A PCB for Rxcontrol PIC 18F2550 based ADS-B decoder

OBSOLETE! Please check the new development for the FPGA based decoder.

(C) DL4MEA 10/2010

PICADSB Version 8 (click on it to get a higher resolution top view)

Scope: ADS-B telegrams, as they are sent from aircrafts on 1090MHz, are widely used as source for a passive radar system. While in the past the few commercial units that were on the market appeared quite expensive, Andy presented  here  a receiver which has a quite reasonable price and also a very good performance. The receiver's output, which actually is a signal level indication of 10MHz bandwidth around 1090MHz, requires a decoder which transfers the telegrams into a format that can be understood by common MMI software, such as  COAA PlanePlotter. There are mainly two decoders known, an  ATMEL based decoder solution and the PIC 18F2550 solution by Bertrand Velle, which I choose as basis simply because I have a very handsome PICKit-2 programmer. I also added an RS232 to USB converter in order to interface easier to all PCs, to connect several units to one PC, to provide power and, as it developped over time, also to increase the data rate by a significant factor.

Schematic: (click on the image to get PDF)

Description:

Basically this is the decoder circuit as introduced by Bertrand Velle with some extensions.

In the current design, and as the kits with pre-soldered FT232 are delivered, the power to the PIC is always on. In case that you do not like this but have the power applied just after enumeration, download the tool from FTDI's web page and reconfigure your FT232 to get permission for 250mA right and switch on power after getting permission from the PC.

Input connector JP1
To make clear how the input connector JP1 is wired.

PICADSB7/PICADSB8: The board also includes a comparator circuit around IC30. The reference is built through a low pass filter 470k/56pF. With this, the comparator level is aligned upwards when strong signals are received.
The input signal must be shifted below the comparator level, which is done using the poti R7 and resistor R6. Adjust this for highest sensitivty, means a maximum telegram rate at all.

PICADSB9: With the experience of the two versions we found that the comparator's reference doesn't raise quickly enough at strong signals. The combination of D31/R33 now does this much quicker, and so does R31 discharge with around 10 times longer time constant. Additionally the biasing is done in a different way now, removing the DC part from the input signal with C12 and re-biasing below the comparator's reference level through an adjustable DC voltage. This will keep the amplitude of weak signals unchanged!  This is also close to the improvement suggestion for ADS-B reception which is telling to set the comparator level to the top signal level minus 6dB. More about this further down on this webpage.
I regularily see planes in a distance of 220nm with this decoder, and do not have a doughnut effect for the planes passing by for the local Munich airport (and if I say close, they are at around 2nm distance). 

Place plan and bill of materials: (click on the image to get PDF)

 The new PICADSB8 board has some minor cosmetic changes:

• The LEDs and the USB connector are aligned on the long side of the board, so it can easier be mounted into a box
• There is a Push-Pull comparator used, which will provide a better signal to the PIC (this is part of the semi-kit)
• The LED current limiting resistors 680R are SMD and the 100nF blocking capacitors (also part of the semi-kit)
• The two additional resistors that avoid the oscillator problem are part of the PCB
• Instead of the LM311 a new Push-Pull MAX942 is used (later removed, did not improve reception)

PICADSB8 top components

PICADSB8 bottom components

• Returned to the LM311 comparator, which is much cheaper. The pullup was made 820R for a faster raising edge
• Cosmetic change: Leaving a little bit more space for the potentiometer R7

PICADSB9 top components

PICADSB9 bottom components

PICADSB7 PCB top

PICADSB7 PCB bottom with FT232

FT232 soldered to bottom

If you are using the standard serial interface with MAX232, setting up the connection to the PC is pretty common and shall not cause any problems.

In case that you decided to use the Serial over USB, the operating system will probably ask you to install an USB driver when connecting the unit for the first time. You will find the on FTDI's web page here. After you have installed the driver, open the device manager and check which COM port became assigned to the new unit. That is the number you have to use in your MMI application.

In case that you connect the unit for the first time, for testing purposes, open a simple terminal application to the COM port that PICADSB uses. (Most users have Hyperterm (which cannot handle 1MBit) available, but there are also Teraterm or Putty).

Jumper S1C selects the serial interface parameters:

The RxControl software supports several data formats. The AVR raw format (Type "3") is are just a hexdump of the received frame. With this format the decimal conversion of the received data is done in the MMI software, for example PlanePlotter. On the terminal, it looks like
*8D484395990052B2E80B027D687F;
*8D3C65039990FB9E286404A0043E;
*8D4BAA499945161668800268C26E;
*8D3C662B903D006914A84A707D6A;
*A028023C2010C231595820AA9001;

In case that you want to see the messages in the RxControl clear text decoded format (Type "1"), you have to enter your own coordinats in HEX format to the PIC. Here is how you convert your location into hex:

Now, in the terminal application, press 'P' button (capital letter!) and enter the LON and LAT values from these hex values.
Then you will get an output like this:
+,0,17,3C1193,,,1,49.222290,11.221088,20700,,,,,,;
+,0,17,4BAA4B,,,1,47.995777,12.357646,34975,,,,,,;
+,0,17,40600E,,,,,,,,-64,333,-249,,;
+,0,17,3C4984,,,1,48.425453,11.477677,8850,,,,,,;
+,12,17,40600E,,,1,48.533550,11.472689,35025,,,,,,;
+,0,20,300096,,,,,,4525,,,,,,;
+,15,17,40600E,,,,,,,,-64,333,-249,,;
+,10,17,4BAA49,,,,,,,,0,-278,183,,;

What is so special with your AD converter?
The AD converter is using a very short time constant in order to create the comparator's reference and herewith somewhat aligns to the signal strength, strong signals will be decoded as well as weak signals. I live in the approach path to Munich airport's runway 28L and some planes are even flying over my house. These are decoded as well as planes up to a distance of 200nm. There is no doughnut effect.

How can I distribute processing load in the best way?
The RxControl software does not receive and decode in parallel due to processing power. Let Planeplotter do the decimal conversion, use the AVR raw format (format '3').

Click on the pictures for higher resolution (drawing from orignal web page, in order to copy before it disappears there)

A final note for manufacturing: Take a long wire and put markers for the sections on it: 136mm -> 65.5mm -> 205mm -> 65.5mm -> 190mm.
Then first roll the inductors (1.5 turns onto a 9.5mm drill) and second do the 90° bends in order to get it straight.

Note: Add a 10kOhm SMD across the input if you are using such an antenna which is not grounded in order to prevent ESD descruction of the SAW and the first amplifier!

Another good antenna, which is easy to build (you may leave away the loop with acceptable degradation) can be found on Edward's page.

Green: ADS-B signal Magenta: Comparator reference Yellow: Output of AD converter It is clearly visible that the PIC, which samples the input at 25% and 75%, reads a '1' at both times and so does not recognize the sync at all.

As you see, the RC combination was made a little bit faster, using R31 (47k) and C32 (82p). But it will even be charged faster using the diode D31 and R33 (4k7). Due to the voltage drop across the diode the reference always stays around 200mV below the voltage peaks. The result can be seen on the right side. The reference level (in the middle) raises quickly and drops slowly. Even in the middle of the signal, it is defined fine in the upper half of the signal; a methode which is also suggested in some official documents. This modification can be done very easily by exchanging components on the PCB and soldering D31 and R33 across R31. NOTE: The modifications are shown for an LM311 comparator. Pins at the MAX942 in PICADSB8 are different!

1. For this modification I suggest to bridge R6 in PICADSB7 and PICADSB8.
2. Also, you have to cut two traces: The one that goes from the signal input (JP1, Pin1) to the poti (solder side), and the wire from the poti's slider to the comparator (PICADSB7: solder side, PICADSB8: top side, below the poti).
3. Use a 100nF (prefered axial -like resistors-) capacitor and connect the input to the comparator input
4. Use a 470k resistor and connect the potentiometer's middle pin to the comparator input.
5. Use a 4k7 resistor and connect the potentiometer's pin that was connected to the signal input before to any pin that carries +5V (or use a leakage at the bottom copper plane)
6. As a check: with a DC voltmeter you now must be able to set the DC voltage at the comparator's input from 0V to 2.5V.
   

This shows how I modified the PICADSB9 PCB in a semi-perfect way: R31 is replaced by a standard 47k resistor, the 1N4148 diode is soldered just over it, between the cathode of the diode and R31 there is a small SMD 0805 4k7 recovered from miniADSB because I only install the one between Pin 7 and Pin 8 and ommit the pullup and protection resistor there. Besides the SMD resistors, you can also see the PCB trace cut off that leads from the poti's wiper to Pin3 of the comparator.

Last updated 04.10.2010 07:00 ((X) home () Munich, St. Martin Straße, S-Bahn station) by DL4MEA