|Back to INAG Homepage||Back to UAG-104 Contents page|
Computer--controlled operation of the IPS-42 ionosonde.
Physics Department, University of Auckland, New Zealand.
Sick of buying film for your ionosonde, processing it, scaling it -- and then finding (next month) that the exposure was wrong...? So was I. If you have the money, buying one of the new IPS-71 digital ionosondes from KEL Aerospace is the answer. This appears to be a very good and very flexible instrument. If you are poor, or just prefer doing things yourself, read on.
Our IPS-42 has proved very reliable, with camera problems being the main source of lost data. It would be nice to bypass the film, and have all information stored directly in a computer. With the rapidly decreasing cost of microcomputers, computer control has become quite inexpensive and for most experimental work we now use a digital system built round a cheap IBM--clone. We therefore began work (in 1987) on the development of a system to control the IPS-42, to collect the ionograms in digital form, and to store, display and scale these as required. This project was initially quite ambitious with facilities for choosing our own sounding frequencies, and for adjusting the number of soundings made on each frequency to get a reliable echo. After much interesting experimentation the equipment has now been reduced to a simpler form which fulfils all our current needs. This Version 3 of the Auckland "DIGION" system uses an optimised mix of hardware and software, as described below.
The IPS-42 is already, in essence, a digital instrument. Sounding occurs at a fixed sequence of 576 frequencies, logarithmically spaced from 1.0 to 22.6 MHz and set by a digitally-controlled synthesiser. Originally we read (and controlled) the lines driving the synthesiser. For routine operation, however, it has proved fully satisfactory to just store data as it comes, from each of the normal sounding frequencies.
The ionosonde pulses 3 times on each frequency. Echoes are detected using a comparator which gives a logical 1 when the amplitude exceeds the mean noise level by a fixed amount, and 0 otherwise. This binary signal is clocked into a shift register at a rate which stores data from the first 800 km of effective height. The data is re-circulated during the 2nd and 3rd soundings, on the same frequency, to zero any bits which do not correspond to a consistent return. The result is then passed out to the video display during a fourth (quiet) interval. For computer acquisition we intercept this final data stream, and the associated clock signal.
Figure 1 shows a block diagram of the overall "Digion" system. A small printed-circuit board is added at the rear of the IPS-42 ionosonde, drawing its power (1 Watt) from the ionosonde. 5 signal leads from this board are clipped on to edge connectors in the rear of the ionosonde, to extract the Data, Clock, Scan, Xmr (transmitter on) and Yb (display blanking) signals. As shown in Figure. 1, the Scan, Xmr and Yb lines are combined to give a single control line which contains all the necessary timing information. A "minute" signal is also extracted, so that the computer can synchronise itself to the ionosonde clock. One further lead is soldered to the "Monitor Sweep" push button inside the ionosonde; this is used by the computer to initiate a sweep when required.
Standard RS232 drivers are used to feed the ionosonde signals to a cable, which may be tens of metres long, connected to the computer. In the computer is an added printed-circuit board containing, in essence, the circuitry shown at the right of Figure. 1. The critical Y-control line is fed to the computer in both true and complementary form, to reduce noise. These two lines are processed by a circuit which records a transition (in the single Y output) only if both lines switch simultaneously over at least 75% of their full range. This completely avoids common mode interference from the transmitter pulses. A monostable is also used to eliminate rapid transitions. As a final safeguard, the computer keeps track of when the transmitter is on and disconnects the Y signal at these times (using the `inhibit' line in Figure. 1). These precautions make the system fully immune to interference of any sort.
Data and Clock lines from the ionosonde enter the computer through normal RS232 receivers (with added hysteresis), and the Clock is cleaned up to remove any noise pulses. The Data line is then clocked into a serial to parallel converter, and a "Data-Ready" signal is activated when a full byte is available (in a latched register) to be read by the computer. The decoded Y signal is used to synchronise data collection with the transmitter pulses. 64 bytes are read by the computer at each frequency. This gives data bits at each of 512 heights in the range 0 to 800 km, for a height resolution of 1.56 km (double that available on film, or using the KEL DBD-43 controller for the IPS-42).
Within the computer an assembler subroutine is used to monitor the `DataReady' signal, and collect the 64 bytes of data at each frequency. After each byte the computer ignores further signals until the next byte is almost due. After 64 successive bytes, the program again shuts down (to avoid R.F.I.) until just before the time when data should be becoming available at the next frequency. This process is repeated until height scans have been collected at each of the 576 frequencies used by the IS-42 ionosonde. The ionogram is then displayed while the program waits for the next scheduled recording time. Any errors in the time, or in the number of heights or frequencies recorded, is flagged and recorded in a log file. However, with everything now double checked, we get error messages only during power failures. These record the time at which the mains power returns, when the equipment automatically does a full reset, resynchronising the clocks, and recommences recording.
Data Collection and Storage
When the computer is switched on (or reset) it connects to the ionosonde, and waits for a one-minute time signal. This is used to readjust the computer's clock to agree with the ionosonde. Synchronisation is also carried out after the first ionogram in each hour, so that the poor long-term accuracy of the typical computer clock is replaced by the good time-keeping of the IPS-42. The main recording routine is then entered to obtain ionograms at any required intervals. Collection of digital data uses only the `monitor' display on the ionosonde, and runs in parallel to any film recording program. Thus independent data sets are obtained; eg. digital data at 5 min intervals and film ionograms (if you still want them!) at 30 min intervals. Additional ionograms can be collected at any time by pressing `Enter' on the keyboard, or by activating a logic input to the computer. Holding this input low causes ionograms to be collected at an increased rate for periods of special interest.
With a 1.56 km height resolution the digital ionograms occupy 37 kB. The binary data consists mainly of long runs of `0' (no echoes), with shorter patterns of `1' when echoes occurred. This is rapidly and
efficiently compacted using run-length encoding for the zeros (ie. a count of the number of zero bytes), and byte-long bit patterns for the echoes. This reduces the storage per ionogram to generally between 5 and 9 kB, depending on the echo density. Each ionogram is stored with header information giving the date, time, and the frequency and height ranges used. Each day's ionograms are stored in a different file with header information giving the date, time and station identification.
Storage requirements are reduced by `cleaning' the ionograms before they are compacted. This involves deleting the date-time numerals, all information below 75 km, and all of the graticules apart from 4 reference marks which serve to verify the frequency and height scales. We also normally delete data outside the range 1.3 to 16 MHz. Some isolated dots are deleted using a very conservative algorithm which retains any information which might be useful (as near a critical frequency). Hourly ionograms are left uncleaned to provide a full check on ionosonde operation, including the AGC trace. The negligible data loss is seen in Figure. 4, which gives the fully cleaned ionogram (with added scaling lines) taken 5 min after the uncleaned hourly record of Figure. 3. After this cleaning process, the modified run-length encoding reduces the size of the ionogram data by a factor of about 10--12.
For five-min ionograms, where one in 12 is not cleaned, the average size of the stored ionograms is typically about 3.5 kB. Data for one day (288 ionograms, with a 5--min recording program) then occupies 0.9 to 1.3 MB, depending on the density of spread echoes. This is stored in a separate file for each day, labelled according to the date (eg. FEB93.01, FEB93.02,..). Every five days a standard data-compression program (PKZIP) runs to reduce the files further, by a factor of about 1.8, and copy them to tape. Thus the final storage requirement for high-resolution ionograms, recorded every 5 mins, is about 15--20 MB per month or 180--240 MB per year. This fits comfortably on two 120 MB tape cartridges (costing less than $30 each) which are swapped monthly so that data can be scaled.
A separate software package is used for display and scaling of the data. This is far more convenient than handling film ionograms. Several months' data can be stored on a hard disc, with any ionogram rapidly accessible. Holding down an arrow key gives a fast forward or backward scan of successive ionograms (`Movie' mode, at rates of about 2 per second) to dramatise changes and to select ionograms for further study. PageUp and PageDown keys are programmed to show only every sixth (half-hourly) ionogram, for a quick overview. Control--Page skips in two-hour steps for rapid access. We can also jump directly to the first or last ionogram in the current day, or on the preceding or following day. Another key will display, within one second, the ionogram at the same time on the preceding or following day. This feature, impossible with filmed data, is invaluable for studying unusual effects near sunrise and sunset. The display of digital ionograms is also much more detailed and accurate than anything available using films. This is seen by comparing Figure. 3, dumped from a computer screen, with the same ionogram printed from film ( Figure. 2).
Scaling of data is carried out using a computer aided approach. Thoughts of fully automatic scaling were discarded because of the difficulty of identifying correctly the O and X traces. Also it is hazardous to use automatically scaled data without verification, so we may as well verify as it is scaled. The current display initially shows, as vertical or horizontal lines, the critical frequencies and minimum heights scaled from the previous ionogram ( Figure. 4). Horizontal lines are shown simultaneously at h, 2h and 3h so that multiple echoes can be used to increase scaling accuracy. Similarly the critical frequencies for the x component, calculated exactly from the corresponding fo, are shown as broken lines. The MUF curve is also shown. Each line appears in a different colour, and coloured letters (following `SCALING' in Figure. 4) indicate the corresponding parameter. The rectangle at the top right in Figure. 4 is coloured to show the `current' parameter. (Original colour prints for Figures. 3 to 5 are available from the author.)
Cursor keys are used to adjust the positions of the lines. Tapping `return' saves the scaled values and displays the next ionogram. Scaling is more accurate and very much faster than from film, and the results are fully checked and reliable. Simultaneous display of fo and fx for each layer, always at the correct (calculated) separation, greatly increases the ease and accuracy with which critical frequencies can be determined. The vertical fo, fx lines will be replaced later by virtual-height curves calculated for a simple (adjustable) model ionosphere. This should provide maximum accuracy for foF2, and allow reasonable estimates of both the scale height and the peak height of the layer.
The ionosonde software is written primarily in C++, with assembler routines for the actual data collection, and occupies less than 200 kB. As a result it runs happily on the slowest and smallest IBM-compatible microcomputer. 10 Mb of free disk space will allow storage of up to one month of 15--min ionograms, using only the modified run-length compaction built in to the program. Further compaction (using PKZIP) fits a month's worth of 5--min ionograms into 15--20 MB. A cartridge tape drive is convenient for long-term storage.
An obsolete XT computer with a 40 MB hard disk is ideal for running the `Digion' system; these can be obtained for less than US$250, or $500 with a new tape drive. We can supply copies of our hardware (including plug-in printed circuit boards for the computer and the ionosonde, plus the cable and all programs) for about US$1,600. The new system has operated flawlessly for over two years at Auckland, and five copies have been provided to other satisfied users who have tried (and failed) to make the system `bomb'. Data is stored in a simple but efficient format, and source code is provided for all programs so that a user can get what he wants from the data, in whatever form he wants. It is hoped that cooperation between users will lead to an increasing amount of useful, freely available software for various purposes (such as the h'(t) plots shown in Figure. 5).
The main advantages of changing an IPS-42 to computer control, with digital data collection and storage, may be summarised as:
·No Film-- No more buying and processing film. Store ionograms for a decade in the space of one book. Share copies made cheaply in your computer.
·No Hassles-- Immediate display of final stored data to see that everything is correct. You don't have to wait a month to find the exposure was wrong.
·Viewing-- Scan data rapidly on your own computer. Skip through hourly ionograms for a quick overview. At the touch of a key, make rapid comparisons with data on adjacent ionograms, or at the same time on adjacent days.
·Accuracy-- Appreciably higher detail and resolution than film ionograms, and always in sharp focus. The height and frequency of each displayed dot is known exactly by the computer (with no calibration).
·Scaling-- is faster and much more accurate, without the eyestrain. Simultaneous display of both fo and fx, always at the correct separation, gives more accurate values of fc. For height measurements, rules at h, 2h and 3h allow convenient use of multiple echoes.
·Savings-- To store 5--min ionograms costs about $50 per annum. For 2--min ionograms it will cost you another $1 per week. Why be restricted by 15--min (filmed) ionograms?
·New Uses-- Plot h'(t) at any number of frequencies for a clear display of ionospheric changes, TID's and gravity waves (as in Figure. 5). Do real-height calculations with a simultaneous display of the full ionogram, the scaled data points, the calculated profile and the corresponding virtual heights.