---

Techniques used in this broadcast 📋 🤓

HamDRM features various transmission parameters that need to be selected for best results:

Mode: A, B or E
Mode A is for channels with no selective fading (groundwave/upper HF/VHF/UHF). Fastest.
Mode B tolerates selective fading well - generally the best for most HF use. A little slower than Mode A.
Mode E tolerates selective fading with significant doppler shift. Slower than Mode B.
(Mode E is based on DRM Mode D. It is not the Mode E of DRM+)

MSC protection level: Normal or Low
This defines the amount of error protection used. Low is faster, but has less error protection.

Interleave: 400ms or 2s
400ms has less delay, but this doesn't matter for sending files.
2s helps the decoder work better during long fades.

There are some other ways to help increase the chances of successful HamDRM data transmission on shortwave:

• Code the data efficiently, and use data compression techniques. Less data allows slower modes and more error correction.
• Use lower numbered QAM where possible - like QAM4. Lower numbered QAM tolerates noise and distortion better, but is slower.
• Use the heaviest error correction that time allows.
• Process the modulated data signal to increase it's average power level using PAPR (Peak-to-Average Power Ratio) techniques (peak limiting and filtering). Lower numbered QAM modes tolerate a higher degree of PAPR processing distortion, and can be processed more heavily for more SNR gain. QAM64 starts to be degraded by only 1dB of clipping.
• Compensate for non-flat transmitter EQ - Many shortwave transmitters use processing that applies EQ boost to the treble and bass ends of the spectrum. The COFDM data modulation needs to be as flat as possible so all of it can be recovered. If a transmitter's response is known, it is possible to at least partly compensate for it by applying inverse EQ in production. This is always done last.

QAM type	PAPR dB used	Notes
QAM4:	6dB	Tolerates up to ~8dB
QAM16:	3dB	Tolerates up to ~4dB
QAM64:	0dB	Tolerates up to ~1dB

Links of interest:

https://en.wikipedia.org/wiki/Data_compression

https://en.wikipedia.org/wiki/Reed-Solomon_error_correction

https://en.wikipedia.org/wiki/Quadrature_amplitude_modulation

https://en.wikipedia.org/wiki/Crest_factor

YouTube - Reed Solomon Encoding - Computerphile

Luckily, we can all use these techniques without needing to be math experts! 😀

---

"Digital Weatherfax" demo: 🌀 QAM4 and RS3

Here is a demonstration of how these digital methods could be used for improved weatherfax broadcasts.

A weatherfax (WEFAX) is a weather chart sent by analog facsimile (fax) over radio.

Weatherfax is an old analog standard that has been used for decades to send timely graphical weather information over HF radio - primarily for maritime and aeronautical use.

Weatherfax is coded as an FM audio tone that varies in frequency to represent brightness information. The most common speed of transmission is two lines per second.

HF weatherfax is often plagued by degradation from multipath propagation and noise, which damages the charts - sometimes making them illegible. To help combat this, most services are simultaneously broadcast on a range of frequencies to allow reception on the highest frequency that will propagate to the receiver. Using higher frequencies results in fewer ionospheric signal hops, and reduces multipath effects. This is especially important when the receiver is at a great distance from the transmitter.

Analog image modes are very inefficient, because they scan the image at a constant rate. A typical weather chart fax spends most of it's time transmitting the white background of the image. This is wasteful of both bandwidth and energy.

Digital methods allow efficient information coding techniques, data compression, and error correction to be used. This avoids transmitting useless information, and instead allows valuable error correcting data to be sent. Digital modems rely on the signal being a certain number of dB above the noise floor before decoding is possible - and fading can frequently take the signal below this decoding threshold, causing data loss. Good error correction allows complete decoding of data, even when a significant portion of it is lost.

Weather charts are high resolution line drawings that are originally drawn in vector graphics format. The browser SVG vector graphics format is an XML text file mode that describes the image elements as geometric shapes. This is more efficient for line drawings than a bitmapped image, and text files also compress extremely well. This allows a complex image to be saved using relatively little data, and the image can be scaled up to any size while remaining sharp and clear.

This SVG image was converted from a vector-based PDF image, and then "cleaned up" to remove unnecessary content using both free software and a free online tool. This was an important step to reduce the data size. The chart remains accurate after this process, and this is easy to check by graphically differencing it with the original image file.

This particular colour weather chart was 165kB in PDF format, 187kB when converted to uncompressed SVG format, 165kB after SVG optimization, and 41kB after LZMA compression. In contrast, the LZMA compressed PDF was 125kB.

Example weather chart from the Australian Bureau of Meteorology website:

The RS3 mode used here reduces the data transfer speed to 63%, but tolerates up to 36% data loss.

Links of interest:

https://en.wikipedia.org/wiki/Radiofax

NOAA weatherfax charts

Australian BoM weather charts

Weatherfax video

Please send reception reports and screenshots to: qsl@rnei.org

---

Slow and steady: 🐘 QAM4 with RS3

QAM4 has two states of amplitude, and two states of phase information. It tolerates noise and distortion very well. These images should be easy to decode, but the practical image data size is limited due to the lower data rate. These next two images have a low data size because of their large blurred areas, so they could be sent in high resolution. Using RS3 allows a bit more speed than the super-robust RS4 that was used for the HTML.

Image 2
An Arabian Red Fox in the Kuwait region:
https://newatlas.com/photography/nature-infocus-photo-contest-winners-2021-gallery/


Image 3
An ant carrying water drops on a piece of thread appears to be carrying the earth:
https://newatlas.com/digital-cameras/agora-top-50-photographs-of-2019/

Please send reception reports and screenshots to: qsl@rnei.org

---

Revving up: 🚗 QAM16, RS3...

QAM16 has four states of amplitude and four states of phase information, and is almost 1.8 times the speed of QAM4. It needs a higher signal-to-noise ratio to decode correctly, but the RS3 coding will help to avoid lost data. The faster data flow makes it more practical to send larger files.

Image 4
Sand wasp with prey:
https://newatlas.com/digital-cameras/agora-top-50-photographs-of-2019/

Please send reception reports and screenshots to: qsl@rnei.org

---

Let's try some fast data... 🚀 QAM64, RS4

I hope you have a low noise floor!

QAM64 has eight states of amplitude and eight states of phase information, and is about 1.5 times the speed of QAM16. It needs a good quality signal to decode correctly, but the RS4 coding will help bridge over up to 49% lost data.

QAM64 is the fastest HamDRM QAM level, making even larger files more practical to send. If the received signal-to-noise ratio is high enough most of the time, there will be adequate data recovered for a good decode.

Hold on to your hat! 🎩

Image 5
Tractor in a field in Woodburn, USA (or... "If Rose had a tractor...")
https://www.pexels.com/photo/photo-of-ride-on-tractor-during-sunset-693857/


Image 6
Nye County, USA
https://www.pexels.com/photo/road-heading-towards-mountain-490411/

Please send reception reports and screenshots to: qsl@rnei.org