"Mobile Amateur (Doppler) Rain Radar" Project
Is it possible to make your own Doppler rain radar? Yes, I think
so but nobody did it until yet. It is my target to build a mobile amateur
rain radar for 10.4 GHz under 2000 Euro. 3 cm rain scatter is very similar
and OE5VRL (Rudi) gave me some ideas with his rain scatter detector http://www.dl6nci.de/oe5vrl.htm
The Doppler effect, named after Austrian physicist Christian Andreas Doppler who proposed it in 1842, is the change in frequency of a wave for an observer moving relative to the source of the wave.
Christian Doppler was born in my city Salzburg.
1. Time schedule & status
- March 2010 : Start of project
- June 2010 : Made my radio license
* finished with my call OE2IGL
- Summer 2010 : Finishing system concept and calculations
* finished (see system overview)
- Summer 2010 : Finishing prototype computer and µ-controller software
* first tests with PIC32 and C-compiler finished
* computer software "dBZ calculation" with USB (via RS232) connection finished
* computer software "Radar visualisation" finished
- Spring 2011 : Finishing prototype feed/dish/RSSI & software
* homemade 10.4 GHz feed finished
* prime focus dish mounting on tripod finished
* µ-controller software with high speed ADC (0.5 MSps), USB connection, LCD display finished
* RSSI circuit, impedance converter finished
- Summer 2011 : Finishing of conical feed/dish/receiver without azimuth rotor
* µ-controller software with 12 bit HH-12 encoder, call sign generator
* modified SAT LNB downconverter
* mounting head and rotor concept finished in more detail
- Summer 2012 : Module tests of Rain Radar system with azimuth rotor
* azimuth rotor with worm gear, H-bridge and HH-12 encoder and µ-controller finished and tested
* power supply (12V accu) finished and tested
* Kuhne beacon transmitter modified (10.402 GHz and pulsed) and tested
* Receiver, RSSI and µ-controller tested
* Dish/conical feed/LNB at 10.402 GHz tested
- Autumn 2013:
* Test with ready finished downconverter instead of modified LNB
- Spring 2014:
* Optimized feed for offset dish
* Test phase of system (dish-feed-circulator-transmitter-down converter-receiver-RSSI-µcontroller-computer software)
- Autumn 2014:
* System analysis/measurements/improvements
* Upgrade from 220mW to 4W transmitter power
- Planned summer 2015 : Start operation of my Rain Radar with "radar vis" software and azimuth rotor
2. System overview
3. Technical details
Frequency : 10.402 GHz, vertical polarisation
Dish diameter : 0.82 m
Dish gain : 37.0 dBi
Dish beam width : 2.5 °
Dish rotation : 2 per minute
Azimuth resolution : 0.9 °
Time per azimuth step : 75 msec
Echo time : 2 msec
Pulse : 8.88 µsec
Effective radiated power : 40 dBi
Sensitivity : 0.5 mm/h at 100 km
4. Details of modules
10.2 - 10.8 GHz circulator
- Isolation: 30 dB
- VSWR: 1.12
- Insertion loss: 0.35 dB
Homemade 10.4 GHz conical feed (wave guide)
- Copper tube with inner diameter D: 20.1 mm
- VSWR adjustment by moveable back wall
- Wave guide formula:
1 / Lamda02 = 1 / Lamdac2 + 1 / Lamdag2
Lamdac = 1.706 x D = 34.3 mm
Lamda0 = 28.82 mm (10.402 GHz)
-> Lamdag = 53.2 mm
- Wave guide drawing:
I use this feed to transmit and receive signals in combination with a circulator. Therefore return loss must be as low as possible (< -30dB). After first measurements of return loss and adjusting back wall with the screw I got around -20dB. Then I adjusted the length from the open end to the pin to 54mm (1x Lamdag). Now the return loss is better than -40dB.
Return loss of 45 dB (lower curve) @ 10.42 GHz, >42 dB @ 10.402 GHz
- For prime focus dish with f / D = 0.36
Closed end has a moveable back wall to adjust return loss.
- Corrugated horn feed (part of a standard sat tv LNB) for standard offset dish with f / D = 0.67
Closed end has also a moveable back wall and a moveable
tube between probe and horn to adjust return loss.
Return loss more than 50 dB.
- Dual mode horn feed for standard offset dish with f / D = 0.67
This type gives the best dish matching.
- Optimized dual mode horn feed for standard offset dish with f / D = 0.67
Now this dual mode horn is my standard feed. Johannes
Falk DC5GV has simulated this feed with CST microwave for best performance
and measurements confirm his simulations.
It is robust, easy to adjust for best matching (with M2 screw), fits into standard LNB holder and is optimized for offset dish.
Microwave transmitter 10.402 GHz
This is the original Kuhne 10.368 GHz beacon transmitter with approx. 220mW.
As this transmitter has only F1 (frequency) modulation
I looked for simple A1 (amplitude) solution to switch on/off my pulse
radar. To my surprise I found a p-MOSFET transistor (4 pin IC on the left
above the output stage) doing this on the board. After little changes (see
red marked area) and
an additional n-MOSFET transistor it is possible to use F1 input as A1 input.
In addition last transistor multiplier stage (before power amplifier stage) also has such a n-channel/p-channel combination to switch transistor power. Otherwise you will see a small signal from oscillator stage during pulse off periode at the 10.4Ghz output. Don't switch off/on oscillator stage!
Here is a Multisim simulation and the real oscilloscope image. Yellow line = µcontroller pulse, blue line = TX pulse (power supply of amplifier).
Switch off edge of p-MOSFET is delayed and not sharp-edged. This could be adjusted with R3.
Improved version: better pulse shaping of output pulse (blue signal), rounded rise and fall edges.
Instead of the original 108.009 MHz quartz crystal I use a 108.354 MHz quartz crystal by Krystaly in Czech (thanks to OE5VRL for organizing). I only had to adjust the oscillator and then the output shows 10.402 GHz with 250 mW. Behind the quartz you can see the little quartz heater board.
Downconverter 10.402 GHz -> 652 MHz
made good experiences with standard satellite LNB's. They have low noise
figure, high gain and are very cheap. My first tests were done with such a modified LNB.
At first I removed the horn and then I made a hole into
the waveguide opposite the probe. A short semirigid coax cable was soldered
to the probe and ground.
The spectrum analyzer shows 10.7 MHz I.F. output of the AR8600 receiver. The peak in the middle is my 10.402 GHz signal running through the modified LNB.
In March 2013 I found first LNB with a PLL instead
of a resonator pill. Stability of frequency should be much better with a
The internal 27 MHz quartz is multiplied by 361.111 inside the RD3560 chip to get L.O. of 9750 MHz.
Big disadvantage of all modified LNBs is a poor noise figure at 10.4 GHz. Bigger than 2.8 dB against 0.55 dB at 11.5 GHz, measured with cold sky - hot sun method.
Now this is my standard downconverter and a noise figure of 1.2 dB is doable. Roberto Zech DG0VE has good converters, e.g. the KON-DWN 97107.
As I will not switch off the downconverter/receiver stage during Tx pulse the downconverter detects the reflected Tx pulse of the feed horn.
3 amplifiers in front of the mixer are good for my weak echo signal but gain is too high for my strong dircet Tx pulse signal. Therefore I removed the 3rd amplifier (red one) to reduce overload of mixer.
green = sub-harmonic diode mixer
red = amplifier
blue = low noise amplifier
Received pulse spectrum after down converter with 18µsec Tx pulses.
Echo signal against cold sky (no cloud, no rain, no hindrance). Peak is reflected Tx pulse from horn feed (first 20µsec) and after 20µsec there is only noise level.
vertical -> signal strength of echo
horizontal -> time/distance
FINAL COMBINATION (without Tx amplifier): conical dual mode horn - circulator - Rx down converter - Tx transmitter
To get more Tx power I use a 10.4 GHz PA with 4 W output made by Dirk Fischer DK2FD. That gives additional 13 dB.
PA is switched on/off with same n-channel/p-channel combination as used in beacon transmitter (switching drain power of amplifier).
AR8600 scanner with 10.7 MHz I.F. output and external RSSI (logarithmic received signal strength indicator)
A RSSI can be built with e.g. SA627 and buffer amplifier
to drive A/D of µ-controller. SA627 has a fast rise and fall time
of 1-2µsec and a RSSI range of 90dB.
RSSI (mV) of SA627 chip versus RF input (dBmW):
PIC32 ethernet or USB starter kit could be used. This controller has a speed of 72 MHz.
USB starter kit with I/O expansion board and LCD/4x4 keypad
Sampling pulses every 2.22µsec and echo time up to 2ms
µ-controller, RSSI meter, LCD, push buttons, motor driver in one box.
Pulses every 2ms in Tx mode.
Mounting head for azimuth and elevation
First tests with a small dish and motor controller.
Finished azimuth rotor with elevation sensor (small black box), azimuth encoder and tripod.
I use this tripod also for noise measurements (hot sun - cold sky method). This tripod can be extended to a top height of almost 3m without azimuth rotor system, 3.5m with it.
The adjustable cross-ties give better stability.
This was my first test of the pulse radar equipment in June 2014. First test was made without azimuth rotor. 20µsec pulsed transmitter, receiver and software are working but the system needs some improvements for best performance.
DC motor with encoder for azimuth (0 to 360° degrees)
12V worm gear motor with additional worm gear to get only 2 rpm.
Torque is approx. 100 Nm.
HH-12 Encoder has a resolution of 12 bit and an accuracy of approx. 0.2°.
DC linear motor for elevation (0 to 90° degrees)
Not yet installed but I use the inclinometer sensor SCA61T .
5. Computer software
"dBZcalc" is a VB.net software running under Windows. It is connected to the µ-controller via USB cable (virtual COM port, RS232) and transfers echo data from the RSSI analog-to-digital 10bit converter to the computer. These data are buffered in a file and the second VB.net software called "MobileRadar" reads this file. "MobileRadar" also converts echo data into dBZ values representing rain rate [mm/h] and makes a rain rate overlay.
dBZcalc with some test data coming from AR8600 receiver, RSSI and µ-controller
First real test outside. First peak is Tx pulse (around
20µsec pulse width) coming back to receiver and the next peak comes
from mountain echos (distance around 22km).
vertical -> signal strength of echo
horizontal -> time/distance
Screenshot of MobileRadar coming soon
6. Place of test/operation
The red area shows the free sight of the rain radar from
"Haunsberg" (730m), 47.936N, 13.014E
Red point shows the location of the station and the red circle shows the range of 100 km (up to Linz and Munich).
© Panorama http://www.udeuschle.de