Passive Ionospheric Sounding and Ranging
Chirp Sounding is a relatively new technique used to measure the ionosphere by bouncing signals off it, in much the same way as a RADAR system. Indeed, older Ionosondes (equipment for sounding the ionosphere) worked in exactly that way, by sending high powered pulses and listening for the response. The modern Chirp Sounder uses lower power and a continuous signal which changes frequency at a steady rate.
Peter Martinez G3PLX and others have used Digital Signal Processing (DSP) and Doppler techniques to measure small differences in carrier frequency that result from movements in the radio propagation path (see the Precision Carrier Analysis page). While interesting for meteor scatter, aircraft and satellite reflections, and more gross or localized ionospheric effects, this technique gives no direct information about reflection layer height, and it becomes difficult to infer information about the propagation medium over more complex paths. What was required was a time domain - rather than frequency domain - technique, for example measuring the propagation time of pulses. It was soon established that a wideband technique, rather than a carrier based technique would be necessary, in order to achieve sufficient time resolution.
In searching for suitable pulse transmissions to use, preferably transmissions available from all over the world on a 24 hour basis, Peter stumbled across a family of transmitters that are used as swept frequency ionospheric sounders. In their normal application, research, professional and military groups use these low power devices to probe the ionosphere to measure propagation. The signal consists of a single long 'chirp', sweeping up in frequency at a constant rate, and repeated periodically. These transmissions are tracked by a companion receiver which is zero beat with the transmitter, and so ionospheric reflections that are returned with short delays are heard as lower sideband audio beats of a few hundred Hz. The equipment then builds an 'ionogram' or two dimensional graphical representation of the ionosphere's reflection height or delay against frequency. The adjacent picture illustrates a typical commercial 50W FM/CW (chirped) ionospheric sounding transmitter.
The first step was to discover how these chirped signals could be used in a passive manner, i.e. without reference to the transmitter oscillators or timing reference. To do this, Peter developed a very clever chirped filter, which not only sweeps in frequency at 100 kHz/second, but has properties not possible in a conventional filter - a bandwidth of only 66 Hz, but with a pulse resolution of 0.66ms. This filter and matching detection software formed the basis of the adventure to follow, and allowed these chirped sounders to be analysed on a single frequency using a fixed receiver. The remaining problem was to determine when the chirp transmissions started, in order to know when it should sweep past the receiver, in order to calibrate the fixed frequency receiver for range.
The purpose of this project has been to explore the use of publicly available chirped ionospheric sounding transmissions to study the ionosphere. We now know that these transmissions are made regularly from most parts of the world, and cover much of the HF spectrum from about 3 to 30 MHz on a 24 hour basis. The unique aspect of this project is that it is wholly passive - it makes use of the sounding transmissions made by other agencies. This means that anyone can potentially receive and interpret the transmissions.
The transmissions used by this project are chirped sounders which transmit at a constant rate of 100 kHz/second, and have reliable chirp start times. Most of these sounders are commercial and research transmissions for vertical or oblique ionospheric sounding purposes. Some are probably military sounders with a similar purpose.
Once the capabilities of this passive sounding technique were understood, the next step was to develop a solution which would enable anyone with an interest in HF propagation to study it in real time with a minimum of equipment and expense - for example using nothing more than a PC with sound card. In addition, one reference sounder might be used to calibrate reception of others, avoiding the need for an expensive high precision time reference. It was also discovered that many of the sounders, especially those used for oblique sounding (transmitter and receiver at widely separated sites) used GPS timing references in order to maintain calibration at multiple sites, and these have been of the greatest interest.
The first phase of this project took place during the latter half of 1999, and proved that it is possible to receive and accurately measure these sounding transmissions in a passive manner. Assessment of the results identified a number of areas of worthwhile improvement. It was possible during this phase to set and synchronise clocks on opposite sides of the earth, and to measure arrival times of signals to ±1 ms. The system concentrated on sounders with periods of 5 and 15 minutes.
Several sounder transmissions were identified and their locations discovered by hyperbolic triangulation (measurement of arrival times at different locations and plotting lines of equal delay). Improved tracking of sounders with different chirp periods, improved time resolution and simpler setup, clock synchonisation and system calibration were perceived as the main areas for improvement. Stations were in many cases able to receive the same signals, thus making distance measurements possible. Some stations had high precision references, making single-station distance measurements possible on some known sounders. Both long and short path transmissions could be identified by timing, and on occasions it was possible to resolve long path and short path signals simultaneously. On a few occasions round-the-world delays were detected.
Transmission from Cyprus received in New Zealand,
showing long path (left) and short path (right).
Vertical scale is milliseconds.
In a later development, new software and tighter hardware requirements allowed measurements were made to ±125us resolution, using sounders with a wide range of periods from 5 to 30 minutes. High precision GPS time references were used for the first time, providing ±1us clock accuracy and similar precision of synchronism between sites.
Better data analysis allowed more accurate delay measurements to be made, making possible the identification of individual propagation paths. At this point 10 or more stations were equipped for passive sounding, but the number was limited by the non-availability of suitable DSP hardware. More observers were needed, with a better geographic spread so that more accurate triangulation of the sounder sites would be possible, but that meant finding a hardware independent receiver solution possible.
Some sounders were found to drift in time, jump in reference time, or could only be heard in some locations, or were not available continuously. An army of 'chirp spotters' would be required to help solve these problems, so the need for a PC sound card solution became very apparent.
Thanks to Andy G0TJZ, we now have an excellent platform independent solution - the PC sound card Chirpview software. As a result more stations have been able to explore these fascinating sounders. Some really good information is coming to light on propagation over paths that have been previously difficult to study. An email group and a superb Chirp-Sounder's Web Site has been put in place to provide news, up-to-date statistics, software, and a database of known sounders.
The PC software requires only a modern Pentium™ class computer with a Sound Blaster™ compatible sound card, a stable and accurate HF SSB receiver, and a GPS system with precise 1PPS pulse output.
Given sufficiently accurate clocks, or the same clock reference used at the transmitter and at the receiver, it is possible to measure the time it takes a radio signal to travel from one place to another. At a speed of about 300,000 km/sec (3 x 108m.s-1), a radio signal can travel right around the world in about 138 ms.
If the transmitter and receiver are in fixed locations, you would expect this delay to be constant. However, it is not, and this variation is the principle on which this project is based. The arrival time of the signal will depend on which way around the world it went, how many times it bounced off the ionosphere and the earth, and which ionospheric layers were involved.
Ionogram of a UK sounder showing groundwave
(straight line) and skywave signals, range 50 km.
Most applications of ionospheric sounding are either vertical or oblique, i.e. with the transmitter and receiver either co-located or separated by up to a few thousand km. This project takes oblique sounding to the extreme - the receiver can be anywhere on the globe. This places the highest demands on stability of the receiver and especially on the time references used. As the distance between transmitter and receiver increases, of course the number of possible paths increases and complexity of the returned signal is therefore increased.
An explanation of the above image is in order. This graph is called a 'waterfall', a type of ionogram where the image axes are both time - horizontally in UTC hours (time of day), and vertically in milliseconds, the delay time from some fixed reference point. Since it is not practical to display more than a short period of time vertically with high resolution, the vertical size is limited to ±40 ms (in this example) or up to +150ms (in other examples). The image displays the strength of the signal during the receiving "window", using white for no signal, and black for very strong signals. 256 grey levels are displayed, 0.25dB per step, over a range of 64 dB. Imagine that a waterfall is set to a time of 2.5 seconds, with a period of 300 seconds (five minutes). Any signal that appears within 40ms of the UTC five minute points plus 2.5 seconds (00:02.5, 05:02.5 minutes:seconds etc) will be displayed in the waterfall window. This technique is extremely sensitive, as no digital detection process is involved - interpretation is left to the eye.
Unlike most examples shown, the image immediately above shows a constant horizontal line - this is because the transmitter was within groundwave range of the receiver. Much of the day, this is the only signal received; however, between 0600 and 2100 UTC, faint lines first with decreasing and later increasing delay appear. These are caused by scatter to the receiving site from an ionospheric skip occuring to some other part of the world. As the skip zone moves closer, the delay is reduced. The signal suddenly becomes very strong and with a stable short delay between 1200 and 1900 UTC. This is the F layer reflection which occurs during daytime, where the transmitted signal is reflected from the ionosphere and directly received at the observing receiver. The additional delay (i.e. the time later than the ground wave arrival time) is an indirect measure of the height of the reflective layer. If you look closely, you can see that there are actually two separate lines between 1200 and 1900 UTC, the ground wave signal and the F-layer skip signal. The fuzzy stuff with longer delays above these strong lines come about because the reflective layer is diffuse, causing some diffraction (scatter), which occurs through a mechanism not unlike the scattering of sunlight reflections from ripples on a pond.
To measure differences between short and long paths around the world requires the ability to measure delay times of about 1 to 140 ms with a resolution of about ±1 ms. To measure with sufficient resolution to resolve individual reflection paths requires rather higher resolution.
However, resolution is only part of the story. The project involves measuring the delay times of signals for hours on end, and if the clock was to wander off in that time, the accuracy of the result would be lost. For the delay to be measured with an accuracy to match the resolution would require an accumulated clock error of less than 125 ±us per day, or nearly one part in 10 7. Only the most expensive rubidium or caesium standards can achieve this low order of drift over long periods, so the decision was made to utilise the timing provided by the GPS (Global Positioning Satellite) system. We have discovered that many of the sounder transmissions are also controlled in this way. The one second references pulses generated by a good GPS receiver are accurate to easily ±1 us on a continuous basis, providing a high quality time reference anywhere in the world (well, except near the poles).
In addition to the precision second pulse, the system makes use of the GPS NMEA messages, which allow the equipment to recognise which UTC second each pulse refers to. The NMEA information on its own is not sufficiently accurate, since it suffers unreliable delays in serial transmission and reception. The timing can also be affected by the actual data transmitted.
Up to this point, the ionosonde signal has been described as though it was a simple pulse. Thinking of it in this way makes understanding the process easier. However, it is difficult to transmit a narrow enough pulse to provide good time resolution, and at the same time provide sufficient energy in the pulse for good sensitivity. This problem is shared with radar systems, and the solutions are similar. In addition, the sounders require to measure the ionosphere throughout the HF spectrum, which is again not so easy to achieve with a pulse.
The ionosonde transmitter in fact sends a continuous carrier, but with smoothly changing frequency, at a fixed but accurate rate (in the case of most chirp sounders we use, with increasing frequency at 100 kHz/second). Peter's design uses the special chirped filter previously described, with properties not attainable with a conventional filter, and so is able to detect the transmissions with 0.66 ms time resolution, and with very narrow bandwidth that provides high sensitivity.
Knowing the chirp rate of the transmission, and what frequency the chirp is being received on, one can work out the nominal chirp time at which the received signal apparently started at zero frequency - by simply counting back at 100 kHz/second.
There are two useful advantages of this chirped filter technique:
The disadvantage is that you can only look at the ionosphere on one frequency at a time, unless you have multiple receivers, or a frequency agile receiver and appropriate control software.
- The receiver will have narrow bandwidth, so will work with low power sounder transmissions.
- The pulse response of the filter provides high time resolution.
In the case of conventional chirped ionosondes, the receiver is more conventional, but follows (tracks) its matching transmitter throughout the HF spectrum. In this passive sounding project, the receiver tracks many different transmitters using the chirped filter, but only over the width of an SSB receiver bandpass - about 2.4 kHz - since the receiver frequency is fixed. This approach is more than sufficient for sensitive single frequency measurements. You simply set the receiver frequency to suit the band you wish to know about.
Listen to a typical chirp received in a 2.4 kHz bandwidth (44kB)
There are two possible equipment solutions. The first system developed was the G3PLX system, which uses the Motorola DSP56002 EVM development kit. The new PC software system by Andrew Taylor G0TJZ requires only a fast PC with sound card. This is good news for those not able to find the (now discontinued) Motorola unit.
To take part in chirp tracking, you will need the following components:The DSP software (in the DSP unit or the computer) is at the heart of the system - it receives the audio from the receiver, filters out the chirps, and measures their amplitude and time. It also receives time information, in the form of seconds pulses and NMEA format serial coded time data, which are derived from the GPS unit. The NMEA (serial data) message allows the processor to decide which seconds pulse is which. The time and amplitude data for each chirp is then analysed on the PC and displayed. Since the data is sampled at 8 kHz, the system resolution is 125 us.
- Stable and accurate HF receiver, in USB mode
- 5m vertical antenna in a noise-free environment
- DSP unit, the Motorola DSP56002 EVM development kit (for G3PLX software)
- PC, 486 or better with Win3.1/95/98 and EVMCHIRP software by Peter Martinez G3PLX
- PC, Pentium™ or better with Win98 or later and sound card and Chirpview software by Andrew Taylor G0TJZ
- GPS receiver with seconds pulse and NMEA data
Calibration and setup tools allow for choice of serial port, setting a calibration delay to compensate for the delay in the receiver, and also to compensate for when the NMEA message arrives relative to the seconds pulse. The receiving frequency is entered (so the software can extrapolate the chirptime), and the signal level and UTC time (from the GPS) are displayed. Three main products are provided by the PC:
- The Chirp Log
- A log of detected chirps, which can be saved to file. This contains UTC time, signal strength, period and measured delay (chirptime), for every detected pulse. It does not contain entries for signals too weak or distorted to be detected as a chirp, but will contain occasional "hits" caused by strong noise pulses and other interference. Pulses as close as 1ms are logged independently, so provided the signal is strong enough and not too distorted, scatter and long path hits will also record.
- The Chirp Statistics
- A log of chirp statistics, which gives a summary of each known chirptime, with period, number of detected pulses, and the time of the first and last detected pulses. The chirptime is averaged over the pulses for the previous two hours, which enhances the precision. Random noise hits are eliminated.
- The Waterfall Displays
- Multiple waterfalls can be set up, in order to monitor any suitable period and chirptime. The waterfall gives a graphical display of the signal, and is much more sensitive than the logs. The waterfall relies on visual interpretation, not software detection of the chirp properties. The scale of the waterfall is 1 ms/pixel vertically, and 5 minutes/pixel horizontally. Very complex reception conditions can be displayed.
The Waterfall display (G3PLX software)
All the software (both versions) are available from the Chirp-Sounder's Web Site.
There is also a range plotting utility, which allows you to enter positions of receiving sites with their delay times, and will plot lines of equal delay on which the transmitter must lie. It will also plot antipodal circles from long path - short path differential delays. This software is ideal for locating unknown transmitters.
It is not practical to distribute a definitive list of all known chirp sounders on a web site, since many of them change from day to day, or week to week. Current observations are generally distributed by email, by posting logs to the chirps mailing list. These logs are helpful in identifying or locating new sounders.
In the log example below, the times quoted here are the source chirp time, i.e. the time that would be measured by a receiver on the transmitter site. You can estimate your chirptime by adding 1 ms for every 300 km of range from the source. For example, for a chirp time of 300:245.000 and a range of 10,000 km, expect the signal to arrive at about 300:245.030.
Chirp times are typically quoted in the form period:chirp time, where the period is in seconds, and the chirp time is the zero frequency extrapolated time of the first chirp transmitted each hour.Period:Chirp Lat Long Approx location -------------+--------+-------------------- 900:178.5496 33S 149E Canberra, Australia 300:250.0000 35N 34E Cyprus 300:224.14595 52S 59W Stanley, Falkland Is 300:77.94247 54N 03W Inskip UK