Amateur Pulsar Detection

on a shoestring

using the
RTL2832U DVB-T
Dongle

Peter East
www.y1pwe.co.uk/RAProgs/Pulsars.html

Introduction

• Pulsars
• Motivation
• Detection Issues and the RTL
• Some Successes
• An Affordable Pulsar Telescope
• Challenges and Improvements
• Conclusions

1
 Pulsar Basics
In 1967 Jocelyn Bell
Discovered the first pulsar
PSR J1919+21 with a
4 acre array of 2000 dipoles
tuned to 81.5MHz
Neutron stars
Fast rotation
High Magnetic field
Wide band energy bursts

The Plan - 2014

Beg real Data from a Big Antenna

Write some software, check SNR
Understand Radio Telescope Performance
Scale the System Parameters
Design Cheap Receiver and Antenna
Try it out

2
Argentine 30m Telescope

Vela Pulsar B0833-45

Data Source (30m): Guillermo Gancio

3
 Vela Video Spectrum
0.1

0.09

0.08

0.07

0.06

0.05

0.04

0.03

0.02

0.01

0
0 500 1000 1500 2000 2500 3000 3500 4000 4500

pafft2 500.bin 500.txt 2 1 8192

Data Source (30m): Guillermo Gancio

Vela Individual Pulses

Data Source (30m): Guillermo Gancio

4
 Pulsar Flux

• Pulsar power ~ 25 Jansky peak

• 1 Jansky = 10−26 watts per square metre per hertz.
• TJ = J x A/1380 °K
• 1 Jansky = 0.00072°K/m2
• 25J (both polarisations) = 0.018°K/m2

http://www.y1pwe.co.uk/RAProgs/AmateurPulsarDetectionF.doc

Digital Radiometer Equation

Tsys
ΔT = Tsys = TLNA + Tsky
Bt B = RF bandwidth
t = Integration time
Tsys N = Number of digital bins
=
Bt / N

5
 Detected Pulsar Data

Noise ~ Tsys - say 100°K and Pulsar 0.04°K

Detected Video

B = 2MHz, t = 1ms, and here, (Tsys + Tsky)/ΔT = 50

6
 Detection Process
Within the pulsar pulse:
The receiver noise = k(Tpul + Tsys)B
and outside = k(Tsys)B
Tpul and Tsys - pulsar and system noise temperatures.
k is Boltzsmann’s Constant, B the RF bandwidth
Squaring the I and Q components (square-law detection)
results in both AC (√BBv) and DC (B) components.

AC
Tpul
DC Tsys

Folding

* Pulse adds linearly

* Noise adds as square root

* SNR improves as √(No. Folds)

SNR= √ (BT/N) x Tp/Tsys

* Optimum No. bins = Period/Pwidth

* Highly tuned period filter

7
 RTL2832U USB Dongle

• The RTL2832U is a ‘high-performance’ DVB-T

(Digital Video Broadcasting - Terrestrial)
demodulator with a USB 2.0 interface.
• It outputs 8-bit I/Q-samples at bandwidths up to
2.4MHz and tunes over 25-1800MHz

RTL SDR Features

RTL SDR is cheap and cheerful

- but can be made better
Main Limitations:
• Basic crystal accuracy - few parts /million
• - replace with TCXO ~ 0.1ppm
• Temperature drift - dissipates heat in use
• - heatsinking and fans

8
 Available Software
Data: Osmocom rtl tools rtl_sdr.exe
from: sdr.osmocom.org
Folding: rapulsar2.exe
from: y1pwe.co.uk/RAProgs
Display: Excel/MathCad
Testing: SDR# + Zadig Driver
from: sdrsharp.com
Professional: Tempo, Presto, Sigproc
from: pulsarastronomy.net

SDR#

9
 Amateur
Pulsar Detection Systems

Dish Diam 30m 3m 3m

RF Bandwidth 2MHz 2MHz 6 Mhz
Pulsar: 25Jy Peak 4ºK 0.04ºK 0.04ºK

Observation Time 100sec 10800sec 7200sec

ΔT (100ºK Tsys) 0.07ºK 0.007 ºK 0.005 ºK

SNR (100bin Fold) 56 6 8

RTL Pulsar Radio Telescope

Dish RTL
RFA
USB
Antenna Filter Chain
Dongle
PC
LNA

Software Processing:
Data De-dispersion
.bin Files Demodulation
Folding Integration

10
 Hannes - Austria
1296MHz + 2MHz + 5Hrs
3m offset dish

Andrea - Italy
422MHz + 2.4MHz + 3Hrs
2m Corner reflector
Automatic Daily Observatory

11
 Steve - Australia
436MHz + 2.4MHz + 3Hrs
5.7m CP 42-element Yagi
Planning Quad Glitch Detection

A Minimal RTL SDR System

• What can be achieved with the minimum

outlay?
• Easy home construction
• Free software
• Just enough signal to detect and identify
• Can add daily results to improve SNR
• Open to all

12
 RT Band Options
• 322-329MHz
• 406-410MHz ~ 0.75 Jansky mean
• 608-614MHz
• 1400-1427MHz ~ 0.1 Jansky mean
• 1660-1668MHz

RFI

3
4000
3.077 ×10

3000

RF xx ,12000

1000

306.739 0
4000 3000 2000 1000 0 1000 2000 3000 4000
− 4096 RF xx ,0 4096

610MHz to 612MHz

13
 Antenna Choice
• Dish - Drift scan or tracking
– ~ 50-65% efficient
– Apertures >1.5m need solid base
– Can be motorised - may need to track source
– Reflector wideband

• Yagi
– ~ 90% efficiency
– Electrical aperture greater than physical area
– Cheap, light and portable - can be stacked
– Narrow band

611MHz + 6MHz (3x2.4MHz) + 2Hrs

Twin 2.5m Yagis

Minimal Affordable System

B0329+54 Pulsar period: 714.47694ms

SNR = 2.98
.70495

0.7049

kr1 k+ kr2 k ⋅1 + kr3 k0.7048

0.7047

.70465
0 20 40 60 80 100 120 140 160 180 200 220

mod ⎛⎜ k , ⎞⎟
0 K 210
⎝ W ⎠

14
 Basic Pulsar RT
Twin
17-element Yagis Simple Radio Telescope
28.8MHz

Minicircuits TCXO
ZX60- 2x Minicircuits
P33ULN+ ZX60-P162LN+ 4-way RF
LNAs splitter
LNA
4-USB
Laptop
Power
Combiner
5-element RasPi
Interdigital Switch Quad
Filter Control RTL2832

ethernet

The Control Room

15
 Antenna

http://www.yagicad.com/yagicad/YagiCAD.htm

611MHz Filter

http://www.changpuak.ch/electronics/interdigital_bandpass_filter_designer.php

16
 Quad RTL Rx
http://www.y1pwe.co.uk/RAProgs/pdf/QuadRTLReceiver.pdf

Data Processing – DOS cmd.exe

OsmoCom rtlsdr library & capture tool: ‘rtl_sdr.exe’.
The capture tool generates files containing raw IQ ADC data
from the dongle in hex form (viewing software: ‘hexdump.exe’).

rtl_sdr22r data.bin -f 611e6 -g 42 -n 18e9

rapulsar2.exe processes this to carry out folding.

• It breaks data into blocks equal in time to the pulsar period
• Sums the blocks.
• Outputs a text file that can be viewed in Excel or MathCAD.

rapulsar2 data.bin data.txt 109 714.46389

17
 RTL Data Processing Software
• rafft2.exe – RF spectrum analysis
• filetrim2.exe – Data file trimming
• cor_tim2_n – File start correlation
• pdetect2.exe – square-law video detection
• rapulsar2.exe – Pulsar period folding
• pafft2.exe – Pulsar video spectrum analysis
• RFImit.exe – RFI spectral line blanking
• pdetfilt2.exe – Video RFI spectrum blanking
• de-dispers2co.exe – Pulsar data de-dispersion
http://www.y1pwe.co.uk/RAProgs/RTLSoftwareToolsU4-6.doc

Digital De-dispersion

Inverse
FFT FFT
fL
De-dispersed
.bin .bin file
File

fH
FFT period Delays

http://www.y1pwe.co.uk/RAProgs/pdf/SoftDispers.pdf

18
 PSR B0329+54

Data Source (25m): Michiel Klaassen

Finding and Validating

• Strong signal – period search peak
• Weak Signal
– Period search peaks
– Shift data by one record - optimise and compare
– Change number of bins
– Divide record in two and correlate sections
– Cross correlate multiple bands
– Two/three-period fold
– Check improvement with within band de-dispersion
– Check degradation with negative de-dispersion
– Check pulse width

19
 Trial 26 B0329+54 Results 1
0.4552
0.455

0.45515

0.4551

kr1k+ kr2k+ kr3k

0.45505

0.455

0.4550.45495
0 20 40 60 80 100 120 140 160 180

mod⎛⎜ k ,
0 K⎞ 180
⎟
⎝ W⎠

summed detected data

4
−4
2 .10
1.2 ×10

4
1 .10
pmLk
− mnL
couLk
0
pmUk
− mnU
couUk
4
1 .10

−4 4
− 1.422 ×10 2 .10
0 20 40 60 80 100 120 140 160 180

mod ⎛⎜ k ,
0 K⎞ 180
⎟
⎝ W⎠

First/second half-file correlation

Results 2
5
−5
5 .10
4.532 ×10

kr1k− kr1m

kr2k− kr2m 0
0

kr3k− kr3m

−5
− 5 ⋅10
0 20 40 60 80 100 120 140 160 180

mod ⎛⎜ k ,
0 K⎞ 180
⎟
⎝ W⎠

3-band correlation
0.4553
0.455

0.4552

kr1k+ kr2k+ kr3k0.4551

0.455

0.4550.4549
0 20 40 60 80 100 120 140 160 180

mod⎛⎜ k ,
0 K⎞ 180
⎟
⎝ W⎠

Three-period fold and overlap

20
 Results 3
0.4552
0.455

0.45515

0.4551
kr1k+ kr2k+ kr3k
0.45505

0.455

0.4550.45495
0 20 40 60 80 100 120 140 160 180

mod⎛⎜ k ,
0 K⎞ 180
⎟
⎝ W⎠

De-dispersed response
0.45515
0.455

0.4551

kr1k+ kr2k+ kr3k0.45505

0.455

0.4550.45495
0 20 40 60 80 100 120 140 160 180

mod ⎛⎜ k ,
0 K⎞ 178
⎟
⎝ W⎠

Raw data response

Results 4

0.4553
0.455

0.4552

kr1 k+ kr2 k+ kr3 k0.4551

0.455

0.455 0.4549
0 20 40 60 80 100 120 140 160 180 200
0
mod
⎛⎜ k , K ⎞⎟ 200
⎝ W ⎠

6ms Pulse visible on 1ms scale

21
 Challenges
• Scintillation - frequency and time
• RF Interference - RF and Video
• Transient spikes
• Weak Signals - Folding process can find peaks
in noise
• Validating low SNR Detections

Improvements
• Lower Tsys - direct SNR improvement
• RTL Band flattening
• Longer Data Records
• Spectrum folding
• Rubidium/GPS locking - multiple sessions
• User-Friendly/Automatic GUI

22
 Conclusion
• Amateurs can detect strong Pulsars
• 3m Dish systems work well
• Detection with home-made Yagis is
possible but more difficult
• RTLs make for an inexpensive Receiver
• Freely Available Acquisition and
Processing Software
Or, if you have a problem,
• Find a friend with a BIGGER DISH

Pulsar Amateur Links

• Neutronstar Group
– http://neutronstar.joataman.net/index.html
• Barga Observatory
– http://iw5bhy.altervista.org/
• Y1PWE
– http://www.y1pwe.co.uk/RAProgs/Pulsars.html

23