GSBC, RandomRace.ru and radio beacons
As it was mentioned, the very first primary engagement for our team is not a GSBC launches, but organization of true-adventure races here, in St. Petersburg, Russia. The competition is named RandomRace, and it's main idea is to deploy orienteering GSM/GPS-equipped checkpoints to unpredictable positions with helium balloons. Nevertheless, we joined GSBC initiative since the stratospheric launches are also very interesting for us. Surely, we reuse some of equipment for both types of our events.
For example, radio beacons. Here, near St. Petersburg, big areas are well-covered with swamps and forests but, at the same time, they are poorly covered with GSM networks. As a result, time to time, our sondes falling into such areas couldn't send their exact coordinates on the ground, but only latest air position. Looking for a sonde in this case became long and annoying process. So we decided to use some kind of short-range radio navigation while searching sondes. I'm a team member, most qualified in electronics, and I have implemented that idea for us. Naturally, we use the same beacon-finder system for GSBC balloons.
For this purpose, it was decided to use radio location finding so that each sonde carries a radio transmitter and the organizers have a receiver or radio location finder.
Transmitter general requirements:
- low weight and volume;
- single-cell lithium-polymer cell power source;
- autonomous operation at least 24 hours;
- low cost due to a possible device loss.
Receiver general requirements:
- moderate weight and volume;
- directional antenna;
- audible and visual indication of signal strength;
- autonomous operation at least 2 hours;
- detection range of at least 100 m in a forest.
Frequencies and antennas
I had to reject standard radio orienteering frequencies (3.5MHz and 144 MHz) due to dimensions of antennas and decided to use 433 MHz frequency. A transmitter antenna shall be omni-directional and a receiver antenna, on the contrary, shall be a narrow-beam one. For a transmitter there was no wide choice between a spring or PCB type antenna.
Receiver antenna was built of Quad type The overall dimensions are 18*18*8 cm. The forward square is an open circuit vibrator while the backward one, being slightly bigger, is a closed circuit reflector. The receiver circuit board was mounted into the vibrator circuit.
The antenna was made of a car brake pipe and some pieces of organic glass. The organic glass proved too brittle material and quickly broke. For the second time, the antenna non-metal parts were made of plywood to the same dimensions by laser cutting.
For all electronics in the project I have used STM8 series microcontrollers — I like them though they are not highly popular.
A programming unit can be purchased separately but it is easier and cheaper to buy demonstration board like STM8S-DISCOVERY and use it's onboard programmer.
I used Raisonance C compiler and stvd IDE.
The first version of electronics
For the first device version I have purchased modules produced by Telecontrolli.
Modules for transmitters were RT4-433 (later
For the radio location finder I have used RRQ14-433 modules by the same Telecontrolli. The module has two outputs — the received data and the analog RSSI signal. RSSI level is continuously polled by the controller's internal ADC.
For indication there was chosen a two-digit 7-segment LED indicator.
I have used an integral LED driver STP16CP05. This is a 16-channel shift register with a current stabilization on every output.
The indicator is just a common anode indicator of a sufficient size. The sound system was made of broken headphones. The headphones were connected to the true and inverted output of the microcontroller timer via a resistor (to prevent high currents on the microcontroller). The controller firmware is constantly measuring RSSI level, trying to find peaks there while receiving radio signal, recalculating them in some arbitrary units (‘parrots’) and indicating. Also length of headphone beeps are proportional to calculated level. At 0 ‘parrots' the location finder is silent and at 99 ones it is continuously beeping.
The first tests and competitions with the use of the developed ‘find me’ system were generally successful — the location finder made us possible to find two landed sondes in a wood. The antenna demonstrated its acceptable directionality and the sondes could be heard by a walkie-talkie, as expected. A location range was also acceptable. It was unpleasant to find a quite high performance variation of transmitter modules — both a frequency and a power.
For the second version of the system, I used HC-11 modules, found on aliexpress.com. This is the UART radioextenders made of the same STM8S003F3P6 and the subgigahertz CC1101 transceiver produced by TI. The transceiver can receive and transmit data in a wide range of frequencies of 300-900 MHz (roughly) and supports various modulation, speed and power. I used that modules on both sides — as a transmitter and finder radio part.
I performed partial reversed engineering of HC-11 schematics.
Basically, everything turned out as it had been expected — MCU's hardware SPI is connected to the transceiver, UART wired out through level converters, reflashing pins are on the test pads beneath the module.
The module has an LDO, so that the transmitter can be directly connected to lithium batteries 1s with a maximum voltage of 4.2V. The TX UART output I connected to a LED emitting short impulses.
Making a transmitter is therefore comes to soldering just a LED, a power wire and a standard antenna. To prevent wires from damage and the circuit from moisture, transmitters were covered with a hot-melt adhesive and wrapped with heat shrink tube.
The C1101 chip is controlled by the standard spi protocol by means of register reading and writing. There is also a FIFO buffer for packet data exchange. The adjustment of chip settings is not recommended to do by a rule of thumb, but with a standard SmartRF utility software downloadable from the TI website.
The next step is a structure of data being transmitted. On one hand, it was desirable to identify the beacon's sound aurally with a walkie-talkie. On the other hand, transmitted sequence should carry some payload.
The C1101 digital chip first transmits a bit pattern of a code word, then a synchronize word, then a data packet and an optional CRC. I have come up with the following transmission format — about 3 times per second the transmitter sends a series of 5 packets of impulses. Each packet is composed of 2 bit patterns with 3 bytes of payload in between. This is a transmitter’s number, its current power in dB and a check byte — inverted power value. The bit patterns are 101010… and 110110… with GFSK modulation. This sounds through standard LPD walkie-talkie like a two-tone signal of about 300 and 200 Hz that can be easily identified against natural and man-made noise. Each packet is transmitted at a different power of -30, -20, -10, 0, 10 dB. As a man with a walkie-talkie approaches the transmitter, more packets within a sequence start to override noises and the man can hear an extending sequence of signals.
Therefore, this allows for a rough estimation of distance to a beacon by a walkie-talkie, which is inherently inconsistent with a location finder by its FM origin. Impulses are transmitted every 3 seconds and a sequence length is about half a second. The CC1101 in a transmission mode draws 20 to 30 mA depending on a transmission power. An average power drain of the transmitter is thus about 5 mA. We have used various batteries for transmitters, but spare batteries for mobile phones proves the best price-to-capacity ratio. A 1350 mAh Nokia battery can power the transmitter for 11 days. To lower a power drain, the controller after impulse sequence transmission turns the transceiver into a stand-by mode and then turns itself into a stop mode. The controller is restarted by the IWDG watchdog which has its own generator and able to ‘awake' a dead hung-up microcontroller. A dead hangup can not be excluded as a sonde with a beacon can reach a high altitude where a temperature can be up to -60Ñ. Unfortunately, stm8s controllers provide a maximum operation duration of this watchdog a bit more than one second, which is surely not enough. It is therefore necessary to keep in memory a watchdog operation counter and transmit a sequence once per three awakes.
I have used the same HC-11 module as a receiving unit of the location finder, but with a different firmware. It lacks output pins to control both a LED driver and sound, but I already had a half-done first generation location finder with a microcontroller, a driver and an indicator. As a result, the location finder became a 'dual core’ system. One stm8 in the transceiver receives a signal, sends a result via UART interface to another stm8, which, in turn, operates an indicator and a piezoelectric radiator. I have left an antenna intact, removed an old receiver from the board. The new module was soldered to the controller pins and the antenna and fixed to the board with a double-sided adhesive tape. For sound generation, I used piezoelectric beeper.
As already mentioned above, the receiver had two cores and two firmwares. Reflashed hc-11 module continuously reads the RSSI value and checks if a correct data packet is received. The controller reports all its observations via UART interface to the second MCU. The second one reads the data, recalculates RSSI values into some ‘parrots’, forms digits on an indicator and beeps. The indicator shows arbitrary signal level and one digit of beacon number.
Unfortunately, a comprehensive testing was not done due to a lack of time. Everything was tested just 'in the field'. Though, some tests were made in the city. Along with the transmitter, one of the most available LPD walkie-talkies — Midland LTX-325 was used for the test.
The maximum transmitter-to-walkie-talkie range of audibility on a straight line was 600 m (with noise damping off) and 280 m (noise damping on). For sure, the 433 MHz range in a city is badly interfered with car alarms as well as building and security companies’ radios. The receiver indicates 15 units (‘parrots’) all the time.
In a natural environment, base signal level is a bit lower, usually 12-13 units. Usually a radio beacon could be detected with a walkie-talkie from about 300 meters. One of competition participants said he had been able to hear a beacon from 1500 meters using portable universal radio Yaesu. One of our flying balloons was accessible during several minutes after launch. Taking average wind speed about 50 km/h that day, the distance was about several kilometers of open air. Usually a finder detects a beacon at a distance a bit shorter than walkie-talkie, in any case that was about 150-300 meters. Antenna quality surely allows to detect direction to the beacon, while the finder's indicator shows about 12-15 units at a maximum distance of detection, and 80-99 near the target. In rare cases, the indicator showed about 60 near the beacon.
We used this equipment in both events — randomrace competitions and stratospheric ballon launches. The complete beacon-finder system has shown moderate performance. In some situations, it saved us many hours of searching sondes in a swampy forest. All the components are relatively cheap, and the design could be easily replicated by an electronics hobbyst
Firmware sources are published at https://sourceforge.net/projects/randomracebeacon/