This project is done in cooperation with the Budapest ham radio club HA5KDR
SSTV over ham radio. The computing part is a Raspberry Pi Model A, and a Ricofunk UHF transceiver is used to carry the images over the air. The first beta operation can be watched on a YouTube video. Working installations use a Raspberry Pi camera module to transmit live images, larger resolution photos of the build can be opened by clicking on the thumbnails, the full list of images can be found in a DropBox folder.
- MH Altiszti Akadémia, Szendendre, Hungary
- Frequency: 433.425 MHz
- Modulation: FM / Martin M2
- Timing: every 15 minutes
- QTH locator: JN97MP 86uv
- Coordinates: N 47° 39.28′ E 19° 04.51′
Image received during assembly using speaker/microphone coupling
Image received from installed beacon using a cheap handheld radio and a smartphone from 12 km away
PySSTV to convert images to sound. As of v0.1.9, a (2 minutes long) Martin M1 image is generated in 4. A more effective solution is in progress at UNIXSSTV, this is currently combined with PySSTV. The software pipeline can be seen below, source code can be found in the rpi-sstv GitHub repository.
raspistill --(BMP image)-> PySSTV --(freq+time tuples)-> UNIXSSTV --(raw samples)-> sox
BMP was chosen since it wouldn't make sense compressing an image (wasting CPU cycles) right before parsing it in another component, and discarding the image forever. PySSTV is used for generating frequency-time tuples (e.g. 1500 Hz for 10 msec, then 2000 Hz for 25 msec, ...) as it's fast enough and it's easy to use PIL to manipulate the image, which in our case consists of resizing it and putting the callsign (HA5KDR) in the top left part of the image. UNIXSSTV performs the parts which are not that complicated, but would be too time consuming in pure Python in a constrained environment as the Raspberry Pi, and SoX (more precisely play) is used to divert the samples towards the audio output.
To avoid screwing up the SD card, by default, every part of the FS is either mounted in read-only mode (things like /home and /usr) or uses a ramdisk (things like /tmp and logs). The beacon is activated periodically using cron; even though the Raspberry Pi doesn't have a real-time clock (RTC), it can measure the time since it booted, so if configured to run every n minutes, it will do that, but it will not be synchronized to any clock.
- it exposes the Raspberry Pi serial console to a 6-pin header compatible with the USB TTL Serial cable, using a voltage divider, so the device can be accessed without network connectivity
- it connects the audio output of the Raspberry Pi to the microphone input of the transceiver
- it makes possible for the Raspberry Pi to switch the transceiver into transmit mode by putting GPIO 18 into high state (there could be a resistor between the GPIO and the LED, but the actual LED used didn't require one)
select the appropriate output device if you have HDMI connected!
$ sudo python >>> import RPi.GPIO as GPIO >>> GPIO.setmode(GPIO.BCM) >>> GPIO.setup(18, GPIO.OUT) >>> GPIO.output(18, True)
The final placement of the Raspberry Pi requires to place the radio 10-15 meters away from the transceiver, and a UTP cable is used to transmit signal, TX control and power. The wire assignment is the following:
|orange||+5V power supply||GPIO 1-2|
|orange-white||transceiver PTT/TX||GPIO 6|
|green||transceiver audio signal||3.5 mm TRS jack|
|green-white||transceiver audio signal GND||GPIO 3|
|blue||UART TX||GPIO 4|
|blue-white||UART GND||GPIO 3|
|brown||UART RX||GPIO 5|
|brown-white||UART GND||GPIO 3|
- install more cameras around the city
- and in other cities (countries?)
- synchronize them to send images over a single frequency
- using RTC?
- listening into each others' signals? what about timeouts?
- since the resolution of the cam is much better than SSTV, it could simulate panning
- add 2-3 seconds of silence after and before sending the image
- add (enable) FSKID
- handle pitch dark night
- system info
- long exposure