The Woomera AN-FPS-16 Radars
by Ken Anderson
In the early 1960s two AN-FPS-16 Instrumentation radar sets were installed on the Weapons Research Establishment [WRE] Range, the longest over land range in the free world, some 2400 km long.
One of these radars was located at the Red Lake site, call sign R38, approximately 40 kms north east of the range head, and the other at Mirikata, R39, a down range Instrumentation site, approximately 115 kms south of Coober Pedy.
The AN-FPS-16, an amplitude comparison monopulse device, working at what was then called C-band, with a fixed frequency , nominally 5.5 GHz, 1 MW transmitter, and a tunable 250 KW transmitter which covered 5.48GHz to 5.825 GHZ – one or other transmitter being used depending on the particular mission. The receiver system used balanced mixers and had a NF of 10 dB. The maximum unambiguous range of the radar was 1,048,576 yd, that is, 524.3 nautical miles, or 971 km.
These radars had been developed from the original monopulse radar set which the US Naval Research laboratories had built in 1943. RCA and the NRL built two prototypes of the set, the first is believed to have had a lens type antenna, whilst the second, had a more familiar dish type antenna. The latter proved to be the more suitable and the production radars were equipped with such an antenna. The AN-FPS 16 was first built in 1958. [there was also a XN3 version-with a 3 MW transmitter, which may well have been subsumed into the later AN-FPQ 6 radar]
The radars were installed on the WRE range, in approximately 1960, to provide precision tracking of the various high speed, long-range missiles being tested. They were very high precision being capable of tracking with angle errors of 0.1 milliradian (approximately 0.006 degree) and range errors of less than 5 yards (5 m) with a signal-to-noise ratio of 20 decibels or greater. [The angle readout of the AN-FPS 15 was in mils, there being 6400 mils in a circle. For small values the error incurred in treating a mil as a milliradian is very small, as 10 mils are equal to 9.8 milliradians.] They had a maximum un-ambiguous range of 500 nautical mile. [A nautical mile is equal, by definition, to 1852 metres.] The improvement in performance of the AN-FPS 16 radar over the VERLORT [VERy LOng Range Tracking] S-band conical scanning radar also used for the early Mercury space missions is shown by typical errors in the VERLORT data of 30 yds in range and 1.7 milliradians in angles.
We can think of the target position as reported by a radar system as being an uncertainty volume, in the case of the radar under discussion, cylindrical in shape, which is twice the range error in length and the angle error in diameter. So, at a satellite range of 700 nautical miles, the AN-FPS 16 would have a cell 10 metres long and 70 metres in diameter, whilst the equivalent cell for the VERLORT radar would be 600 metres long and 1190 metres in diameter, a 1000 fold increase in the uncertainty volume. The much better performance of the AN-FPS 16 was due to the monopulse tracking technique and to the very precise ranging system of the more modern radar.
Of course, at the shorter ranges typical of missiles on the WRE range the position uncertainty was much smaller. If we take a range of 50 nautical miles then the cell size is still 10 metres long but is only 5 metres in diameter. The target would be assumed to be at the centre of the cell. [Strictly, the uncertainty volume would be related to Standard Deviation of the data stream, and is best visualized as a cloud with its relatively dense centre fading to a clear area at 3 SD.]
The AN-FPS 16 radars employed a 3.9 metre [12 ft] diameter Az-El antenna which resulted in a beam width of 21 milliradians [1.2 Degs], with a precision pedestal capable of a 6400 mil rotation in just under 9 seconds. The pedestal was mounted on 4 metre square concrete pillar which was inside the building, but isolated completely from the building, penetrating down to bedrock.
The electronic equipment for the radar was installed in standard racks on the top floor of a two storey brick building. The ground floor of the building contained the air conditioning system, the primary power switchboards, which included motor driven variable transformers allowing adjustment of the supply voltage, the waveguide pressurization sub-system and the antenna drive amplidyne machines. This floor also contained the WRE data transmission and receiving system.
A cabinet on the rotating antenna pedestal contained the receiver mixers, local oscillators and the intermediate frequency pre-amplifiers, amplifiers and the Tranmsit/Receive [TR] switch, which automatically disconnects the receiver from, and connects the transmitter to, the antenna during the transmitter pulse.
The monopulse system employed three receiver trains-the SUM or reference channel, the AZ channel and the EL channel. Because the radar could be used in either skin or beacon mode there were also beacon IF amplifiers. In Skin mode the radar tracked the direct reflection from the target of its transmitter signal which is on the same frequency as the transmitter. In the beacon mode the coded transmitted signal from the radar triggered an electronic transponder in the target. This device then transmitted a signal, on a frequency different to the radar signal, back to the radar. The transponder output signal was typically 400W, a much higher amplitude signal than the reflection from the target.
By using this technique, with the target co-operating in the track, much greater ranges could be achieved. In fact without it, it would have been impracticable for the AN-FPS 16 radar to track the early manned space Mercury capsules, the reflected signal would have been well down into the noise, even at the point of closest approach of the target, typically 200 nautical miles.
At the Red Lake site NASA established a Telemetry and Acquisition Aid system, in a separate building from the radar, as the Red Lake site formed part of the NASA Manned Space Flight Network. The buildings were linked by a synchro network so that the radar antenna could be slaved to the ACQ aid antenna. Because the VHF signal from the spacecraft was detected before the spacecraft had penetrated the radar horizon, the VHF signal following the curvature of the earth slightly, the radar antenna was thereby pointed towards the actual position of the spacecraft. The radar operator could switch the radar scan generator which caused the radar antenna to sweep over a search pattern around the position designated to it, thus allowing for slight errors in the data.
The Red Lake radar was upgraded by the addition of the IRACQ [Instrumentation Radar ACQuisition] modification and of a data transmission sub-system which transmitted,as a teletype signal, the radar range, azimuth, elevation, and time once every 6 seconds when the radar was tracking a NASA spacecraft. This real time data, and, in this broadband day and age we would hardly regard it as real time data, it was very cutting edge at the time, went to the NASA computer system at Goddard Space Flight Centre. [GSFC]
The IRACQ modification consisted of an all electronic range tracking system, an auxiliary to the precision mechanical range gear of the basic radar, and, therefore referred to as the AUXTRACK, and an additional operator console. This was an all electronic system, which used a Voltage controlled Crystal Oscillator [VCXO] as the clock generator. The range tracking gate was actually comprised of an early and a late gate, the early gate starting with the leading edge of the tracking gate and finishing half way through that gate, at which point the late gate began and ran to the end of the range gate. A comparator circuit evaluated the signal in each of the early and late gates and developed an error voltage which, in turn, varied the frequency of the VCXO. This varied the counter count time and thus, the time at which the tracking gate occurred, in the correct sense to track the signal. This modification was provided to overcome the limitations of the main range system, which resulted from the fact that the Mercury and later capsules were typically, being equipped with transponders, acquired by the radar at a range of 700 nautical miles.
Because the radar had a maximum un-ambiguous range of , for the purposes of this discussion, 500 nautical miles [very nearl1070 kms], set by the time between two successive transmitter pulses, with a Pulse Repetition Frequency [PRF] of 142 pulses per second being employed. The transponder signal being tracked was due to the transmitter pulse for the period before the present one, so, a signal from an object 700 nautical miles away appeared to be at a range of 200 nautical miles, the maximum un-ambiguous range of radar, plus the additional 200 nautical miles representing the travel time of the radar signal as it went out to the target and returned to the radar.
As the target closes in range it rapidly apparently approaches zero indicated range, an actual range of 500 nautical miles. If allowed to continue the signal will be received at the instant the next transmitter pulse occurs. At the instant the transmitter emits the next pulse, the Transmit/Receive switch [T/R Switch] disconnects the receiver from the antenna and connects the transmitter to the antenna, when that happened track would be lost. The IRACQ sub-system used the added electronic range system to track the target, and so provide the necessary gating pulses to the AZ and El receive channels, which allowed the radar to track, in angles, whilst the operator slewed the range gear box from zero to maximum un-ambiguous range and then re-acquire range track at the [real] 499 nautical mile range.
To allow the signal resulting from the n-1 pulse to be tracked at the time the nth pulse was being sent. [doing so being called ‘through the Big Bang] as the closing target reached an indicated range of less than 8 nautical miles the IRACQ sub-system took over the generation of transmitter trigger pulses and produced these at a time equivalent to 8 nautical miles later than they normal trigger train. At the same time the auxiliary range track system delayed the range gate so that the gate and the desired echo coincided in time. After the closing target reached an indicated 499 nautical mile range, and the operator had matched the radar tracking gate to the target position, the normal transmitter pulse train once again became the active source. In the data reduction process at the Cape a corrected range, corrected for the fact that the radar data was due to what is called nth time around tracking was arrived at. In Nth time around tracking the radar received signal results from the transmitter pulse N PRF periods ago.
The IRACQ Console had, on the left, two A-scope range displays, one 178 mm [7 inch] diameter tube, which displayed the entire un-ambiguous range of the radar on two traces. Each showing half the total un-ambiguous range, and a 125 mm [5 inch] diameter tube with two traces. The upper of these showed a 50 kyd [45 km]segment of the total un-ambiguous range centered on the radar range tracking gate, and the lower trace a 6 kyd [5.5 km] segment of the total range centered on the radar range tracking gate.
In its centre the IRACQ console had a wave meter which was designed to allow measurement of the receiver local oscillator. Then on the right the console included a B/E scope. This had two traces at right angles to each other. The vertical trace began at the bottom of the scope at 0 range and with increasing range towards the top of the scope and moved left or right as the radar antenna azimuth was moved away from the designated acquisition point of the space craft. The horizontal trace began at the left hand side of the scope, representing 0 range and with increasing range to the right. It moved vertically as the antenna elevation moved from the designated position. If we take the case of the spacecraft traveling towards the expected point AOS point, then radar operator would have the antenna slaved to, as mentioned above, the the Acq Aid antenna. The B/E scope operator could switch on the IRACQ scan generator [note that this is different to the radar scan generator] which would cause the radar antenna to search through the selected pattern, there being four of these for use under different circumstances. When a signal was detected bright pips would show on each trace to indicate where the spacecraft was with respect to the designated position. The B/E scope operator could, then switch off the scan and use his joystick to bring the antenna onto the spacecraft.
Parts for the IRACQ mod must have cost an absolute fortune. Because they had been conceived of, designed, constructed, and installed in 13 months, they were essentially prototypes. Spares came wrapped in very sturdy boxes with an RCA part number, and so on, and with values like 346.723 Ohms. It was reasonable to assume that the nearest preferred value resistor would have probably done the job, but, somewhere in RCA someone was employed to make these odd value resistors to meet the NASA specification.
The radar range tracking system employed an early/late gate tracking technique. The range gate was the equal in length to the transmitter pulse width. In the range sub-system an early gate, half that length started at the beginning of the range gate, and a late gate started at the half way point of the range gate. The range sub-system employed a comparator circuit which measured the relative signal strength in each of these two gates and derived an error voltage which drove the range servo system to reduce any inequality.
The Mirikata radar was a not employed in NASA tracking and was thus, a stock standard AN-FPS 16 device. Its crew were flown to the site, usually in an RAAF De Havilland Otter aircraft, on Monday of each week, accommodated in the Mirikata camp, which could house approximately 50 people, and flown back to Woomera on Friday afternoon, when a DC3 or Dakota, and occasionally, a Bristol Freighter, known to one and all, because of the way its clamshell nose door flexed in flight as the “Frighter” was employed.
The crews at each site operated and maintained the radars in order to support the WRE trials or NASA mission as required. The radar performance was checked at least weekly by the tracking of a 150 mm diameter metal sphere suspended from hydrogen filled weather balloon. [Filling of which was a very hazardous procedure in the very low humidity conditions. The danger was that electrostatic build up on the filled balloon would lead to a spark which would explosively ignite the hydrogen. The balloon filling hut was remote from other structures and was fitted with a water spray system. The drill was that you dipped the empty balloon in water, attached to the filling nozzle, retired to outside the shed, turned on the water spray, then filled the balloon, turned off the sprays, into the shed, tie off the balloon with the metal sphere pre-attached to the twine, outside and let it go. No one ever got hurt in the process.
One of these spheres caused somewhat of a stir when recovered from Lake Eyre by the local station owner. This occurred just after Donald Campbell had set out to break the land speed record on Lake Eyre. Because of the interest in this attempt the station owner had many contacts in the media.
Also, just before this event the so-called “Boulia Ball” had been found and generated great media interest. It was determined to be a spherical pressure vessel which had survived the re-entry of a rocket launched by the USSR. So, one of the radar crew had taken it upon himself to write on a calibration sphere the letters CCCP, and “If found return to Joe, The Kremlin.” in Red Texta. When he found this the station owner immediately called his media contacts in Adelaide. A great fuss followed, and, in no time flat the radar crew found themselves being very sternly admonished by WRE Security people, threatened with loss of their security clearances etc etc. No one else ever wrote anything on a sphere.
In preparation for a mission the radar had to be calibrated, a process which took about two hours. It involved checking that the radar pointed in the right spot, a process which involved y tracking a test signal from the Bore site Tower [BST] whilst one of the crew rode the antenna, [an exciting thing to do, slewing at 44 Degs per second through 90 degrees requires you to brace yourself and hang on tightly.] and checked its alignment through a precision telescopic sight.
The receiver system was then calibrated by reducing the BST signal in small steps and recording the resultant AGC voltage onto one channel of the Sanborn 8 channel pen recorder. The reduction continued until the receiver noise level was reached, at approximately – 110 dBm. The angle error and range voltages were also recorded onto the strip chart recorder, the angle errors being calibrated by offsetting the radar 1 mil from the BST locked on position. The antenna was also checked for level by having the antenna rider talked the operator onto each of two accurately surveyed zero elevation markers, one on the BST and the other at 90 Degs in azimuth away from that tower.
Then the range system was calibrated by transmitting to the Range marker, a corner reflector mounted on a 10 metre high tower. The range to the marker was known because of very accurate survey. The actual calibrations required and the various adjustments of the several radar parameters were detailed in,either, the WRE Trials Instruction or the NASA Mission Plan.
Having calibrated the system the crew then prepared for the actual mission. In the case of the NASA missions a teletype message called an Inter-Range Vector [IRV] provided the best data available as to the expected position of the spacecraft at Acquisition of Signal [AOS]. So, as the predicted AOS time approached the radar would be manually positioned to the IRV designated point, and the operator would wait for the Acq Aid crew to report that the Acq Aid had AOS,
The radar operator would then slave the radar antenna to the Acq Aid antenna and start the range system searching around the IRV designated range. Usually, the spacecraft transponder signal bobbed up in the range gate, the radar angle and range servo locked on, and the data started flowing to the GSFC, at a line of teletype giving time, range, azimuth and elevation, every 6 seconds.
In the case of the WRE trials the WRE pointing system provided pointing data derived from the kine-theodolite cameras and other devices. In this case the operator was able to slave the radar antenna to the WRE data, so that the antenna was then pointing at the target position. Hand wheels on the radar console allowed him, and it was, in those days always him, to adjust antenna position and range gate settings to correct for any discrepancy. The range system could be set to sweep around the designated point and to automatically lock onto the target.
Data from the radar went to the Range Head computer via the WRE data system. The operator’s skill was tested when he had to switch the radar from one target to another nearby one. This happened in trials employing the Skylark, a rocket 7.6 metres(25ft) long, 0.44 metres (17 in) in diameter and with a fin span of 0.96 metres (38 in). It was used for many trials at Woomera. In some of these the Skylark was fired after local dusk to such a height that it emerged into the sunlight, at which point it ejected a falling body, accompanied by a puff of talcum powder. This produced a cloud visible form the ground. The radar operator was able to see the echo of both the Skylark and the falling body, and had, by vigorously winding the range hand wheel, transfer the range tracking gate from the Skylark to the falling body – the object of these trials being to study wind profiles as detected by the horizontal movement of the falling body.
Each radar had what was called an optical tracker, a pair of WWII German high power binoculars mounted onto a pedestal such that the user»s movements were repeated into the radar building, with the operator having the ability to slave the antenna to the binoculars. Should the signal transfer fail to take place at ejection, a rare event, then the optical tracker provided a second chance to find the falling body before it disappeared into the earth’s shadow.
The WRE data system had two transmission channels, usually one carried target data and the other missile data. At the Red Lake site the radar operator was able to select either channel on a switch box at his left hand. But, for some reason or another, at Mirikata, the channel selection was controlled in the Instrumentation Building, some 1 km away from the radar.
Usually, the Instrumentation Building control room was manned, but, one night when an important re-entry body trial, which required colour photographs of the re-entry of an accelerated body, the Control room was not manned, the operator being prostrate and non compos mentis, due, as it turned out, to excessive consumption of alcohol that afternoon.
The Trials Instruction required the radar to track the missile during ascent and then transfer to the re-entry body after it separated from the missile and was accelerated back into the atmosphere by the third stage rocket. The pointing data for the missile was transmitted on one channel and that for the re-entry body on the other.
So, in the absence of the Control Room operator a member of the radar crew was detailed to drive to the Control Room and operate the switch. Launch, successful track, the moment to switch – crew member rushes out of the building, leaps into vehicle, turns headlights on to high and spins around from the parking lot, departs at high speed for the Control Room. Oops!! When he spun around with lights on high they swept across the fields of view of the cameras waiting, with open shutters, for the bright green re-entry flare. Oh dear, no photos!!!!
What a fuss followed, former Control room operator, with rapidly packed goods and chattels, was escorted onto a plane and taken to Adelaide, and hurled into outer darkness. Many question asked of the radar crew, why, wherefore, how, why not????? But the switches remained in the Instrumentation Building!
Curiously, and quite by co-incidence, the Range Marker for the Red Lake radar was at a distance of 2505 yds, or in Octal 4711 yds-easily remembered because of the fact that the octal value is brand of of the well know eau-de-Cologne. Just a co-incidence.
It should be pointed out, that in order to achieve the required precision each angle co-ordinate and the range system employed two digital encoders, the higher order bits employing a mechanical encoder and the lower order an optical device. The encoders were arranged so that the Most Significant Bit [MSB] of the optical encoder and the Least Significant Bit [LSB] of the mechanical encoder had the same weight, that is, represented the same value. This was simply due to the fact that there were no single encoders available to achieve the required precision.
The encoders employed a special binary code known as Gray Code or Reflected Binary, in which only 1 bit changes for an increase or decrease of 1 unit. In normal binary code this is not the case,
Normal binary code.
0 = 000
1 = 001
2 = 010
3 = 011
4 = 100
So, the increase from 3 to 4 requires 3 bits to change. Bits 1 and 2 turning off and bit 3 turning on.
Gray or Reflected binary code.
0 = 000
1 = 001
2 = 011
3 = 010
4 = 110
Here the change from any digit to the next changes only one bit
In this age of integrated circuits, it is difficult to imagine why the digital logic interfacing the digital encoders, displays and the outside world, occupied four standard 6 ft racks. Rather than the binary circuits we are familiar with today , they utilized decimal counting tubes, called Magnetron Beam Switch Tubes [in some documents called Magnetic Beam Switch Tubes ]with a clock speed of [probably]1 MHz, a very high speed for the time. The outputs of a whole range of these tubes were fed through AND gates to provide the various timing pulses required. So, should a pulse of 8 time units be required starting at time 150, the the 10 unit pulse from time 150 to 160 and the 8 unit pulse from time 150 to 158 would be ANDED together to give the required gate time. The console displays showed azimuth and elevation in mils to some fractional part of a mil, and range in yds, but to base 8, that is in Octal, the conversion of the binary encoder outputs to octal being a trivial problem, whereas conversion to decimal, in those days long before today’s microprocessors were available, would have required much more complex circuitry.
One slight problem
In those pre- electronic calculator days the Loss of Signal azimuth and range co-ordinates of a test body had to be converted from Octal to decimal manually, the radar crew used perspex covered tables to rapidly do the necessary conversions so that the recovery crews had a starting point for their searches within couple of minutes of the body hitting the ground. The crews became very fast in carrying out this part of the mission,
Data was fed, via WRE designed error correcting digital data system, to the Range Head for use in data reduction and for the derivation of pointing data for other instrumentation. For one particular series of missile launches, the WRE staff designed and built a general purpose digital computer, as one did at the time. One of the functions of this machine was to send the Self Destruct signal to the missile should it stray outside the safe path limits. To that end the front panel of the machine had a very large red lamp labeled ‘One million pounds’, equivalent to over $22 million dollars today. If the lamp came on the missile was automatically blown up.
On the big day, the launch proceeded, the radars locked on, the missile rose majestically, all seemed well. Then, as programmed, the missile began to tilt onto the planned trajectory-the R38 crew reported “Signal level falling!”, followed by the same report from R39. Rapidly followed by almost simultaneous reports of signal loss. The missile had been destroyed, and fell to the ground in a cloud of metal fragments.
The post mortem revealed that, a significant fact had slipped through the cracks in the planning process, namely that the radars were linearly polarized. The horn antenna on the missile was also linearly polarized. When the missile was vertical the polarisations matched. But, when the missile tilted the signals were no longer of the same polarization, the received signal level fell, the computer switched from R38 to R39, found it also had a low signal level, and as programmed , checked R38 again. Finding the signal was getting lower, the program lit the red lamp and sent the Destruct signal, because, once track was lost the missile impact point was no longer predictable.
There was a deal of fuss, the R38 people had to carry out a test to prove that antennas have to be of the same polarization. This involved hauling to the top of the BST, a 45 metre high tower, a very beautifully made horn antenna with a carefully graduated scale showing rotation in degrees etc etc. Any textbook would have yielded the answer. Oh dear!!!!
The radars were beautifully built to Military Specifications, absolutely precision devices. One fact which sticks in my mind was that the pedestals were equipped with precision levels for alignment to true local horizontal. The bubble levels were so sensitive that alignment checks were carried out late at night, when the pedestal had cooled to ambient temperature. Personnel carrying out the checks had to even use indirect torch light to examine the bubbles in the levels as the energy from a direct torch was sufficient to minutely raise the temperatures of the bubbles and thus distort them, distortion which would lead to erroneous results.
From time to time very precise checks of the antenna positions were carried out by performing star calibrations. These utilized a film camera being mounted on the side of the antenna, positioning the antenna to designated star positions, exposing the film frame by frame, with a timing reference. The film was then processed and the data reduction carried out. If all was right, the star appeared in the correct location on the film.