paradar is a tiny, handheld ADS-B receiver for paramotor, paraglider, drone and general aviation pilots.
- 36 ultrabright LEDs clearly indicate the direction of other aircraft in a ~20km radius
- Fully self contained handheld receiver and display - can be used without a linked phone or tablet
- Much faster to understand at glance vs an app on your phone
- Optionally feeds high quality GPS and traffic information to SkyDemon or other GDL90-compatible EFB apps
- Easily readable in full sunlight, on the ground or in the air on your flight deck
- Receives ADS-B on 1090Mhz (UK/EU) and 978Mhz (US)
- Colour-blind-friendly colours
- Built-in 1950mAh battery (2.5 hours battery life) with USB-C fast charge
- Only 88mm x 88mm and 175g with antenna
- Open source hardware and software
It can be used standalone (with or without a phone or tablet), on the ground as a handheld compass that indicates air traffic, or in the air on your flight deck to improve your situational awareness.
Paradar is the result of a short two-week project that grew into a monster. I fly in a busy area, and I wanted to see aircraft in the sky around me - and apps designed for fixed-wing pilots are tricky to read while flying an ultralight. Paradar started out as a Raspberry Pi attached to a USB software-defined radio dongle, and some LEDs from Amazon. Six months later, it's now a professionally manufactured PCB and case that fits in the palm of your hand. The journey to get here was long and arduous.
2020-05-03 Not too much to report recently: I'm mostly waiting for manufactured PCBs from PCBWay to arrive. They're making progress, which is good - but COVID-19 is causing huge delays. Assembled boards should ship soon - then I get to do a ton of testing!
2020-03-29 A lot of work this week to finalise assembly, fix bugs after feedback from the beta testers, and start the process of finding someone I can outsource the PCB manufacture to. To put together the first few boards I'm working with PCBWay (click for some neat videos) in Shenzhen, China, who are a large PCB manufacturer and assembler - alongside a distributor in the UK to sell and source the devices. It's been a pretty straightforward experience working with them so far - most of the pain has been around sourcing parts, but we've settled on a final BOM that they say they can deliver.
Obviously the massive global disruption from COVID-19 is causing a few headaches, mostly around shipping and delivery of parts - getting certain microUSB connectors has been a nightmare. However, the cursed package arrived this week (the final component I was waiting on) and now it's possible to assemble the boards together in one unit.
The bottom photo below is of the new optimised assembly. For the beta units, we soldered wires by hand from the USB port to the Raspberry Pi (top photo) - and now there's a neat little custom adapter board to do this!
Some other news - paradar branded antennas! 1090Mhz antennas are really hard to get at a reasonable price outside of a manufacturer - they're fairly specialist - and manufacturers won't sell you less than a few hundred. So we've found a factory and ordered a load. But as a bonus, they'll engrave a logo! Behold the new paradar logo and antenna design:
2020-03-24 Coronavirus disruption aside... my distribution partner in the UK is shipping the first run of Paradars! I have a small number ready to go, and enough components to make up a few more around the ongoing supply chain disruption. Get them while they're hot!
2020-02-25 DHL confirms that the first run of boards has arrived in the UK! With luck, they'll be with my distrubution partner by the end of the week. Assuming there's no unexpected hardware issues, there should then be enough boards and components to make up the set of 10 units for the beta test group over the weekend.
Case design has also been ticking along, and my ghetto towel-covered 3D printer has been working overtime to print up prototypes. So many broken prototypes. 3D design is hard. The good news is that I've settled on a case that encapsulates a larger but less-power-hungry SDR, featuring a sturdier SMA connector. It's a touch larger overall but worth it from a reliability perspective.
The next step is to print some cases in transparent ABS to verify the LEDs can be clearly seen. Not much left to do here!
2020-02-12 After (too) many iterations, the final, tested PCB layout is done 🎉 🎉 and has been sent for an initial manufacturing run of 40 boards.
The case design is close to complete, but requires some tweaking to add holes for the SD card and antenna.
2020-03-24 Some good bugfixes this week from loads of testing.
- Fix an off-by-one error that meant the 36th LED wasn't getting used
- Improve compass stability a ton - by adding a moving average over the sensor readings
- Improve wifi reliability (fix GPIO pullup issues)
- Improve SkyDemon compatibility - send messages slightly more often
Battery life tests out at 2-2.5 hours without wifi, which is probably as good as we're going to get without a bigger battery. That SDR is pretty power hungry.
2020-03-24 Software is complete and well-tested. Test units have been shipped to a small group of beta testers for their feedback
2020-02-25 We're just about feature-complete! 978MHz ADS-B is working, LED colours are now red/green-colourblind-friendly, take-me-home feature is implemented and working. The user can configure the following features via the six configuration switches on the top of the board:
- high_brightness: switch LEDs between medium and high brightness
- wifi_enabled: enable/disable Wifi AP mode (for SkyDemon) - to be implemented
- track_home: enable/disable recording GPS location at startup (or when switch is activated), and indicate the direction with a light blue LED.
- show_north: enable/disable indication of compass north as a white LED on the display
- led_test: enable/disable test mode, which cycles through red/green/blue/white to help with manufacturing (spotting dead LEDs)
- enable_978: enable/disable 978MHz reception. The device only has one radio, so when this mode is enabled it will listen for about 10 seconds at a time on each frequency. Aircraft broadcast their position every 1-2 seconds, so there's a low probability that any will be missed. This is only useful in the USA, where 978MHz is in use.
The main remaining feature here is GDL90/SkyDemon support and testing. It's really exciting to be so close to the finish line!
2020-02-12 The software supports receiving ADS-B on 1090Mhz (EU frequency), indicating it on the display, and mapping the display to the user's position using the compass. The following extra features are planned for the initial release in April/May:
- WiFi base station and support for SkyDemon via GDL90
- Support for US ADS-B (978MHz)
- A ton of Ansible automation to ensure the board runs reliably
- Take a Raspbian image and deploy the software to it
- Convert an SD card between read-only and read-write mode
- Start software on boot
- Watchdog to ensure software is running correctly
- User configuration via the DIP switches on the board
- WiFi on/off to save battery
- Listen on 1090Mhz, 978Mhz, or both (switch between frequencies periodically)
- LED brightness
Board version v1.2 without the USB right-angle connector
paradar is built around the Raspberry Pi Zero, a cheap and fast general purpose embedded computer (1GHz ARM, 512MB RAM) with built-in WiFi. The software runs on an SD card booting Linux (based on Raspbian), which provides drivers for the SDR and supporting software.
Part | Where to buy |
---|---|
Raspberry Pi Zero W | Adafruit, Pi Supply |
paradar main PCB | Order on PCBway, or shoot me an email |
paradar USB adapter PCB (fits the SDR neatly next to the Pi Zero) | Order on PCBway, or shoot me an email |
Nooelec NESDR SMArt v4 SDR | Amazon, Nooelec |
2x20 2.54mm header | RS, Amazon and many other suppliers |
Battery | RS, TME |
Power switch | RS |
3D-printed case | Design files |
The major SMD components are the PNI RM3100 high-resolution compass module, Quectel L86 GPS, and 36x WS2813 LEDs.
The concept here is pretty simple - securely connect a compass, GPS and LED ring (and in version 1.5, a pressure sensor and accelerometer) to the Raspberry Pi. The board incorporates:
- PNI RM3100 high-resolution compass module. Works well in environments with lots of electrical noise.
- Quectel L86 GPS module with integrated antenna. GPS backup power is supplied from the main battery (speeds up GPS startup time)
- A ring of WS2813 programmable LEDs. These are very, very bright (2300mcd white, 480mcd red) and all LEDs can be easily controlled with one GPIO pin.
- 6-way DIP switch connected to Pi GPIO pins, to allow the user to configure the device in the field.
- USB-C power supply with TPS61090 boost converter and MCP73871 LiPo battery charger. Allows powering the unit via USB-C @ 5V/2A (with the battery providing for load spikes if necessary), battery charging, and powering direct from battery. The TPS61090 handles the boost of battery voltage (~3V) to the 5V required by the Raspberry Pi. Includes a resettable fuse and a TVS diode to absorb voltage spikes on the USB line.
- LEDs to indicate USB power connected, charging, charge done, low battery.
- A 500mA LDO regulator supplies 3.3V to the compass and GPS (not connected to the Raspberry Pi 3.3V power supply)
- A pad for a JST 6-GH connector is incorporated on the bottom of the board to permit future support of a FLARM Atom.
- A variety of jumpers exposing GPIOs, power lines, serial & SPI to allow future expansion.
- Mounting holes to suit an undermounted Pi 3 A+ or Pi Zero.
The first mockup featured a few components from Amazon velcroed to a plastic lid. I figured I'd get a PCB made to make things nicer. Then scope creep and silly mistakes crept in... and five board revisions later there's 36 LEDs and a switchmode power supply with QFN20 ICs. I got the SPI MISO/MOSI pins for the compass backwards no less than three times.
The original design used an off-board modular power supply; I decided I wanted to incorporate it onto the main PCB, to reduce weight and size. My friends all told me that switchmode design was hard (as if I was going to listen to that). I threw a design together based loosely on the Adafruit PowerBoost and sent it off to be manufactured. Without bothering to test it first.
Hubris caught up with me: it turns out, board layout matters. In board version 1.3, the ground return from the negative terminal of the output capacitor had a poor connection to the power ground on the boost converter. As a result, sometimes the boost converter would work perfectly: and sometimes it would fry the IC's internal MOSFET after an hour or so. The dead boost converter would short battery voltage to ground, which made things even more sad (read: toasty). Dear TI: this is an incredibly unhelpful failure mode.
I spent weeks debugging this problem, because it presented as a short on the 5V bus that only appeared after some time in operation, about 50% of the time. I was so desperate I considered buying a FLIR camera to look at current flows to locate the problem. But careful deduction and my hot air rework station saved the day (desoldering components one by one until the short went away), and I was then able to show that a wire soldered between the two problematic components reduced the failure rate (though not to zero). In one evening I replaced and then instantly fried 9 boost converters on the same board. They cost $3 each. It hurt.
As a result, board v1.4 features a redesigned boost converter layout. There's some compromises (ground return is across the other side of the board), but I was unwilling to change too much at the risk of introducing another bug. It works so well and I am so happy.
v1.3 boost converter design (left) compared with v1.4. The ground path back to the boost converter is now via the other side of the board: this isn't ideal from a noise perspective, but there weren't easier solutions without a larger redesign.
tl;dr - boost converters are REALLY HARD THEY WILL DRIVE YOU INSANE >:{
- Sourcing components in the middle of a global pandemic is hard.
- Testing a device that detects aircraft - when there are no aircraft flying due to aforementioned pandemic - is also hard.
- Soldering a WS2813 LED sideways shorts +5V to ground, and turns the LED power trace into a resistive heater. A board at 200 degrees Celcius for an hour effectively liquifies a PETg plastic case (on the upside, the power supply worked great)
- Designing against components commonly available from RS in the UK means that assembly in China will be a challenge :'(
- Soldering USB-C connectors using anything at home is a nightmare. I tried two manufacturers and still can't get soldering reliable, using a stainless steel stencil and an oven. I gave up and outsourced manufacturing.
- Scope creep
Wonder what this is for? Hint, not chicken
To get a more compact case, the USB connection needs to make a sharp right-hand bend so the SDR can sit neatly alongside the Pi. Does anyone make a tight right-hand USB adapter? Of course not; people want pink android-shaped USB adapters. And the market provides. So this board connects to both Pi Zero microUSB ports and exposes a compact USB A socket for the SDR.
The Pi Zero's power input port (the one closer to the edge of the board) is not connected - it's just used for mechanical support. The Pi is powered via the GPIO pins connected to the main PCB.
Manually soldering tiny wires. So many tiny wires. Don't do this.
The case is 3D printed and follows the shape of the board. It incorporates a hole to permit easy removal of the SD card (filled with a rubber grommet), slot for the power switch, and a hole for the antenna connector off the SDR. The lid has holes for M3 hex screws, and the base 4mm shafts to hold brass inserts for the M3 threads.
I learned how to use Blender to make this. I use the word "learned" loosely.
Dimensions without antenna are 88 x 88 x 30mm. The antenna connector is SMA, and the included antenna is 110mm in length with a flexible elbow.
Both bottom and lid are 3D-printed in PETg (the lid is translucent). This is resistant to most solvents and fuels, though the case is not waterproof. Maybe in a future version (scope creep!).
The case was arrived at after a lot of iterations. Obviously I justified (I use the word "justified" loosely) buying a 3D printer for this project (my trusty Anycubic Mega-S) and it churned out prototypes, and was a gamechanger in terms of rapid iteration.
The current version of the case features posts to support the switch and support the PCB, plus posts in the lid to hold the PCB down (so it's sandwiched in a fixed position). The SDR (which gets hot) is kept separated from the battery by a small tab. The lid locks to the base to permit easy assembly, and there's a hole to give access to the SD card to permit firmware upgrades in the field.
It took a lot of iteration to get here.
I'm still in really early stages here, and if you have thoughts (particularly if you fly fixed-wing aircraft, ultralights, or drones) I'd love to hear your thoughts. Shoot me an email, or raise an issue on GitHub.
You can find the user manual here.
Please read the disclaimer and license section below. By purchasing paradar, you are indicating that you have read and agree to the disclaimer and license.
I don't sell or distribute these devices - I only maintain the open source project. paradar.co.uk manufactures paradar units, comes recommended by me, and you can grab one on their website! Discounts are available for volume purchases.
paradar does not replace your responsibility to look for, see and avoid other aircraft. This device will not save your life. It is intended only to provide additional situational awareness of some traffic around you when flying VFR. It is not designed for and must not be used for IFR flight, and you must not rely on it when flying in any conditions.
The pilot is responsible for the safe conduct of any flight, and for obeying all applicable laws.
Copyright (C) 2020 Patrick Coleman
paradar is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 3 of the License, or
(at your option) any later version.
paradar is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.