Signals Museum
Radar
home page

Please note:

There are multiple references to the 4 diagrams that come with this document. Due to the large size of 3 of these, you will find them in an Adobe Acrobat document (they do not come out well on a web page). I would recommend that you download this pdf document so that you can view it in Adobe Acrobat as opposed to in your web browser. The diagrams are here. (358kb)

Despite there being a list of illustrations, there are no illustrations with the document. I assume (but haven't checked) that these are in the A.P.s (see the Radar Home page)

This page was generated by using Optical Character Recognition on the scanned (typewritten) pages. I've tried to remove all the errors, but wouldn't guarantee that I found them all. This document brought back a lot of memories for me as I worked with this equipment (among others) for some 10 years at Staxton Wold.

Colin Hinson 21st February 2018.


MARCONI'S WIRELESS TELEGRAPH COMPANY LIMITED
RESEARCH AND DEVELOPMENT LABORATORIES
GREAT BADDOW,
ESSEX.

GROUP: Radar Data Generation Laboratories.
SECTION: MTI Systems.
Report: RD/ G.1528 Date August 1961.
Copy: No. 22
Security Grading: was Secret, then Restricted, now unclassified.
Period Covered: Up to 1961.
Job No.: 40020

TYPE 84 RADAR,
DESCRIPTION OF SIGNAL PROCESSING FACILITIES

Work Carried out by  W.S. Mortley.
                                  A.D. Slocombe.
                                  A. Miller.
                                  and others.

Report Prepared by A. Miller.

 

Approved by Dr. G.N. Coop.

MARCONI'S WIRELESS TELEGRAPH COMPANY LIMITED

Research Division,
RD/G.1528                                                                                                                                                                          Gt. Baddow, Essex.

MTI Systems Section. 9th August, 1961.

TYPE 84 RADAR,
DESCRIPTION OF SIGNAL PROCESSING FACILITIES

ABSTRACT
A description of Type 84 Signal Processing is given, with an indication of the theoretical and practical limitations of the system. Photographs show the results achievable in various clutter conditions.

CONTENTS

SECTION  
1. INTRODUCTION,
2. CLUTTER PROCESSING- BRIEF DESCRIPTION.
  2.1Circular Polarisation.
  2.2 Swept Gain.
  2.3 Logarithmic Receiver +PLD.
  2.4 Moving Target Indication.
3. ANTI-JAMMING - BRIEF DESCRIPTION.
  3.1 Noise Jamming.
  3.2 CW Jamming.
  3.3 Impulsive Jamming.
  3.4 Window.
4. GENERAL PRINCIPLES OF MTI.
  4.1 Single or 2-Pulse Cancellation
  4.2 Degree of Cancellation Obtainable.
        4.2.1 System Stability.
        4.2.2 Clutter Fluctuations.
        4.2.3 Scanning Effect.
  4.3 Double or 3-Pulse Cancellation.
  4.4 Target Visibility.
        4.4.1 Blind Velocities.
        4.4.2 Blind Velocity Bands.
        4.4.3 Blind Phases.
  4.5 Doppler Compensation for Moving Clutter.


LIST OF ILLUSTRATIONS.
(There are no illustrations with the document I have - try the A.P.s)

SK - SP 119 Discrimination against Angels by means of Swept Gain.
   
RD 23382  
Sheet 1. MTI Systems.
2. Velocity Response of Moving Targets.
3. Response of Phase Detector.
4. Fully Processed Display.
5. PRFD Effect on Single Pulse.
6. PRFD Effect on Trains of Pulses.
7. Principle of Clutter Switching.
8. Principle of Angel Switching.
9. Principle of Pulse Length Discriminator.
10. Doppler Compensation, Vector Diagrams.
11. Proposed Swept Gain Law.
12. PRF Generator, Frequency Division.
RD 14376/D Signal Processing Scheme for Type 84 Radar.
RD 10189/B Simultaneous Cancellation of Fixed and Moving Clutter.
RD 14377 Doppler Compensation Scheme for Type 84 Radar.
E/RAD 47814 Generator Sweep (Video).
E/RAD 35459 Comparator, Signal.


SECTION
 
5. GENERAL DESCRIPTION OF SYSTEM.
  5.1 AJ Channel.
  5.2 Linear Channel.
  5.3 Log+PLD Channel.
  5.4 MTI Channel.
  5.5 Video Switching.
  5.6 Description of Operator's Facilities.
6. MORE DETAILED DESCRIPTIONS.
  6.1 PRF Generation.
  6.2 Cancellation System.
  6.3 PRF Discrimination.
  6.4 Clutter-Operated Switching Unit.
    6.4.1 Clutter Switching
    6.4.2 Angel Switching.
  6.5 Doppler Compensation System.
  6.6 Pulse Length Discriminator (PLD).
  6.7 Generation of Swept Gain Waveforms.
  6.8 Application of Swept Gain Waveforms.
    6.8.1 In Log+PLD Channel.
    6.8.2 In MITI Channel
7. TEST EQUIPMENT.
  7.1 Test Pulse Generator.
  7.2 Noise Generator.
  7.3 Artificial Permanent Echo Generator.


CROSS REFERENCE, BLOCK DIAGRAM TO UNITS

RD 14376/D

SYMBOL
UNITS
AIJ 10D/22628, 10D/22646, 10L/16758
AJ 10U/17516-7-9
ASD 10D/22652
CD 10D/22630
COMP. 10U/17485
COSU 10F/20585
DC See RD 14377
ESA 10U/17487, 10D/22646
FDC 10D/22645
G 10D/22630, 10V/16468
GC 10L/16757
GSG 10L/16757, 10V/16456
IF 10U/17488
IFSU 10F/20586
LD 10D/22642
LIM 10F/20590
ND 10L/16757
OR 101/16757, 10D/22628, 10U/17487, 10D/22646
PLD 10D/22647
QO 10V/16461
RG 10V/16464
RO 10V/16468
RY 10F/20589-90
SAN 10V/16460
SM 10V/16461, 10D/22700
TED 10L/16756-8
TSG 10S/17707, 10S/17716
VR 10L/16757, 10F/20585, 10U/17487
VS 10F/20588
+ 10D/22624
- 10D/22624-5

RD 14377

SYMBOL
UNITS
AMT 10AE/2147
BPF 10D/22629
ES 10F/20591
FD 10L/16760
M 10D/22629
MR 10L/16763
QO 10V/16463
RO 10V/16468
SR 10L/16760
Y 10V/116463


TYPE 84 RADAR.
DESCRIPTION OF SIGNAL PROCESSING FACILITIES.

1. INTRODUCTION.
The Type 84 is a high power long range radar, operating on L Band, about 23 cms., at a nominal PRF of 250 c/s. and a normal scanning speed of 4 r.p.m. The pulse length is 10μsecs. and the peak power 5 megawatts. The aerial beamwidth is 0.9° which results in 9 pulses per beamwidth, between 3 dB points.

Signal Processing is employed in order to obtain a radar display showing the maximum number of targets both in the clear and in the presence of clutter or jamming.

Clutter may consist of ground-clutter or permanent echoes, sea-clutter, rain, "angels", or any of the above at long range visible under conditions of anomalous propagation, i.e. "anoprop". Jamming may be caused by pulses, CW., "window", or noise, as generated, for example, by a carcinotron jammer. For the purpose of signal processing, window is treated as moving clutter.

Owing to the high power of the radar, clutter would often make the display virtually useless if only simple unprocessed radar were used. At an experimental site in Essex some ground-clutter echoes are 70 dB above noise.

Heavy rain is commonly 30 to 40 dB above noise and angels are frequently troublesome up to about 80 miles from the station. Anoprop is much less frequent but it has been observed at 320 miles range. The photographs show the clutter which may be encountered and the effects of signal processing.

The anti-clutter facilities of the equipment are:-

Each of these methods has its advantages and limitations so it is essential to use the best combination to suit the prevailing conditions. The methods of achieving this in the Type 84 result in the "Fully-Processed" display.

Of the above measures, MTI is the most effective against ground-clutter and window, as it can in normal conditions provide "sub-clutter visibility" (SCV).

That is, targets may be seen in clutter even though their echo strength is less than that of clutter. It may be used against fixed and moving clutter, but the equipment for the latter is complex and requires a fair degree of skill on the part of the operator. A limitation of MTI is the loss of a proportion of targets owing to blind velocities. This loss occurs whether clutter is present or not. Ideally therefore, MTI should only be used when the presence of clutter makes it necessary.

Rain is often difficult to deal with effectively by MTI, owing to its turbulence, which limits the SCV obtainable, and Circular Polarisation is more effective. It provides SCV of 12 to 14 dBs., requires no adjustments by the operator, and the targets are not subject to blind velocities. Its use cannot be confined to selected areas of rain however; it must be applied over tie whole screen. As it causes a loss of target strength of 3 to 5 dBs. provision must be made to switch it off when the maximum sensitivity of the radar is required.

The same remark applies to Swept Gain, which obviously causes a loss of target strength in the region where it is used. It must be switched off therefore, when maximum sensitivity is required in this region.

The principle of confining the use of MTI to the clutter areas is adhered to in the Type 84. The fully-processed display has a "background" of Log+PLD. This causes no loss due to blind volocities but gives no sub-clutter visibility. In areas of clutter MTI is switched in to the display. The switching is normally effected by a voltage which is derived from the clutter itself. This technique is referred to as "clutter-switching" and is an elegant means of limiting the use of MI to the clutter areas. In some cases clutter may be of such a nature that it does not provide an effective clutter-switching voltage. In such a condition MTI may be applied all over a selected area. As targets in such an area will be subject to blind volocities, it is essential that the selected area should be as small as possible.

Briefly, therefore, the fully-processed display consists of a background of Log+PLD with clutter-switched MTI in areas of clutter. Swept Gain, variable in degree by means of a switch, is used for a limited area round the station, and Circular Polarisation is applied all over the screen. The last two facilities will be dispensed with when maximum sensitivity is required.

It may be wondered why Linear Radar is not used as the background instead of Log+PLD. This is because Log+PLD gives target visibility in the clear which is comparable with that of Linear, with the addition of several anti-clutter and anti-jamming features which will be described later.

2. CLUTTER PROCESSING.
Brief Description.

2.1 Circular Polarisation.
Experiment seems to indicate a diminution of rain echoes of about 18 dB and a loss of target strength of 3 to 5 dB depending on the aspect of the target. A useful gain of visibility in rain is thus obtained. As already noted, CP must be used all over the screen or not at all, so provision is made to switch it off when maximum sensitivity is desired.

It is important to note that as CP is effected at the aerial, it is in series with any other signal processing which may be used. If we consider for instance, strong rain 40 dB above noise, masking a target 20 dB above noise, a sub-clutter visibility of 20 dB would be required for the target to be clearly visible. With rain returns of normal turbulence this would be too much for MTI to achieve on its own. Assume however, that CP reduces the rain to +22 dB and the target to +15 dB. The rain would still be very visible and the target invisible. Now however MTI would only be required to provide rather more than 7 dB sub-clutter visibility, which would be well within its capabilities.

2.2 Swept Gain.
As ground-clutter, sea-clutter and angels are usually only troublesome in a restricted area round the station, it is sometimes useful to attenuate the signal for a limited range only. The degree of attenuation applied, and its variation with range, must be arranged according to the site, characteristics and purpose of the radar, and the vertical polar diagram of the aerial.

Swept gain must obviously cause the loss of some targets but it is sometimes the only effective means of reducing the echoes from angels, as their velocities may be too random to allow cancellation by MTI. Swept gain is available on both the MTI and the Log+PLD channels but the law to be applied is as yet only provisional and may be modified after flight trials on site. SK-SP 119, which has been derived from the Aerial VPD, RD/18749 Sheet 1, shows the response of the desired target and of what are considered to be typical angels in the main lobe. Their responses are also shown with the proposed swept gain law applied.

In the MTI channel swept gain is applied to an amplifier valve in the IF circuit. Swept gain cannot precede the log. amplifier, as the log. characteristic would neutralise the change of gain. The desired effect is achieved therefore by baseline clipping of the video signals after the log. amplifier, the amount of clipping being dependent on range.

2.3 Logarithmic Receiver + PLD.
It is known that most of the troughs of a noise voltage are not more than 20 dB below the r.m.s. value while most of the peaks are not more than 10 dB above the r.m.s. value. If therefore any level of noise be applied to a logarithmic amplifier, the output excursion will be unchanged, although the "DC" will be increased with increased input. This DC level may be eliminated by differentiation and the resultant on the screen is a constant noise level giving a C.F.A.R. characteristic. For this to be so the log. law must extend to 20 dB below the normal thermal r.m.s. noise level and 10 dB above the highest r.m.s. noise level for which it is desired to maintain C.F.A.R. characteristic. Clutter is also greatly reduced by removal of the DC component of the output.

The simple use of a short time-constant however, reduces the signal amplitude and produces leading and trailing edges of the clutter, only one of which may be removed by rectification. The pulse length discriminatic used after the log, receiver is a refinement of the differentiation process. Signals up to 10μsecs., the transmitted pulse length, are passed at full amplitude. Any signals of twice the normal pulse length or more are suppressed. Any clutter signals therefore of more than 20μsecs. are eliminated, without leading or trailing edges. As the PLD operates by virtue of pulse length it is equally effective whether the clutter is moving or not.

The advantages of Log+PLD compared with MTI are:-

(a) There is no loss of targets due to blind velocities and target visibility is comparable with that from a linear receiver.
(b) It is effective against moving clutter without any adjustment of controls by the operator.
(c) It gives automatic protection against noise and CW jamming. The word "protection" is used here in a limited sense. The screen will remain clean in the presence of the jamming, but target visibility will be reduced by a degree corresponding to the jamming. The effect is, in fact, that of jamming-operated AGC .
(d) It suppresses impulsive jamming of more than twice the transmitted pulse length.
(e) It is relatively simple and inexpensive.

The limitations of Log+PLD are:-

(a) A target in clutter must be 5 to 10 dBs stronger than the clutter in order to be visible. With MTI a target may be visible even though it is weaker than the clutter.
(b) Permanent echoes from isolated masts, towers, etc., are passed at full strength. As areas of ground clutter contain many such echoes this results in a very dirty display at short range.
(c) There is no protection afforded against angels, which are discrete echoes. They may be quite dense up to 80 miles or more from the station. There is no doubt that they are often caused by birds and possibly sometimes by abnormal refractive effects. They move at low speeds and may often be usefully reduced by MTI with Doppler compensation.

The above comparison leads to the conclusion that MTI should always be used in areas of ground clutter and MTI with Doppler compensation is usually desirable in areas of rain or window. In areas free from clutter either Linear or Log+PLD should be used.

Linear gives slightly better visibility of weak signals, but this advantage is more than outweighed by the protection that Log+PLD gives against saturation of the screen by jamming. Log+PLD also suppresses fixed and moving clutter, though giving no SCV. It is thus a very useful emergency display in the event of a failure of the lel . For these reasons it has been chosen as the background of the fully-processed display in preference to Linear.

Circular Polarisation should always be used when rain is present except in conditions which render the reduction in signal strength unacceptable. Such conditions would be, for example, the presence of jamming, or the need to detect very weak targets.

2.4 Moving Target Indication.
The general principles of MTI are discussed in the published literaturel,2 In order to make this report self-contained however, they are briefly stated in Section 4, with some reference to 3-pulse cancellation, which is the system used in the Type 84.

In this report the term "MTI with Doppler" means MTI which is compensated for moving clutter.

The parameters of the Type 84 were dictated by considerations other than optimum MTI performance. Owing to the low PRF, which results in only 9 pulses per beamwidth, the "Scanning Effect" limits the cancellation, even of ideal fixed clutter, to about 20 dB with 3-pulse cancellation. This means that in order that the scanning effect shall not spoil the cancellation of permanent echoes, the "Pre-Cancellation Limit Level" (see Sect. 4.3), must not exceed 20 dB signal to shoulder noise. This rather low limit level results in "blind velocity gaps" of about 9 knots, centred at each multiple of 56 knots, and entails the loss of 9/56 or approximately 16% of strong targets due to blind velocities, assuming a uniform distribution of radial velocities.

For the cancellation of moving clutter however, a blind gap of 9 knots is often not sufficient, owing to the turbulence of the clutter itself. Rain in average conditions needs a blind gap of 13 knots, corresponding to a cancellation ratio of 14 dBs and a loss of strong targets of 13/56, or 23%, again assuming uniform distribution of radial velocities. To accommodate varying conditions three cancellation ratios are available, 20, 14 or 8 dBs. They are selected by means of a switch at the monitor console.

In the absence of rain, window, or angels, MTI would be applied to the fully-processed display only in the area round the station containing ground-clutter. This area is defined by a circle known as the "PE Circle". It is adjustable in radius to suit the site and is indicated - on the Monitor Console only - by a video marker.

If moving clutter is present however, MTI with Doppler may be used in any or all of three rectangles, which may be set up in any position and of any size. Each rectangle may have its own Doppler-compensating velocity applied within it. If two rectangles overlap, No. 1 takes precedence over No. 2, in the sense that the Doppler compensation set up in No. 1 rectangle applies in the area of overlap. Similarly No. 2 takes precedence over No. 3. This is indicated by the video markers.

A special case arises when any rectangle overlaps the PE circle and actually encroaches on the PE's themselves. It is evident that if the rectangle took precedence over the circle, it would spoil the cancel of the fixed clutter, as it would be set up for cancelling moving clutter. This is avoided by using two cancellation channels. Within the circle one channel always cancels the permanent echoes, as it can never have any Doppler compensation applied within the circle. The other channel cancels the moving clutter. The two outputs are passed to a coincidence detector which passes the smaller of the two signals. Both types of clutter may therefore be cancelled, without degrading either. If however, the fixed and moving clutter actually coincide in position on the screen, the result on the PPI will be the weaker of the two, so if they are of equal strength no improvement is obtained.

It should be noted that when a rectangle overlaps the circle, in the area of overlap two series of blind velocity gaps are operative. In the worst case this would lead to twice the usual percentage of strong targets which would be lost due to blind velocities.

As already stated in Section 1, the MTI is usually clutter-switched in order to avoid blind velocity losses where no clutter is present. The circle and the rectangles merely define the areas within which clutter-switching may take place. It is not therefore necessary that a rectangle be made as small as possible. If however "block MTI" is used in any rectangle, the area should be kept as small as possible.

The signal which acts as the switching voltage for clutter-switching is derived from an "addition" circuit located in, the first cancellation unit. This provides a gain of 3 dBs in signal to noise, compared with a linear receiver. It also has the advantage that when the Doppler compensation is correct, the clutter-switching is optimised, as the output of the addition circuit is then at its maximum.

3. ANTI-JAMMING PROCESSING.

Brief Description.

3.1 Noise Jamming.
As stated in Sect. 2.3, the Log+PLD channel is automatically protected by the logarithmic amplifier against saturation of the screen by noise jamming.

Saturation of the MTI channel is prevented by means of noise-operated AGC. After detection of the incoming IF signal a circuit incorporated in the Limiter Electrical Noise unit counts the number of noise spikes occurring between two specific voltage levels. The resulting count, after integration and amplification, provides an AGC voltage which operates on the Swept Gain amplifier valve, via an OR gate to which the swept gain waveform is also coupled. This ensures that only the larger of the two bias voltages determines the gain of the valve and prevents the application of unnecessary swept gain to signals which are already being reduced in strength by jamming. In order that the AGC circuit shall not be actuated by clutter, counting only takes place over the second half of the trace, that is, after 2 milliseconds. An IAGC amplifier is used in the MTI channel and the combination of this and the noise-operated AGC maintains the screen in a usable condition in the presence of up to 50 dB of jamming.

The above measures, i.e., log. receiver and noise operated AGC, will only prevent saturation and will not preserve the signal to noise ratio of the targets. In the presence of heavy jamming therefore, practically all targets would be lost on both channels. In this condition it would be necessary to operate the ANTI-JAM switch on the Control Console. This would result in the Log+PLD and the MTI channels being fed with IF from the AJ (Dicke Fix) receiver, which is designed to deal with carcinotron-type noise jamming. This receiver, besides preventing saturation, preserves a signal-to-noise ratio of about 17 dB more than the other methods described.

The AJ receiver will feed to the system a noise level about 50 dB higher than will be supplied by the linear head amplifier, with a maximum signal-to-noise ratio of 17 dBs. The log. receiver will cope with this high noise input automatically. In the MTI channel, the noise-operated AGC will reduce the excess noise to the normal level used in the MTI circuits.

The Dicke Fix receiver is the subject of a separate Report3.

3.2 CW Jamming.
This type of jamming is suppressed by the PLD circuit, so the Log+PLD channel is automatically protected against saturation of the PPI. In the event of strong jamming, however, either the head amplifier or the logarithmic amplifier may be overloaded, in which case all targets would be lost. In the case of weaker jamming, target visibility would depend on the relative strength of target and jamming. It is probable that a target would need to be 5 to 10 dB stronger than the jamming in order to be visible.

In the MTI channel, the signals are passed through a PLD circuit before they go to the fully processed display. This avoids saturation of the PPI screen. In the presence of strong jamming, the IAGC receiver would reduce the overloading of the MTI circuits, and retain some incremental gain for targets which exceed the strength of the jamming. The effect of the noise-operated AGC would be purely fortuitous, depending on the jamming frequency. It should be noted, however, that cancellation would not be effective in the presence of CW jamming, as the jamming would produce a non-coherent beat frequency with the IF signals. This means in effect that the MTI is rendered virtually useless in the clutter areas. In the presence of CW jamming, therefore, it would be necessary to rely on Log+PLD all over the screen.

3.3 Impulsive Jamming.
This is dealt with by the method known as PRF Discrimination, in the case of both the MTI and the Log+PLD channels. Signals are fed to a coincidence detector together with the same signals delayed by one pulse period. The target signals, being a train of several pulses with the correct spacing, are passed by the coincidence detector with the loss of only one pulse. The interfering pulses are suppressed unless they happen to be nearly synchronous with the PRF. This system is fairly easily applied to equipment which incorporates MTI, as delay lines accurately matched to the PRF are necessary for the MTI circuits. The application of "PRFD" is complicated in the case of the MTI channel by the fact that in a 3-pulse MTI system each single pulse is converted into three pulses. In the Type 84 the three pulses are reduced to a single pulse again in order that PRFD as described in Section 6.3 may be used. The system gives excellent protection against impulsive jamming without any loss of signals. There is in fact a gain in signal-to-noise ratio, but at the cost of increased spottiness of the noise.

3.4 Window.
"Window" is the name given to the echoes from masses of dipoles, made of metal foil, which are dropped by enemy aircraft.

When first dropped, the window is in a compact mass with all its elements subject to the same wind conditions. It is therefore very amenable to cancellation by MTI with Doppler compensation. In this early stage, however, it does not constitute a very grave threat, being too small to obscure aircraft efficiently.

After being dropped the window starts to fall and is dispersed over a wider area by the wind. The rate of fall, measured during tests, is about 300 feet per minute. After several drops have been made therefore, a considerable difference in altitude exists between the early drops and the later ones. This complicates the process of cancellation owing to the different wind conditions which may prevail at different heights, After several drops have been made therefore, it is often necessary to use more than one Doppler correction condition in order to obtain good cancellation. In general, however, it is found that window is easier to cancel than rain. It is found in fact, that if a sub-clutter visibility of 12 dB is acceptable, with corresponding blind gaps of 13 knots, window may be eliminated over large areas. In order to achieve this however, the necessary velocity compensation must be found by a skilled operator, aided preferably by some form of simple calculator.

4. GENERAL PRINCIPLES OF MTI.
Although the following is mainly a statement of principles reference will be made to their embodiment in the Type 84.

The object of MTI is to remove from the PPI echoes from fixed objects such as surrounding hills, masts, towers etc., while retaining maximum visibility of moving targets. The echoes to be suppressed are called fixed clutter or permanent echoes. An extension of the basic MTI principle enables the echoes from moving clutter such as rain or "window" to be suppressed. This refinement of the system is known as Doppler Compensation.

4.1 Single or 2-Pulse Cancellation.
The principle of permenent echo suppression is based on the fact that the interval between the start of a pulse at the transmitter and the arrival at the receiver of the echo pulse depends on the radial distance of the target. The interval is therefore constant for a fixed target or permanent echo, but varies from pulse to pulse if the target has a radial component of velocity.

If therefore a CW oscillator, of high frequency stability, is set in oscillation by each transmitter pulse, so that its starting phase bears a, constant relation to that of the transmitter pulse the phase of the echo pulse from a fixed target will also bear a constant relation to that of the reference oscillator. The oscillator may in fact be regarded as a clock, of high accuracy and discrimination, for measuring the echo time.

The phase comparison between the echo pulses and the reference oscillator is made in a balanced phase detector and it is evident that the successive output pulses from the detector will be similar for a permanent echo but will differ if the target is moving with any radial component of velocity, except for certain velocities which are discussed in Sect. 4.4.

If two successive outputs from the phase detector be applied simultaneously to a subtraction circuit, the output from a permanent echo will be ideally zero while the output from a moving target will depend on the radial distance travelled between pulses.

In order to effect the subtraction the first pulse must be memorised or delayed for one pulse period. The delaying and subtraction of successive pulses may be made a continuous process by the arrangement of Sheet 1, Fig. 1. In the Type 84 the delay required is about 4 milliseconds and this is obtained by means of an ultrasonic delay cell using mercury as the medium. To achieve good cancellation the pulses must be passed through the cell with little distortion and the delay time must be accurately matched to the PRF. The delay through the cell depends on temperature, so to avoid the need to control this to close limits the delay time is used as the reference in the Type 84 system, the PRF being automatically controlled to match it.

As the starting phase of the reference oscillator is controlled by the transmitter, it is said to be "coherent" with the transmitter and is known as the "coherent oscillator" or "coho".

A small portion of the energy of each transmitted pulse is used as a "Lock Pulse" to control the starting phase of the oscillator, thereby making it coherent in phase with the transmitter. In order to achieve satisfactory locking the coherent oscillator is stopped after a suitable period by means of an inhibiting pulse. It then remains quiescent until
the arrival of the next lock pulse. In the Type 84 the quiescent period is the last 125μseconds of each pulse period. The output of the coho is fed to the phase detector or "Coherent Detector", where it is compared with the incoming echo signals. Such a cancellation system is shown in Sheet 1, Fig. 2.

In the arrangement of Fig. 2 everything is done at the transmitter frequency and this is not practicable at microwave frequencies. At such frequencies however the superheterodyne principle is used for a normal radar receiver and this may be applied to the lock pulse and coho. Such an arrangement is shown in Fig. 3. The local oscillator which reduces the incoming signals to IF is fed to a second mixer together with the RF lock pulse. This produces an IF lock pulse which controls the starting phase of an IF coherent oscillator. For MTI purposes, the local oscillator must have a high degree of stability, and is normally referred to as the Stable Local Oscillator or "Stalo" and the system is known as "Coho-Stalo MTI".

The starting phases of the transmitter and the stalo are common to both the incoming IF echo signals and the coherent oscillator output. They therefore cancel out when the two beat together in the phase detector. The output of the latter depends therefore only on the number of cycles executed by the coho and the stalo during the echo time. Both these oscillators must be stable, the absolute frequency stability required being the same for each oscillator. In the Type 84 the stalo frequency is derived from a crystal oscillator at about 10 Mcs., multiplied by 128. As the coho operates at the IF of 13.5 Mcs., the stability requirement is easier by a factor of about 100 and a carefully designed LC oscillator is adequate.

It is common practice to make the frequency of the coherent oscillator as nearly as possible equal to that of the incoming IF signals, as this minimises the bandwidth requirements of the following circuits, and makes less stringent demands on the stability of the cancellation circuits. When this is the case, the output of the coherent detector is as shown in Sheet 1, Fig. 4. The output is bipolar, being positive or negative according to the phase of the incoming signals relative to the coho. The figure shows three ideally stable permanent echoes and the output from a moving target which exhibits the "butterfly effect" due to the variation of phase from pulse to pulse. It is evident that the difference of successive sweeps would give no output from the fixed echoes.

It is sometimes advantageous to detune the coherent oscillator so that its frequency differs slightly from that of the IF signals. This is in fact done in the Type 84, for reasons discussed in Section 4.4. The output of the coherent detector then no longer consists of pulses which are either positive, negative or zero. Every pulse is symmetrical about zero, and contains the beat frequency between the oscillator and the signals, the number of cycles displayed depending on the amount of coho detuning and the pulse length. In the Type 84 the coho is detuned about 100 kc/s., so that each PE output pulse, instead of being as in Sheet 1, Fig. 4, would contain a complete cycle of the coherent low IF. The output from a moving target would also contain a complete cycle, but the waveform would not be identifiable owing to the change of phase from pulse to pulse.

4.2 Degree of Cancellation Obtainable.
The degree of clutter cancellation which can be achieved is limited by three factors; the stability of the system, the fluctuations of the clutter itself and the scanning effect due to the aerial.

4.2.1 System Stability.
In designing the Type 84 system, the aim has been to ensure that about 30 dB cancellation would be achieved in the absence of other limitations.

4.2.2 Clutter Fluctuations.
It is evident that with a perfectly stable MTI system, and the aerial stationary, perfect cancellation would be obtained, if the clutter itself were quite stable. This is rarely the case in practice however, even with ground clutter, owing to the movement of tree branches, leaves, etc., in the wind. A fixed object will present a stationary vector to the phase detector, resulting in complete cancellation.

If the object moves an appreciable fraction of a wavelength between pulses, however, the cancellation will be degraded. The effects of clutter fluctuations are obviously minimised by using a long wavelength and a high PRF. With the parameters of the Type 84, an object which moved 1 cm. radially between pulses would cause a vector shift of 30°, resulting in imperfect cancellation. In fact, however, with the Type 84, ground-clutter fluctuations are negligible compared with the effects of scanning. In the case of sea-clutter and rain, however, fluctuation of the clutter itself is often the main factor which limits cancellation.

4.2.3 Scanning Effect.
The scanning effect limits cancellation even though the clutter is absolutely stable. It is due to the fact that the beam illuminates a different area of ground at each pulse. It can be shown that when the beam is scanning extended clutter the echo at any given instant is the sum of the echoes from an area determined by the beamwidth and half the pulse length. For the Type 84 this area would be nearly a mile long by half a mile wide, at a range of 30 miles. At an instant exactly a pulse period later this illuminated area would be at the same range but moved in azimuth about 1/18 of a mile, as there are 9 pulses per beamwidth. The returned echo will therefore differ from the first one, owing to the different terrain which is illuminated. The amplitude variations of the echoes may be removed or minimised by limiting, but the phase variations still remain. It is therefore evident that complete cancellation is impossible, as the two inputs to the subtraction circuit are not equal. In practice, the signal to noise ratio is limited before the signal is fed to the phase detector. This signal to noise ratio is called the "Pre-Cancellation Limit Level" or PCLL. The optimum condition is that which causes the uncancelled residue of the clutter to be indistinguishable from noise on the PPI. If the PCLL is set too high the clutter is visible; if it is too low, black holes are seen in the PPI noise and there is an unnecessary loss of targets.

If the number of pulses per beamwidth can be made large, due to either a high PRF or a low scanning speed, or both, the change in the illuminated area from pulse to pulse is obviously small, resulting in good cancellation. The degree of cancellation obtainable is proportional to the number of pulses per beamwidth for 2-pulse cancellation, if the scanning effect is the limiting factor.

If 2-pulse cancellation were used in the Type 84, the cancellation of ground clutter would be limited to about 15 dB.

4.3 Double or 3-Pulse Cancellation.
The cancellation obtainable with 2-pulse MTI is hardly worth while with the Type 84. Moreover, the efficient cancellation of moving clutter such as rain or angels would be virtually impossible owing to the random motion of the clutter. For these reasons 3-pulse cancellation is used. This employs two cancellation systems in series. The output is
therefore proportional to the second difference of the incoming pulses instead of the first. The effect of amplitude and phase variations due to scanning or other causes is thus reduced. Reasonably stable ground clutter may be cancelled to about 21 dB. The pre-cancellation limit level must not therefore exceed 21 dB, if clutter is to be cancelled to noise level. In practice the output of the phase detector is set to 20 dB maximum signal to shoulder noise.

With a PCLL of 20 dB the sub-clutter visibility of an optimum velocity target at optimum phase is about 22 dB, the average SCV for all phases being 17 to 18 dB.

When dealing with moving clutter such as rain, cancellation is often limited to 14, or even 8 dB. This is due to the need to provide sufficient "Blind Velocity Band" to accommodate the random internal motion of the clutter. The corresponding pre-cancellation limit levels are therefore 14 or 8 dB, and the average SCV's are about 12 dB and 6 dB.

4.4 Target Visibility.
The visibility of a target, whether in clutter or in the clear, is subject to both "blind velocities" and "blind phases".

4.4.1 Blind Velocities.
As MTI suppresses the echoes from a target having no radial velocity, an aircraft which is moving tangentially to the beam will obviously be invisible. Blind velocities other than zero occur however. If a target moves between pulses a radial distance which is an integral number of half wavelengths, the path of the echo pulse will change by a whole number of wavelengths. The phase of the echo with respect to the coho will therefore be constant and the target will be cancelled. The radial velocities at which this occurs are given by:-

v= .0097 nλfr.

v being the velocity in knots, λ the wavelength in cms., fr the PRF in c/s. and n any integer including zero. With the parameters of the Type 84, this gives blind velocities occurring every 56 knots, approximately.

Radial velocities midway between the blind velocities, i.e. at odd multiples of 28 knots, produce 180° phase change at the output of the phase detector and thus maximum output from the subtraction circuit. They are called optimum velocities, therefore. Referring to Sheet 1, Fig. 4, the variation of the moving target output takes place at the Doppler frequency fd. At blind velocities this is an exact multiple of the PRF. Thus if the first pulse from the phase detector is of amplitude A, succeeding ones are also of amplitude A and produce no output from the subtraction circuit. At optimum velocities however, if the first pulse is of amplitude A, succeeding pulses alternate between -A and +A, the output after subtraction being 2A.

4.4.2 Blind Velocity Bands,
The response of an MTI receiver as a function of radial velocity is shown on Sheet 2. The response for both 2-pulse and 3-pulse systems are shown, compared with that of a receiver without MTI. The ordinates show the relative signal to noise ratios of the systems. The constants used are those of the Type 84, giving blind velocities about every 56 knots. Targets at optimum velocity and optimum phase give an improved signal to noise ratio compared with straight radar. The pre-cancellation limit level determines the actual signal to noise ratio of the output. The noise level shown on Sheet 2 indicates a PCLL of 20 dB.

It will be seen that targets become submerged in the noise over a band of velocities occurring every 56 knots. For 3-pulse cancellation this blind, velocity band has a width of 9, 13 or 19 knots, when the PCLL is 20, 14 or 8 dB respectively. The zones of blindness, in which some wanted targets may lie and are therefore lost, have the virtue of providing a velocity tolerance for the cancellation of clutter. This velocity tolerance is necessary in order to achieve cancellation in spite of clutter fluctuations and the scanning effect of the aerial, as discussed in Sections 4.2.2. and 4.2.3.

4.4.3 Blind Phases.
The response of a balanced phase detector, is shown on Sheet 3, as a function of the phase angle between the two inputs. A target at optimum velocity causes a phase change from pulse to pulse of π radians. If the first pulse occurs at zero phase on the diagram therefore, the output will be maximum positive. The next pulse will be at π radians giving an output of maximum negative, and result will be a large output from the subtraction circuit. This is the case of optimum velocity and optimum phase. If the first pulse occurred at π/2 radians however and the next pulse at π+π/2 radians, the output from the phase detector would be zero in both cases. This combination of optimum velocity and blind phase would result therefore in complete loss of the target. Blind phases can occur at velocities other than optimum. If one pulse occurred at A and the next at B the result after subtraction would be zero. This would cause the loss of only one strike however, as the next pulse at C would give appreciable output when subtracted from B.

The effect of blind phases can be avoided by detuning the coherent oscillator so that during the pulse the optimum phase will be passed through at least once. In the Type 84 the coho is detuned 100 kc/s, giving a complete cycle during the pulse. The difference frequency produced by offsetting the reference frequency is called the "Coherent Low IF" or CLIP.

The use of CLIF will avoid blind phases when targets are in the clear, but it is ineffective when they are in strong clutter. In the vicinity of the station clutter may be of very high amplitude, so in order to cancel it effectively it must be limited or compressed before it goes to the phase detector. A moving target in clutter may be regarded as a rotating vector which modulates the fixed clutter vector both in amplitude and phase. The phase detector will respond to both but after a large clutter vector has been limited only phase variations remain. The vector of a moving target at optimum velocity rotates 180° during each pulse period. If it is in phase with the clutter vector the phase detector will see no change of either phase or amplitude and the target will be lost. This is the blind phase. Optimum phase is in quadrature with this. For targets in weak clutter, below limit level, the blind phase is not so serious because when the phase is blind the amplitude change is a maximum for optimum velocity targets.

The effect of blind phases in clutter is mitigated to some extent in the Type 84 by the use of an IAGC receiver, instead of a linear receiver with hard limiting. This compresses strong signals but still allows some amplitude variation to remain, thus avoiding complete blindness when the phase is wrong. A further advantage of the IAGC receiver is that it avoids the sharp discontinuity of pulse amplitude which occurs if a steeply rising aerial response curve is limited hard. Such discontinuities tend to spoil the cancellation.

4.5 Doppler Compensation for Moving Clutter.
Referring to Sheet 1, Fig. 4, the output of the phase detector, the signal from the moving target, which is varying at the Doppler frequency, might be an echo from moving clutter. If the Doppler frequency can be reduced to zero the detector output will be constant and the clutter will be cancelled. This may be done by changing the reference frequency from the coherent oscillator by an amount equal to the Doppler frequency of the target.

If an area of rain surrounds the station and is moving with uniform velocity, its radial velocity will vary according to a cosine law as the aerial revolves through 360 degrees. The modulation of the reference frequency must therefore vary in a similar manner. In the Type 84 the reference frequency may be modulated to cancel clutter velocities up to 100 knots. The modulation of the reference frequency has the effect of shifting the whole velocity response curve of the system along the velocity axis. The shape of the curve will not be altered however, and the blind gaps will remain 56 knots apart. The shift will vary with the frequency modulation and will thus be sinusoidal at aerial speed.

In the Type 84 equipment three different values of Doppler compensation may be used at once, applied in three rectangular areas, adjustable in size and position. Three rectangles are provided because when attempting to cancel rain over a large area it is seldom that a single velocity condition prevails. In stormy conditions even three rectangles are not enough but it would be difficult for an operator to handle more.

It is evident that if Doppler compensation for moving clutter is applied in an area which also contains ground clutter, the latter will not be cancelled. Two cancellation channels are used in the equipment. Within a certain range round the station one channel has no Doppler compensation applied, so that it cancels the fixed but not the moving clutter. The other channel cancels the moving clutter but not the fixed. The two outputs are fed to a coincidence detector which passes the smaller of the two signals, resulting in a display free from both fixed and moving clutter. When both types of clutter actually coincide in slant range and bearing, whichever is the weaker is passed to the display. This system is referred to in the text as "overlapping areas".

5. GENERAL DESCRIPTION OF SYSTEM.
In the following description reference will be made to the Block Schematic Diagram No. RD 14376/D Issue 5.

The IF signals at 13.5 Mcs. are fed to the Signal Processing equipment from two sources: from the Linear Head Amplifier and from the AJ receiver. Both these signals go via pre-set attenuators to the IF Switching Unit IFSU. A switch at the control desk enables the operator to select which of these two signals is fed to the MTI and the Log+PLD channels.

5.1 AJ Channel
The AJ receiver is a Dicke Fix receiver for use in the presence of carcinotron type noise jamming. Its output is fed direct to a linear detector LD and thence via the equalising delay T4 to Channel 1, the user's channel. The linear detector has independent power supplies so that in the event of a major failure of the signal processing racks the user would have an AJ display as a standby.

In the event of heavy jamming, the ANTI-JAM switch on the control console would be operated. This would cause the Log+PLD and the MTI channels to be fed from the AJ receiver instead of from the linear head amplifier. The noise level fed to the two channels by the AJ receiver would be about 50 dB above normal (see Sect. 3.1). This excess noise would be handled by the logarithmic receiver and the noise-operated AGC system.

5.2 Linear Channel.
The signal from the linear head amplifier is at a level of 300μvolts r.m.s. noise with maximum signals 76 dBs above noise. This level is required for satisfactory operation of the logarithmic amplifier but is too high for the other IF ampifiers in the system. The signal is therefore fed via a 15 dB attenuator to the Linear Amplifier and thence via a linear detector LD and delay line T3 to a monitoring console.

The video level supplied to this and other consoles is nominally 1.5 volts peak signal and 0.3 volt shoulder noise, but these levels may be modified if considered advisable.
Swept gain which starts at 200 miles and increases to about 60 dB attenuation in the next 40 miles, may be applied to the linear channel to aid in the identification of a jammer.

5.3 Log+PLD Channel.
The output of the IFSU goes direct to the log. amplifier and thence via the swept gain unit GSG and delay T2 to the PLD unit. This has two outputs, one of which goes direct to Channel 2. The other output feeds three video switches in parallel. VS1 supplies the "background" signal to the Fully Processed display and VS2 that for the Semi-Processed display. The output of VS3 consists of Log+PLD for 200 miles and Linear with Reversed Swept Gain beyond this range. By means of the Lin. Injection switch at the control desk the operator can actuate the relay RY2 and thus change the backgrounds of the fully and semi-processed displays to Linear at 200 miles.

5.4 MTI Channel.
The signal from the IFSU is passed through a 15 dB attenuator to the amplifier GC and then to the IAGC amplifier in the MTI channel. GC is an amplifier the gain which may be controlled by either the Swept Gain Generator GSG or the Noise Detector ND, which operates if noise jamming occurs. The OR gate selects whichever of these two bias voltages is the greater.

The IAGC amplifier input is about 55 microvolts r.m.s. noise and 76 dB maximum signal to noise.

The noise is amplified 62 dB but large signals are compressed so that the maximum output signal to noise ratio is rather more than 20 dB. The output goes to the IF Compressor. This further compresses the output of large signals to either 20, 14 or 8 dB above shoulder noise. These three conditions are selected by means of the Blind Velocity Band Switch on the control desk. The output of the compressor is fed in parallel to the two Coherent Detectors CD.

The Lock Pulse which is derived from the transmitter pulse and the Stalo sets the Reference Oscillator in oscillation so that its starting phase is coherent with the transmitter pulse. The reference oscillator or "coho" is allowed to oscillate for 3.875 milliseconds when an inhibiting pulse stops further oscillation until the next lock pulse arrives. For reasons given in Section 4, the coho is offset about 100 kc/s from the IF frequency in order to produce coherent low IF (CLIF).

The output of the coho is fed to the coherent detectors via the Doppler Compensation Equipment DC. This is quite complex equipment which is described later. For the present it may be considered to be merely splitting the reference frequency into two parallel paths, one to each coherent detector.

The output of each coherent detector is bipolar. With the synchronous reference frequency the output pulses would be either positive or negative according to their phase. With the use of CLIF however, every phase is explored so each signal is both positive and negative. This is of assistance when setting up the equipment by means of artificial signals.

Considering one channel, the output from the coherent detector is rectified by the Video Rectifier VR in order to operate the Noise AGC unit ND. The output from CD is also taken to the first cancellation system which is shown symbolically on the diagram. Delayed and undelayed signals are fed to the subtraction circuit. The output of this is again bipolar and resembles the output of the coherent detector except that permanent echoes have cancelled by about 15 dB.This 2-pulse cancelled output is fed to the second cancellation system, in which further cancellation takes place. Permanent echoes have now been cancelled to noise level.

The output of the second subtraction circuit is converted to unipolar video by a video rectifier VR. A second VR rectifies the 2-pulse cancelled output, and a third rectifies the same output which has been delayed one pulse period. The outputs of the three video rectifiers go to a Coincidence Detector "&". This triple coincidence is the first stage of the anti-impulsive-jamming system, PRFD, described in detail in Section 6.

The other cancellation channel is similar to the one described, except that it is not used to control the Noise AGC . Its outputs end in a similar coincidence detector. The outputs from these two coincidence detectors are then passed to another coincidence detector. This is to provide cancellation in overlapping areas referred to in Sections 2 and 4. In such an area one channel will be cancelling fixed clutter but not moving clutter and vice versa for the other channel. Whenever the outputs of the two channels differ, the coincidence detector passes the smaller of the two signals, resulting in the cancellation of both types of clutter.

It may be wondered what happens to the wanted targets, when so many coincidence detectors are used, all of them passing the smallest of the input signals.

The MTI signals pass through three coincidence detectors: the triple coincidence in each MTI channel, which is the preliminary stage of PRFD, the coincidence between channels just described, which is for cancellation in overlapping areas, and the final coincidence in the PRFD system.

Considering these in order. Two of the inputs to the triple coincidence are from a 2-pulse cancelled system and the third is from a 3-pulse system. RD 23382 Sheet 2 shows the signal-to-noise ratios and velocity response of these two types of signal, compared with that of linear radar. The curves are normalised for noise, that is the video gains of the MTI systems have been reduced so that the noise outputs of all three systems are equal. It is desired that the output of the triple coincidence shall have the characteristics of the 3-pulse system. This is achieved by injecting an artificial moving target of optimum velocity, small enough to avoid limiting, and adjusting the gains of the video rectifiers so that the signal inputs to the coincidence detector are equal. The noise input from the 3-pulse channel will be slightly smaller, 1.2 dB, than from the other two channels. The output of both, signal and noise will therefore be determined by the 3-pulse signal and target visibility will be as good as from a 3-pulse cancellation system alone.

The next coincidence detector, for overlapping area cancellation, does cause some loss of targets as each MTI channel has its own series of blind velocity gaps, and while one series is centred on zero, 56, 112, etc. knots, the other will be centred on V, V+56, V+112, etc. knots, where V is the Doppler compensating velocity applied. There will thus be an increased number of targets lost due to blind velocities, the percentage lost being doubled in the worst case. This can only occur however, over a very small proportion of the screen, and in any case the effect will be minimised by the use of clutter-switching.

The PRFD coincidence detector causes no loss of target visibility whatever. There is in fact a slight gain in signal-to-noise. The only price which must be paid for the benefits of coincidence detectors is some degradation of the quality of noise.

The combined output of the two channels then passes via delay Tl and a PLD unit to VS1 and VS2. The OR gate, subtraction circuit and ESA unit which are shown on the diagram will be ignored for the present as they are part of the PRFD system.

The PLD in series with the MTI serves a dual purpose. Firstly, it prevents saturation of the PPI by CW jamming. Secondly, if an area of rain or window is imperfectly cancelled by the MTI, due for instance to turbulence, the PLD will clean up the screen further.

One of the main tasks of the MTI operator is the adjustment of Doppler compensation for moving clutter by means of the direction and velocity controls. This makes it essential that he be supplied with an MTI display where the imperfections of cancellation have not been masked by the series PLD circuit. The MTI monitor console is therefore fed from the junction of T1 and PLD. As the PLD unit introduces about 12μsecs delay this display will have an error in range of about a mile but this does not matter as it is not an operational display.

The video switches VS1 and VS2 are now each supplied with an MTI signal and a Log+PLD signal. Before considering their outputs we will return to an earlier stage of the MTI channel.

A Timing Pulse derived from the PRF Generator is fed to the first cancellation system of one channel. It is delayed by the mercury delay cell and passed to the Time Error Detector TED, along with the same pulse undelayed. If the delay of the cell is not matched to the PRF, the TED develops an error voltage. This operates the Servo Motor SM and corrects the PRF by changing the frecuency of the master oscillator Q0. The pulse is at -8 microseconds, that is, just before the arrival of the lock pulse. It thus operates undisturbed by signals, as the coherent oscillator is quiescent.

In the first cancellation circuits of both channels, the delayed and undelayed signals, besides being fed to the subtraction circuits, go also to adding circuits "+". The output of the adding circuits is small for optimum velocity targets and large for clutter. The output of each adding circuit goes via a video rectifier VR to a Clutter-Operated Switching Unit, COSU. If the COSU receives a signal of more than 18 microseconds duration i.e. about twice the pulse length, it treats it as clutter and produces an output of about 30 volts which is fed as a switching voltage to VS1. The outputs of the two units are combined in an OR gate so that VS1 receives whichever is the greater signal. The COSU is also sensitive to closely spaced echoes, such as a cloud of angels. They will therefore operate the clutter-switch and can be treated by MI. How this is done will be described in Section 6.

As the COSU only produces an output if the input signal exceeds 18 microseconds, the clutter-switching voltage is delayed by at least this interval. As the switching signal must arrive at the video switch just before the signals which are to be switched delays Tl, T2 and T3 are inserted in the signal channels.

Whichever signal is selected for Channel 1 goes via the PRPD circuit, labelled AIJ on the diagram. This removes impulsive jamming. The exception is the emergency AJ signal, which goes direct, as it is an emergency channel for use in case of an MTI failure.

5.5 Video Switching.
The video switches are so arranged that in the absence of a switching voltage VS1 and VS2 pass the "background" signal. This is normally Log+PLD, but if the Linear Injection switch is operated the background signal is the output of VS3. This consists of Log+PLD for 200 miles and Linear radar beyond this range.

VS1 feeds the Fully Processed display, and is supplied with a switching voltage for the P.E. circle at about 50 miles, generated by the range generator SAN. The radius of the circle may be adjusted to suit the site. Other switching voltages are supplied for the three rectangles by the rectangle generators RG. All these voltages go to the video switch via the area switching distribution unit ASD, This unit combines the switching waveforms and generates the video markers which indicate the position of the circle and rectangles on the PPI.

The clutter-switching voltage from the two COSU's is also fed into the ASD unit and thence to the video switch. A coincidence detector in the ASD unit ensures that the rectangle or circle waveform is passed to the video switch only when the clutter-switching voltage is also present. If desired however, the COSU voltage may be replaced by a DC voltage, thus passing the whole of the circle or rectangle waveform to the video switch and producing "Block MTI".

VS2 feeds the Semi-Processed display. Its switching voltage is from the P.E. circle only, without clutter-switching.

VS3 provides the background signal to the other two video switches, when Linear Injection at 200 miles range is desired. Its switching voltage comes from the appropriate range generator SAN.

5.6 Description of Operator's Facilities.
The MTI operator sits at a suite of three 12 inch consoles. Console 1, on the left hand side, displays Channel 1, which is fed to the user and is the best display available in the prevailing conditions. Channel 2 also goes to the user. This is a standby channel and consists of Log+PLD without any PRFD against impulsive jamming.

Console 2, in the centre, is the MTI monitor console. It is for the use of the operator and a switch enables him to choose any of the following displays:-


(a) Channel 1, namely what is being sent to the user.
(b) A linear radar display limited at a low level to avoid blooming the PPI.
(c) MTI. This displays MTI all over the screen and enables the operator to set up any rectangle he may need for cancelling moving clutter and to optimise the speed and direction conditions in each ractangle. It also assists him to detect any fault in the MTI system. Such a fault might not show up clearly on Channel 1, owing to its composite nature.
(d) PLD. For monitoring the Log+PLD channel.
(e) A/J. For monitoring the output of the Dicke Fix receiver.
(f) Spare. A test position for maintenance purposes.

Console 3 on the right hand side, displays Linear radar in order to show the operator the extent and nature of the clutter and jamming. The signal is subject to reversed swept gain beyond 200 miles in order to give an accurate indication of the direction of a jammer. This is not an operational display and is not passed to the user.

Channel 1.
This may be any one of five displays with different degrees of processing, selected by the operator at the request of the user. They are listed on RD 14376/D and only the first two need description.

(a) Fully Processed.
This is the best picture the operator can produce in given conditions. Sheet 4 shows such a display in which three rectangles are in use. The centre circle defines the area within which Channel A will not be Doppler compensated and so will cancel permanent echoes. Rectangle 3 is shown overlapping part of the circle.
Within this overlapping area only Channel B is compensated with the appropriate velocity for the rain R3. Outside the circle, in area 3 of the rectangle both channels are cancelling rain R3. In rectangle 2 both channels are compensated for "window". It will be seen that the boundary marker of rectangle 3 is obliterated where it overlaps rectangle 2. This indicates the precedence of 2 over 3. That is, in the area common to both rectangles, the Doppler compensation applied to No. 2 prevails. Similarly Rectangle 1 takes precedence over both the others.

Rectangle 1 is shown cancelling "anoprop". This is frequently distant ground clutter made visible by the bending of the transmission path. It can usually be very well cancelled by MTI with no Doppler compensation.

The MTI on the Fully Processed picture is normally clutter-switched, the circle and rectangles merely defining the boundaries of the clutter-switched areas. There are four switches however, which enable clutter-switching to be dispensed with in any of the areas, the result being MTI all over the area.

(b) Semi-Processed.
This is the best picture which can be produced without appreciable manual adjustment by the operator. It has Log+PLD as the background with MTI all over the centre circle.

Adjustment of Rectangles.
The operator can adjust the rectangles to any desired size and position by means of switches and the joystick. The procedure is facilitated by means of corner dots which define each rectangle. These are injected in the intertrace period so that they are always visible during the setting up procedure.

When a rectangle has been set up in the required position, the operator sets the LOCAL/USER switch to LOCAL and proceeds to optimise the cancellation by adjusting the wind-speed and direction controls. When he considers that the best possible result has been obtained, he will set the switch to USER and the effect of the rectangle will only then be seen on the Fully Processed display. This is important as it means that the user's display will not be degraded during the period of trial and error.

6. MORE DETAILED DESCRIPTIONS.

6.1 PRF Generation.
The Type 84 requires for its operation three pulses at a nominal PRF of 250 c/s. and occurring at arbitrary times, "0μsec.", -8μsec. and -125μsec. respectively. The
pulses must be extremely free from jitter and the precise repetition frequency must be accurately controlled by the delay of the mercury cell. Pulses at -250, -500 and -750/μsec are also provided by the equipment for purposes not directly connected with the Type 84.

The most usual method of fixing the PRF automatically from a delay unit is by using some form of "run-round" system. This system is a good one from some points of view, but presents some difficulties, such as jitter arising from noise in the delay cell amplifier and the difficulty of providing jitter-free pulses at various times, without the use of additional and expensive delay cells. The system adopted therefore, uses an automatically frequency-controlled crystal socillator and a frequency divider chain. (See Ref. 4 and British Patents 618967, 663657, 678162).

The crystal oscillator frequency is 4096 kc/s.
(i.e. 214 x 250 c/s.), which is divided by two to feed the main divider chain. This frequency is very suitable for variable frequency crystal oscillators. Also the use of a high initial frequency enables a very jitter-free pulse to be obtained by the methods used.

The divider chain is fed with a nominal frequency of 2048 kc/s. and this is divided in stages of two to 250 c/s. It is not easy to divide from such a frequency using conventional bi-stable circuits. Even if it were, it would be preferable to use some simpler device which does not demand such close tolerances on components and valves.
The circuit used is shown on Sheet 12. Apart from the grid-stopper there are only three components in addition to the valve. This circuit, which is really the ultimate simplification of the divide-by-two "regenerative modulator divider", gives no output unless there is an adequate input. This is obvious if one considers the input to be joined to earth. Then, relative to the centre-tap, any swing of potential on one grid is balanced by an equal and opposite swing on the other. The changes of current balance and there is no feedback into the tuned circuit. This would be upset of course, were the left-hand triode to fail without the right-hand one doing so. For this reason series connected heaters are used.

When the circuit is driven at twice the tuned-circuit frequency it becomes synchronously regenerative. This is best understood by assuming the tuned circuit to be already in oscillation. With no input, this voltage would cut off the left-hand triode when positive and make it conduct when negative. If, at this latter point, a negative half-wave is impressed at the input, it will energise the tuned circuit in a direction to increase the oscillation. The next negative input pulse, being in the wrong direction, is not passed by the valve because it is already cut off. The input and output waveforms shown on the Figure are idealised. In practice there is a good deal of distortion.

The dividers are supplied with reduced HT, (100 volts), from a cathode follower, to save current and prolong valve life.

The waveforms from the dividers do not contain any rapid transitions suitable for the precise determination of time. They are therefore combined by means of AND gates so that the final output is a single half sine wave of the input frequency recurring at the repetition frequency of the last stage. The -8μsecs. and the -125μsec. pulses are produced by phase shifting some of the sine wave components before they are fed to the AND gates. Only three divider outputs need phase-shifting in order to produce each pulse.

The pulses after amplification and shaping are fed to cathode follower stages which provide output signals at 220 ohms impedance for local use and 75 ohms, for remote use. The amplitude of the pulses is not less than 15 volts and the duration 4μsecs. The pulse-to-pulse jitter is very small, being less than one millimicrosecond. It is probable that some of the jitter observed is due to the measuring equipment.

Each cancellation panel has both cancelled and delayed outputs. The 2-pulse cancellation panels also have an "added" output, for supplying the clutter-operated switching units.

The level of the detected carrier from the delayed channel amplifier is +3 volts, modulated about 50%. Besides going to one input of the subtraction circuit, it goes via a DC amplifier to the "delayed" channel output cathode follower. An amplified AGC voltage is therefore obtained from this output valve. It is applied via an RC smoothing circuit to control the gain of the delayed channel amplifier, in order to keep the carrier level substantially constant. Similarly AGC for the undelayed channel is derived from the "cancelled" output valve. Any inequality of the two carriers applied to the subtraction circuit is amplified and causes a change in mean level of the cancelled output, which is corrected by means of the AGC voltage to the undelayed amplifier. The carrier levels applied to the two inputs of the subtraction circuit are thus maintained substantially equal, which is a necessary condition for good cancellation.

Biassed catching diodes ensure that in the event of any component failure the AGC voltages will always be at least 1.7 volts negative, in order to avoid damage to the amplifier valves.

6.3 PRF Discrimination.
As stated in Section 3.3, the method of eliminating non-synchronous impulsive jamming is to pass the signals to a coincidence detector, together with the same signals delayed by one pulse period. A train of pulses from a target is passed with loss of only one pulse. Non-synchronous interfering pulses give no coincidence and are therefore suppressed. As the noise is uncorrelated the noise level is reduced 3 dB. This more than compensates for the loss of one signal pulse, if there are more than two pulses per target. There is in fact a small net gain in signal-to-noise ratio.

The crystal oscillator at 4096 kc/s is based on the "FMQ" principle4,5 which enables the frequency to be varied over a band of about 1% of the mean frequency. The frequency is controlled by the delay of the mercury cell. As the temperature coefficient of the cell is +2.6 in 104/°C., one crystal may be used over the range of temperature likely to be encountered at a given station.

6.2 Cancellation System.
Each MTI channel consists of two cancellation circuits in cascade. Four mercury delay channels are therefore needed. Economy in delay cells is achieved by means of frequency duplexing. The first cancellation is effected by modulating an 8 Mcs. carrier with the output of the coherent detector. After passing through the delay cell, it is amplified and detected and fed to one input of the subtraction circuit. The undelayed signal is attenuated and fed to the subtraction circuit via a similar amplifier and detector.

The cancelled output from the subtraction circuit is used to modulate a 6 Mcs. carrier in a second delay cell driver. This modulated carrier then passes through the same delay cell channel. Hybrid circuits are used at the input and output of the delay channel in order to combine and re-separate the signals. The use of only one mercury delay path for the two cancellations ensures that the delays are equal.

A -8μsecond timing pulse derived from the PRF generator is fed to the 8 Mcs. delay cell driver of one MTI channel. The delayed and undelayed pulses go to a Time Error Detector. If the delay is not matched to the PRF an error voltage is developed which changes the frequency of the master oscillator and corrects the PRF.

A similar process takes place in the other channel except that no timing pulse is used. Two mercury delay paths are therefore required for the complete cancellation system. These are contained, one above the other, in a single case, so that their delays are subjected to very similar temperature effects. Each cancellation channel contains a short delay, switchable in 0.1μsec. steps, to compensate for small differences in cell or circuit delays.

A jamming pulse on either diode of the coincidence detector allows the noise on the other diode to pass at full strength. The jamming pulse is therefore visible on the output as a 3 dB increase of noise. As this is not desirable it has been minimised as follows. Referring to RD 14376/D, the main PRFD circuit follows the relay RY4 and is labelled "AIJ". Signals both delayed and undelayed go via an OR gate to an output of a subtraction circuit, the other input of which is fed from the coincidence detector. Jamming pulses are passed by the OR gate only, so the subtraction circuit receives only one input. It therefore produces an output which operates the electronically switched attenuator ESA and reduces the output 3 dB. The noise which has been increased by the jamming pulse is therefore restored to normal. Target pulses are fed to both inputs of the subtraction circuit, so the ESA is not operated and no attenuation takes place.

The PRFD system needs a 4 millisecond delay line, a delay line driver and amplifier, much the same as in a cancellation system. The delay must be matched to the PRF, but the accuracy demanded is not so great as for MTI. With a 10μsecond pulse, an error of one μsecond would only degrade the signal-to-noise ratio by about ½ dB. As the PRF is already controlled by the MTI delay cell, a variable delay cell is used for PRFD. The output of a time error detector controls a motor which varies the delay path of the cell.

The system just described is not enough in itself to eliminate jamming on the MTI channel. In a 3-pulse MTI system a single jamming pulse is converted into three pulses. These must be reduced to a single pulse again before going to the main PRFD system. This is done by feeding three cancelled outputs to a coincidence detector. The three input signals are; 3-pulse, 2-pulse and 2-pulse, delayed but without further cancellation. The process is illustrated on Sheet 5, it will be seen that a single pulse of interference emerges from the triple coincidence as a single pulse, which is then eliminated by the delay and coincidence detector of the normal PRFD circuit.

The effect of MTI and PRFD on a train of pulses from a target is shown on Sheet 6. Seven input pulses are shown for reference. The output of the triple coincidence detector still contains seven pulses but the "time centre" of the paint has been delayed one pulse period. This is equivalent to about 6' of arc. The final coincidence in the PRFD circuit shifts the centre of the train a further 3' of arc. The fully processed signals are therefore slightly displaced in azimuth, the background by 3' and the MTI by 9' of arc.

The OR gate, subtraction circuit and ESA used in the PRFD system to prevent an increase of noise by jamming pulses, are used also in conjunction with the MTI coincidence detector, for the same purpose.

6.4 Clutter-Operated Switching Unit.

6.4.1 Clutter Switching.
Sheet 7 shows the principle of operation. The bipolar output of the addition circuit is fed to the COSU, where it is rectified and limited. The resulting signal is shown at (a), which shows strong clutter and angels along with noise. The baseline is clipped about 6 dBs, above shoulder noise to remove all but the highest noise peaks. The resulting signal, shown at (b), is fed to a coincidence detector together with the same signal, shown at (c), delayed 18μseconds. The output of the coincidence detector is as (d). It starts 18μseconds later than the input signal and is shorter in duration by the same amount. This output signal is amplified and stretched and is fed as the switching voltage to the video switch VS1. Stretching is necessary in order that the duration of the switching voltage shall be a little greater than the signal which is to be switched. The amount of stretching is about 30μseconds. As the output from the COSU is delayed approximately 20μseconds, the signals which are to be switched are also delayed so that they arrive at the video switch just after the switching voltage.

6.4.2 Angel Switching.
The angels shown on Sheet 7 produce no output from the COSU as just described, as there is no coincidence, owing to their short duration. It is desirable that the fully-processed display should be switched to MTI by angels if they are dense, as even imperfect Doppler compensation is preferable to Log+PLD, which passes them at full strength. One solution is to apply MTI all over a rectangle which encloses the angels. This results in blind velocity losses, however. It is also preferable that the decision as to the density and nuisance value of the angels should be made automatically and not be left to the operator.

It has been found that if the density of the angels corresponds to two miles random spacing, or less, they are sufficiently troublesome to warrant treatment by MTI. Random spacing of two miles corresponds to about eight miles radial separation. The angel switch has therefore been designed to operate if point echoes are separated by less than 100μsecs.

Referring to Sheet 8, four angels are shown at (a) after limiting and clipping, and again at (b) having been delayed 18μsecs. The angel identification device is essentially a pulse-stretching circuit. It stretches the pulses in the delayed channel so that they still have appreciable amplitude after 100μsecs. This is indicated by the dotted line. It will be seen that the residual amplitude of each stretched pulse provides a voltage coincident with the succeeding undelayed pulse. The train of four angels therefore produces three switching pulses. These are amplified and stretched and fed to the video switch. The first angel of the group is not switched to MTI. This is the desired result, as it is important that an isolated target in the clear should not be switched over to MTI.

It is evident that if closely spaced angels are switched to MTI so also will a close formation of aircraft. This would be undesirable in some cases; hence the provision of a control switch at the console. This enables angel-switching to be dispensed with, while retaining normal clutter-switching.

6.5 Doppler Compensation System.
As there is a good deal of confusion as to the operation of Doppler compensation for moving clutter, some clarification will be attempted.

The essential condition for cancellation is that the output from the phase detector shall be constant from pulse to pulse. This requires that the two inputs, the echo signals and the reference signal, shall be of constant amplitude and have a constant phase difference. Referring to Sheet 10, Fig. 1, let the vector OT represent the starting phase of the transmitter pulse and OE the phase of the echo pulse from a fixed object. As the echo time t is constant, the phase angle between the vectors will be constant. Let ORs be the starting phase of the coherent oscillator. The phase angle between OT and ORs will, ideally, be constant. If ORe is the phase of the reference signal after time t, it does not matter what this phase is as long as it remains the same, producing a constant phase angle θ. The coherent oscillator need not therefore be of any specific frequency if the oscillator behaves similarly from pulse to pulse. It is this fact which makes possible the use of coherent low IF.

Considering an element of moving clutter, such as "window". To simplify the argument, it will be assumed that the aerial is stationary, directed at the clutter, which is approaching at about 10 knots.

Assume that the first blind velocity is 56 knots, corresponding to a Doppler frequency fd of 250 c/s. and phase change of 2π radians from pulse to pulse. A radial speed of 10 knots will therefore produce a Doppler frequency of 45 c/s. and a phase change between pulses of 64°. In Fig. 2 OT is again the phase of the transmitter. If OE1 represents the phase of, the first echo pulse from the moving clutter, OE2 and OE3 will be advanced by 64° and 128° respectively. For cancellation the reference frequency vectors must be made to behave in the same manner. This is done by adding 45 c/s. to the coho frequency by any convenient means. When the lock pulse starts the coho the frequency changing system produces a new reference frequency. The starting phase of this is represented by ORs1 and will depend on the lock pulse and on the fortuitous phase change introduced by the frequency changing circuits. Let the vector ORe1, at angle θ to OE1, represent its phase after echo time t. Succeeding lock pulses will continue to start the coho at a fixed phase relation to OT, but the starting phases from the frequency changer will each advance by 64°, owing to the addition of a continuously rotating vector to the coho vector. This is shown by ORs2 and ORs3. Assuming constant behaviour from pulse to pulse, the vectors ORe2 and ORe3 will also advance 64°, so the phase angle presented to the phase detector will always be θ, and the moving clutter will be cancelled. It is seen therefore, that Doppler compensation is achieved by altering the starting phase of the reference frequency from pulse to pulse, rather than by altering the number of cycles during the echo time.

If the moving clutter surrounds the station its radial velocity will vary as the aerial revolves, between plus and minus its true velocity, according to a cosine law. The applied compensation must therefore follow the same law and be adjustable in magnitude and direction. In the Type 84 the velocity control is calibrated from zero to 100 knots. This entails Doppler compensating frequencies up to nearly twice the PRF. If the compensating frequency required is ±fd the actual frequency added to the coho is 4(PRF) ±fd. The datum shift of 4(PRF) is used for convenience and does not affect the compensation as the reference frequency vector is merely moved 4 complete revolutions, thus introducing no effective phase change.

The frequency to be added to the coho frequency varies therefore from about 500 to 1500 c/s. As the output frequency is so near to that of the coho, the frequency change is done in two stages, as shown on RD 10189/B. In this diagram the coho frequency is shown as 13.5 Mcs. In the Type 84 this would be 13.4 Mcs. owing to the use of CLIF, but as stated earlier in this section, neither the PE cancellation nor the Doppler compensation is affected by this detuning. On the diagram it is assumed that 4(PRF) is 1000 c/s. The Doppler compensating frequency required for a given velocity of moving clutter depends on both the PRF and on the wavelength of the transmitter, as given by the expression in Section 4.4. Both these parameters may vary from site to site, but the Doppler compensation circuits may be adjusted empirically to take account of this. It will be seen that two crystal oscillators are used. The first is variable by ±fd, the compensating frequency, and is subtracted from the coho. The second oscillator, of fixed frequency, is then added to the output of the first mixer, to give the desired result.

The use of two RF oscillators in this manner solves the problem of filtering out unwanted sidebands and of providing the necessary frequency modulation, but it creates a problem of frequency stability. In order to achieve a compensation accuracy of ± 1 knot the difference between the two oscillators must be stable to ± 4.4 c/s. To maintain the absolute frequencies of the oscillators to such a degree would be difficult, even with temperature control. Fortunately only the difference frequency is important and this is stabilised as follows. The difference frequency is detected, amplified and limited and then fed to a frequency discriminator, the output of which acts as an error voltage. The system is adjusted so that when no Doppler compensation is desired, i.e., Vd is zero, the error voltage Ve is zero and the difference frequency is 4(PRF). As this corresponds to the fourth blind speed it does not introduce any velocity compensation and permanent echoes are still being cancelled. The difference frequency of 4(PRF) is chosen as a convenient mid-frequency for the discriminator, which thus operates between 500 and 1500 c/s. approximately. If one oscillator tends to drift in frequency relative to the other the error voltage Ve is applied to the reactance valve which changes the frequency of the variable oscillator. The feedback chain includes a high gain amplifier so that the difference frequency is held substantially free from undesired variations. When wind speed and azimuth information is fed to the balanced phase detector the deviation voltage ±Vd, subject to a high degree of negative feedback, frequency modulates the variable oscillator.

Fig. RD 14377 shows an embodiment of the system which employs three rectangles and provides cancellation of both fixed and moving clutter when a rectangle overlaps the centre circle. Each synchronous rectifier is supplied with two voltages; a 500 cycle sine wave and a sine wave which alternates in phase at aerial speed. The direction and speed controls for this are at the Console Control Panel, as are the controls for positioning each rectangle. The output of each synchronous rectifier is a DC voltage varying symmetrically about zero at aerial speed, (±Vd on RD 10189/B.).

A single fixed crystal oscillator is common to the three circuits, each of which has its own variable crystal oscillator. The frequency of these is varied by the now well known FMQ technique.4,5 As the deviation required does not exceed ±1 in 104 good linearity is not difficult to achieve.

If any of the three electronic switches is operated by a switching voltage from the corresponding rectangle generator the frequency variation fd is introduced into both channels. If more than one rectangle switch operates at the same time - that is, if rectangles overlap, 1 will, take precedence over 2 and 2 over 3. For a limited range, say 50 miles, the PE ring operates the electronic switch which supplies one channel. This switch overrides the effect of any of the rectangle switches and thus ensures that this channel has no Doppler compensation applied to it within the PE ring. This provides the facility for cancelling both fixed and moving clutter within this area.

6.6 Pulse Length Discriminator. (PLD).
The PLD passes 10μsec. pulses at full strength but suppresses clutter of more than 20μsecs. duration. The principle of operation is shown on Sheet 9.

The unit is fed with video signals from the logarithmic receiver. A signal consisting of a 10μsec. echo pulse and a block of clutter is shown at (a). This is applied to a 5μsec. shorted delay line with the result shown at (b). The signal at (b) is delayed 10μsecs. as shown at (c). It is also inverted with no delay, as shown at (d). The signals at (c) and (d) are fed to a coincidence detector, with the result that the 10μsec. pulse gives an output but the block of clutter does not, owing to the lack of coincidence. It will be seen that the output pulse is delayed one pulse length compared with the input.

The PLD process produces a rather spiky type of noise, tending to give a high false alarm rate. This is mitigated to some extent by the use of a low pass filter after the coincidence detector. This reduces the high peaks of noise and improves target visibility.

The output amplifier of the PLD unit employs negative feedback to give approximately unity gain. A simple signal intensifying circuit is incorporated in the feedback loop, in order to compensate for the loss of signal-to-noise ratio due to the logarithmic receiver. This starts to operate at approximately 6 dB above shoulder noise.

6.7 Generation of the Swept Gain Waveform.
The main object of applying swept gain is the reduction of the strength of echoes from angels. As these are discrete echoes they are passed at full strength by Log+PLD, and owing to their random velocities they cannot always be usefully reduced by Doppler compensated MTI.

Consideration of the Aerial VPD, the size of target to be detected, and the probable angel returns, led to a provisional swept gain law as follows:- 4th power law for the first 40 miles; a constant 15 dB attenuation for the next 40 miles, and a gradual tapering off of the attenuation to 120 miles. This obviously requires a complex waveform, as shown on Sheet 11. The method of generation is as follows, reference being made to E/RAP/D.47814.

Pulses of 20 volts nominal amplitude are fed in at SKM.

V1 amplifies and inverts them, the negative excursion being limited by the silicon diode V18 and the potential divider R5 and R6.

While the point SKD is held negative the diodes in V2 conduct and cause condensers 05 and C6 to discharge, rendering both anodes of V4 32 volts negative with respect to earth.

When the input pulse ends, after about 5μseconds, V2a and V2b cut off, leaving the potential across C5 to aim for HT+ via R9, R11 and that on C6 to aim for earth via R12 and V3b. When this potential reaches -2.5 volts, which occurs after 500μseconds, corresponding to 40 miles, V4 conducts and holds it at this value due to the potential divider V16, R23 and R24.

In the meantime the voltage on C5 continues to rise until after a millisecond the other half of V4 conducts and permits the rising voltage on C5 to pass to the grid of V5a. V5a and V5b comprise a high gain amplifier causing the grid of V6a to rise sharply upwards at this instant.

As V6a is a cathode follower the output voltage at SKF rises, sharply to a positive value with respect to earth. V3 and V16 are cut off and since the voltage jump at SKF is also felt at the junction of R23 and R24, the bottom half of V4 is also cut off.

The voltage on C6 is now no longer held steady and it resumes its upward rise, the aiming potential now being the positive potential on SKF. The charging route is via R12 and R91 into condenser C9, which is large so that it does not charge up appreciably before the advent of the next input pulse.

It is seen therefore that the voltage on C6 has the desired characteristic, namely, an exponential rise, a flat portion and a further exponential rise. C6 is connected to the grid of the cathode follower V6b, thus providing a working output.

6.8 Application of the Swept Gain Waveform.

6.8.1 In the Log+PLD Channel.
As mentioned in Section 2.2 swept gain cannot be applied to the IF fed to the log. amplifier as the attenuation of the IF would be accompanied by greater sensitivity of the amplifier, so the desired result would not be achieved. The swept gain waveform is therefore used as a baseline clipping voltage operating on the video output of the log. amplifier. This has been done by a method outlined below.

It was pointed out by R. Benjamin,6 that when using the differentiated output of a log. receiver, it is inadvisable to employ the normal amount of swept gain in the presence of interference from clutter or noise. This is because the interference itself causes a loss of signal-to-noise ratio, even though the PPI Screen remains clear. The use of swept gain as well, therefore, leads to an unnecessary loss of targets. The solution is to arrange that sensitivity is lost either by interference or by swept gain, whichever is the greater, but not by both at once. This principle is embodied in the Type 84 as follows, still referring to E/RAD/D.47814.

The DC output of the logarithmic receiver is negative and its amplitude depends on the amplitude of the input noise. Any noise interference increases the negative DC component of the output therefore. The noise fluctuations, however, remain substantially constant. This signal is connected to SKN and is inverted and amplified by a DC amplifier with the gain limited to about 5 by negative feedback. The positive-going signal -at SKH- is then split into two paths, R48 and C12 extracting the DC component and C11 passing the AC component, of constant noise amplitude, to the grid of V10a, and thence to the output at SKL,. The negative swept gain waveform and the positive DC component of the signal are combined by R49 and R50 and the resultant, if negative, is passed by V14 and V15 to the cathode follower V9 and thence via R60 to the grid of V10a where it operates as a baseline-clipping voltage on the AC component of the signal.
The degree of clipping may be selected by relays operated from the control desk, in four positions varying from zero to maximum.

It is seen therefore that if the DC component of the signal increases owing to noise jamming or clutter, the applied clipping voltage is wholly or partially cancelled. The AC output of the log. receiver is thus modified according to the principle stated in Ref. 6. This modified signal is coupled from the output SKP to the input of the PLD unit.

6.8.2 In the MTI Channel,
The swept gain law desired is the same as for the Log+PLD channel and the waveform is generated on the same principle. In this case, however, it is used to vary the gain of an IF amplifier which precedes the IAGC amplifier, The same amplifier is also controlled by an AGC voltage which depends on the incoming noise level and prevents saturation of the screen and the MTI circuits by noise interference. The two controlling voltages, either noise AGC or swept gain waveform, are passed to the variable gain valve via an OR gate, so that only the larger of the two is effective. The switch on the control desk which selects the degree of swept gain in the log. channel operates similarly on the MTI channel.

The above processed take place in the Limiter Electrical, Noise, Ml. The relevant units on the block diagram RD 14376/D are ND, GSG, OR and GC.

7. TEST EQUIPMENT.

7.1 Test Pulse Generator.
This provides a lock pulse, a block of artificial clutter and an artificial moving target, all at 13.5 Mcs. The clutter is coherent with the lock pulse as it is derived from the same crystal oscillator. It is used for testing and optimising the cancellation circuits and for checking the coho stability. The clutter signal extends from zero to about one millisecond. If the cancellation is good over this range the clutter block may be switched so that it starts at 1 msec. and continues to the end of the trace. This enables the coho stability at long range to be tested. The leading edge of the clutter may also be used for testing the operation of the clutter-switching.

The moving target is a 10μsec. pulse adjustable in range from 50 to 1500μsecs. It can therefore be either in the clutter or not, as desired. The moving target is also coherent with the look pulse, but the phase of the oscillations is reversed on alternate traces by a multivibrator operating at half the PRF. The pulse therefore simulates a moving target at optimum speed.

7.2 Noise Generator.
This is provided so that the gains of the various units may be standardised against a known noise output. The output of the generator is high enough to be easily measurable. An attenuator is used to reduce it to 300μvolts r.m.s. which is the noise level required by the logarithmic receiver. Having standardised the gains of the system the noise level from the head amplifier may be adjusted to the correct level.

7.3 Artificial Permanent Echo Generator.
In order to test the performance of the complete system, including the transmitter, stalo and lock pulse, a stable permanent echo at long range is required. As no such real permanent echo can be used, an artificial echo is generated as follows.

A standard Type 84 Stalo is fed into a mixer along with the incoming pulse from the transmitter, reducing them to 20 Mcs. These IF pulses are delayed about 2 milliseconds in a quartz delay cell. After amplification the delayed pulses are restored to transmitter frequency by the Stalo. The resulting RF pulses are then amplified and re-transmitted. The received signal thus appears as a permanent echo at an apparent range of about 160 miles more than the true distance of the test gear from the station.
It is proposed to use this equipment at a short distance from the station, with the aerial locked and beaned upon it. This will eliminate the scanning effect and the results will therefore be a measure of the stability of the overall system.

Author...A. Miller (signed)
Approved by... G.N. Coop (signed)

REFERENCES.

  1. Emslie, A.G. and McConnel, R.A.
    "Moving Target Indication" M.I.T. Radiation Laboratory Series Vol. 1, Chap. 16.
  2. Tanter, Herve.
    "Radio Receivers with Elimination of Fixed Echoes" Electrical Communications. Dec. 1954 p. 235
  3. Steward, K.W.F.
    "Anti-Jamming Facilities of the Type 84". Report RD/G.1505 M.W.T. Co. Ltd. Gt. Baddow.
  4. Mortley, W.S.
    "FMQ". Wireless World, 1951, 57, p. 399.
  5. M.W.T. Co. and Mortley,
    W.S. British Patent No.622140 British Patent No.712349
  6. Benjamin, R.
    "The Use of Swept Gain with Logarithmic Radar Receivers". A.S.R.E. Technical Note R13 - 57 - 2.

Distribution:-

  1. Director of Research.
  2. Mr. M. Birchall.
  3.         "
  4. Mr. L.E. Banks.
  5.         "
  6.         "
  7.         "
  8. Dr. D.H. Davies.
  9. Dr. G.N. Coop.
  10. Mr. C.P. Cooper.
  11. Mr. O.E. Keall.
  12. Mr. P.A. Dutton.
  13. Mr. H.N.C. Ellis-Robinson.
  14. Mr. K.G. Stoker.
  15. Mr. W.S. Mortley.
  16. Mr. A.D. Slocombe.
  17. Mr. W.C. Hall.
  18. Mr. E. Greenhalgh.
  19. Mr. J.D. Bell.
  20. Mr. M. Browning.
  21. Mr. J.D. Crispin.
  22. F/O P.C.F. Healey.
  23. Mr. J. Rogers.
  24. Mr, K.W.F. Steward.
  25. Mr. A. Miller.
  26. Room 21 File.

 



Page last updated on the 21st February 2018 by Colin Hinson.