Notes on a Visit to Hughes Aircrrift Co, Culver City, Los Angeles on May 15th and 16th, 1961.
Los Angeles,
Persons seen:-
Mr. W.E. Hinds - Sales
Mr. W. Dickman - Tarran
Mr. E.V. Phillips - C.W. techniques
Mr. R. A. Gehrke - Advanced Fire Control Systems
Dr. H.T. Osaki - Parametric amplifiers
Visitors: - V/c.Cdr . Biggs - Canada
Mr. H. H. Keefe - Canada
Dr. T.S. England
Mr. R. Aspinall
Mr. R.B. Clayton
1. Introduction
Some difficulty arose concerning clearances and there was no technical discussion on the morning of May 15th. The discussion was mainly concerned with the ASG-18 radar and the associated GAR 9 missile. In addition there was some discussion concerning parametric amplifiers.
2. ASG-18
The ASG-18 is an airborne interception radar intended to work in conjunction with the GAR 9 missile. The ASG-18 has the following characteristics
Frequency X band
Transmitter power 3kW mean, 12kW peak
Pulse repetition frequency 250kc/s
Transmitter pulse length 1 usec
Receiver pulse length 2.4 usec
Doppler frequencies searched +- 40 kc/s relative to the Doppler frequency of the carrier aircraft
Aerial search rate 115deg/sec
Transmitter is frequency modulated for ranging with a modulation frequency of 500c/s and a deviation of 500c/s
Additional information not obtained on present visit -
Aerial diameter 40"
Aerial gain 36.9 dB
(1) Aerial Search +-40deg x 4 bars
time for search 3.8 sec
range 100n.m.
Cumulative probability at this range 90%
on a target of speed Mach 5
and echoing area 10 sq m
(2) Alternative scan patterns such that azimuth search is increased to 70deg
and/or number of bars is increased to 8
Conical scan frequency in range 200-400c/s
2.1 Transmitter
The basic frequency is obtained from a temperature controlled crystal oscillator at about 100 Mc/s. The microwave frequency for the transmitter is obtoined by multiplying this frequency by 120 in 2 or 3 stages. The transmitter and local oscillator are phase locked to this source. The bandwidth of the phase locked loop waas thought to be about 200,000 c/s.
The power is then amplified by the travelling wave tubes in series (types HAX 5A and Hax 4). A block diagram is shown in Figure 1. The transmitter power is 3kW mean and 12kW peak and the overall efficiency is 10%. The efficiency of the final stage is 12.5%.
The transmitter specification was such that targets should be visible within 2 kc/s of a large clutter signal. The specification states that the frequency modulation noise in a 250c/s bandwidth shall be less than 75- 80 bB below carrier.
2.2 Duplexer and crystal protection
The aerial has a V.S.W.R. of 1.4 over a 10% band, and thus the peak power returned may be 500 watts.
A gas tube was not incorporated because with this power level it would have had a short life. A ferrite switch was used using a cooled tube of ferrite. This was followed by a low power T.R. tube.
The loss of the ferrite was 0.5 dB. The circulator was liquid cooled and gave 33-38 dB of isolation and a loss of 0.4 dB. It was 4 in. long and the ferrite was attached to the walls. The ferrite was cut into small lengths in order to avoid magnetostriction resonance at the p.r.f. frequency (see fig.2).
When a parametric amplifier was used, noise from the keep alive was observed.
2.3 Receiver
Intermediate frequencies of 30 Mc/s and 1.4 Mc/s are used.
2.4 Clutter rejection
It is arranged that two of the filter banks search in the clutter-free region above the doppler frequency corresponding to the aircraft velocity and the other two filter banks search in the clutter region.
A clutter rejection filter is provided.It is arranged to give 80 dB rejection over a bandwidth of 700 c/s.
Aircraft velocity and position of the aerial dish are used for initial frequency setting of this filter. A frequency lock loop is also provided in order that the frequency of the clutter rejection filter should coincide with that of the main beam return. This loop has a two second memory ( see fig. 3).
The clutter reject filter is provided only in the acquisition mode.
Once locked, the notch filter is removed in order to enable crossing targets to be tracked.
2.5 Multiple filters
The receiver incorporates 4 banks of multiple filters each covering a bandwidth of 20 kc/s.
The filters have the following characteristics:-
| Number of filters in a bank | 40 |
| Bandwidth of filter to 2 dB points | 500c/s |
| Centre frequency | 1.4Mc/s |
| Insertion loss | 1.5dB |
| Type | 2 crystal lattice |
| Size of 1 filter | 1 cu.in. |
| Long term stability of filter plus detector | 1 dB |
| Cost each | $130 |
| Change in frequency for 50°c temperature change | 10-20c/s. |
Each filter bank is filter. Each amplifier preceded by is provided an amplifier and a 20 kc/s. wide with separate a.g.c.
A separate frequency control loop, previously described, ensures that the doppler frequency from the main beam return appears between the second and third bank of filters.
With the separate a.g.c. control it will be possible to observe a small target signal in the 3rd and 4th bank and it should be possible to observe simultaneously a larger target signal in the 1st or 2nd bank.
A switch, using Trochotron valves, is used to sample the output of each filter five times in the time that the aerial moves through the target. A dummy filter is provided to give an earthed signal once per switch cycle and enable the levels to be set.
One of the difficulties in the multiple filter bank is the provision of adequate drive power. In order that the detector should be linear it is necessary to have 3V r.m.s. at the detector. For this it is necessary to have a 5V at the input of the filters and thus use a generator giving 20 volts and feed each of them in series with a resistance of 5000 ohms equal to the filter impedance, This means 30 watts is required (see fig. 4 )
Now a number of filters are joined in parallel as shown in figure 5. This has meant that the power consumption has boon reduced. The inductances are incorporated in order to match out the capacitance of the filters when off tune.
A complete filter system has been drawn out but not made. 300 filters plus the switch require a volume of 1800 cu.in.
The Gain stability of the filters is better than 1 dB but at the moment it is effectively 2 dB as non precision resistors are used. One of the variations arises because of magnetisation of the ferrite transformers.
The effect of gain variations of the filters is reduced by the circuit shown in figure 6. The threshold setting is obtained by observing the average noise in the filter before the signal arrives. This gives a slight deterioration of performance, say 1/2 dB.
The sensitivity of the filters is such that there is a 50% probability of detecting a signal having a 4 dB signal to noise ratio when the aerial is scanned at a rate of 115°/sec, This corresponds to a time of 22 m.sec. on the target. There was one false alarm whole filter bank per minute.
Hughes said that the multiple filter system was very trouble free and did not need resetting in a period of a month.
When an alarm is received in one filter an attempt is made to stop the aerial and return it to its original position. This takes 0.1 sec.
The position of the trochotron switch when an alarm was received is used to produce a pseudo signal using a digital oscillator which is applied to the input of the receiver. A fast frequency sweep is used in order to lock on in frequency to this signal (fig. 7),
The position of the switch is altered and the real signal is applied. Velocity lock is then obtained on the real signal.
In order to tell whether a signal is present or not an internal coherency check is used using a jitter oscillator.
The particular method of instrumenting range requires that the signal should be at the centre of the narrow band filter. There is a "lean" of 20 c/s at the output of the discriminator and Hughes have considered the use of a servo system to reduce this to zero.
A phase lock loop is provided in order to ensure that the frequency deviation remains at the centre of the narrowban filter.
One difficulty with the system is that it is necessary to ensure that the frequency deviation produced by the jitter oscillator is small. Otherwise, the measurement of range will be incorrect. In order to avoid this difficulty the system of figure 8 has been adopted. This enables the amplitude of the jitter oscillator to be increased. It was stated that this circuit was working.
It is possible to hold lock on to a slowly accelerating target signal that is 6 or 8 dB below the detection level.
2.6 Angular lock
The operator is presented with a display of azimuth versus velocity with a marker showing the elevation.
A strobe is put on in azimuth and elevation by the operator or by outside information.
If there is an output from the filter at the designated azimuth and elevation, the angular position will be remembered and the aerial will return to this position in O.l sec.
A conical scanning receiver is used for providing angular information.
2.7 Ranging
Range is measured when the aerial is locked on the target.
The p.r.f. is fixed as a subharmonic of the intermediate frequency and so this cannot be varied for ranging.
Range is measured by frequency modulating the transmitter when the aerial is locked. The modulating frequency is 500 c/s and the modulation index equals one.
The phase difference between the frequency modulation on the transmitter and on the receiver is measured. It is thus necessary to ensure that the time delay of any narrow band filter does not vary and this may be very difficult to achieve. There appeared to be no advantage in this method of instrumentation compared to that used by Raytheon in the Hawk system and in fact Hughes are considering the possibility of
having one modulation for acquisition and ranging.
The ranging accuracy is of the order of ½ mile corresponding to 1deg of phase error.
2.8 Aircraft trials
Aircraft trials of pulse doppler radar started in July, 1957.
Some of the equipment has been installed in a T-29 (Convair two engine) aircraft. The mean power was 200 watts and a duty cycle of 1/10 was used.
A single frequency swept filter was used. An attempt was made
to observe the scattering coefficient of the ground and the desert and an attempt was made to lock on to a Canberra target fitted with a Lunberg lens having an echoing area of 600 ft2 , They measured the echoing area of the Canberra and found that it was 80 s q.ft. without the lens and 420 sq. ft. (±_ 2 dB accuracy) wih the lens. However, fading having nulls 40 dB down was observed when the Canberra was equipped with the lens. Hughes could not understand this.
An attempt was made to measure the spectrum returned when flying over the sea and over the desert. No quantitative information was provided.
3. GAR 9
GAR 9 is a missile using a semi-active form of homing and is associated with the ASG 18 airborne interception radar.
Hughes considered the possibility of injecting CW, into the transmitter but rejected this because of weight considerations.
The present missile uses the A.I. transmitter as the transmitter of a semi-active radar and the receiver is incorporated in the missile.
The missile has the following characteristics , -
| Weight | 800 lb. |
| Receiver dish diameter | 10-12" |
| Range at launch | 30-50 n.m. |
| Control surfaces | moving tail |
| Polarisation | vertical |
| Roll control of missile | roll rate stabilised |
3.1 Progress
A laboratory model has been vibrated and components had been
flight tested.
3.2 Static split receiver
A static split receiver is used.
There are three channels at the intermediate frequency each
incorporating a narrow band filter, whose bandwidth is about twice that of Sparrow. The outputs of these channels are frequency multiplexed to simulate a conical scan system (see fig.9).
There is no increase in bandwidth in the home on jam mode,
The effect of phase changes in the I.F. amplifiers had not been studied.
3.3 Dish system
Various types of aerial have been tried including,(
(1) a dish with four helical feeds to give circular polarisation, this was not actually used.
(2) a dish with four dipole feeds.
and
(3) slot arrays for each of four quadrants in a circle.
A static split receiver is used and there are three mixers and head amplifiers on the dish assembly.
The dish is stabilised using a free gyro and torque motors are used to precess the gyro.
3.4 Reference power
No special reference aerial is usocl on the launching aircraft, so that reliance is placed on the power recieved from sidolobes of the A.I. transmission.
The transmitter pulses of the A.I. radar are received at a reference dish at the rear of the missile. This signal is effectively passed through a narrow band filter.
Care has, of course, to be taken to ensure that the filter is
centred on the fundamental frequency of the original transmission and not separated from it by the pulse repetition frequency.
3.5 Local oscillator
A solid state local oscillator is used giving a power output of 20 m.w. at X band. This signal is derived from a crystal oscillator in the range 50 to 100 Mc/s.
3.6 Receiver
The receiver is turned off when the transmitter pulses are received provided the range from the launching aircraft to the missile is small. At long ranges from the aircraft this gating is removed.
A clock is being considered as giving some estimate of aircraft missile range.
There is no other gating of the received signal.
The receiver is turned off for the first one or two seconds of
flight and an attempt is made to lock on to the target signal. Two accelerometers are provided in order to predict the doppler frequency, and this means that it is necessary to search 10 kc/s. Without accelerometers it is necessary to search 30 kc/s.
4. Parametric amplifiers
Hughes Aircraft Company were doing experimental work on parametric amplifiers and would supply such amplifiers commercially.
Some manufacturers attempt to use 1N 23 cases for holding the varactor diodes but it was considered that the cases are not suitable.
4.1 Quasi degenerate amplifier
The pumping frequency is twice the actual frequency.
If commercial diodes are used with a cut off of 60,000 Mc/s it is possible to obtain a noise factor of 4.5 - 5.5 dB at X band. This includes the circulator loss and the effect of a 10 dB noise factor in the next stage.
The gain of the amplifier is 17 dB and this can be obtained over a bandwidth of 30 to 100 Mc/s.
The amplifier is linear wth a power input between 10-14 and 10-5 watts. These input levels do not deteriorate the noise factor when measured over a broad band. No measurements have been done with a narrow band receiver.
The amplifiers are at present available for sale.
4.2 Non degenerate amplifier
At X band a 3 dB noise factor has been obtained.
Silver bonded diodes are used with a cut off frequency greater than 100,000 Mc/s. The pump frequency is 35,000 Mc/s.
Hughes make these diodes but at the moment they are only getting a yield of 5%. Cooling of the diodes by Peltier cooling gives an improvement in noise factor. An improvement of O.2 dB was obtained with a 20°c difference.
An S band non-degenerate parametric amplifier giving a 5% bandwidth is available for sale. At X band, a quasi-degenerate amplifier gave 5% bandwidth using selected diodes.
4.3 Use in a C.W. receiver
A parametric amplifier has been incorporated in a C.W. system,
It is satisfactory with a spillover power of 60 dB above noise in a 500 c/s bandwidth.
The normal noise factor was 7.5 dB (using a 1N 23F cost $12) and there uas 2.5 dB improvement with the parametric amplifier.
Dual infra-red/radar system
Hughes have done a paper study, but this was not an optimum system.
)
