This stands for quaternary phase-shift keying. On the standard computer phase constellation display, you will see four points instead of two. This is done through the use of a second BPSK carrier shifted 90 degrees (in quadrature), driving a second receive demodulator.

While this should double throughput, the extra capacity is used for an error check and more robust decoding. The performance of this mode on degraded circuits is impressive, but at the cost of more bandwidth and a tuning precision challenging even modern solid-state radios.

Other BPSK and QPSK modes are at 10, 63, 125, and 250 baud. There's also PSK220F, a 220-baud mode well adapted to broadcasting. It is used by the Cuban numbers operation at times.