This document is dedicated to the engineers at Collins who brought the

technology of analog vacuum tube receivers to its peak, so that the extent of

their work may be more widely known.

 

Part 1: B+ Filter

​

The purpose of this investigation was twofold. The primary purpose was the reduction of receiver temperature caused by location of the type 6082 tubes. The secondary purpose was that of reducing the cost of the B+ filter. The problem was attacked in two ways. One way was to substitute a single 6336 tube or the two 6082s. Since this tube took less space than the other two, it might have been possible to place it on the IF unit. Also, the higher mu of the 6336 made the elimination of one voltage reference tube possible. However this circuit still had about the same power dissipation as the two 6082 tubes. The second method of attack was to build an L-C type of filter and compare its performance with that of the electronically regulated supply. Power saving amounted to about 40 watts and in conjunction with electrolytic capacitors, the L-C filters afforded a considerable saving. Further testing showed that it was necessary to add a VR tube to get the required stability from the VFO and crystal oscillators.

 

Part 2: Elimination of Certain Functions

​

As a result of the conference of September 10, 1953, it was decided to investigate the savings resulting from elimination of the squelch facility and remote control connector J-105. As a result of this investigation, the remote control connector was removed and the squelch facility was made optional.

​

Part 3: IF Amplifier

​

Work had previously been started on mechanical filter designs on Contract DA36-039-sc-56636. This contract called for development of 455 kc filters, having bandwidths of 2 kc, 4 kc, 8 kc and 12 kc. Early in the cost reduction program, work was started to incorporate these filters into an IF strip. Tests on early models clearly showed the improvement in selectivity that the mechanical filters offered, but it was evident that further work was necessary to reduce spurious responses and feedthru around the filters. Blocking on strong off-tune signals also needed improvement. Later models showed substantial improvement in all these respects. Experiments were conducted to

determine the performance of the mechanical filter IF for DF work. These tests showed that the mechanical filter IF was unsuitable for this application. As the receiver is tuned across the passband the apparent direction of the signal varies up to 10° or more.

 

Part 4: Power Line Filter

​

Considerable time was spent investigating the efficiency of the line filter. The filter proved to be desirable in two respects - reducing receiver oscillator radiation into the power line and reduction of signal input to the receiver via the power line. Simpler filters were investigated leading to a design similar to that formerly in use but having fewer inductors. Tests revealed performances nearly as good at low frequencies and considerable improvement over the old filter at high frequencies. For use in the cost

reduction receiver, the filter was repackaged, eliminating the 164-1connector. The line cord now fastens to the filter by means of screw terminals protected by a cover. The power plug was changed at Signal Corps request to a U126/U type.

 

Part 5: Tuning Control Stops

​

As a result of continuous difficulty with the tuning stops on production R-390 and R-391 receivers, this work was accelerated, resulting in a stop of improved design which was immediately incorporated into production receivers, and which will be used virtually unchanged in the cost reduction receivers.

 

Part 6: Crystal Oscillator

​

The frequency scheme of the receiver was carefully studied to determine if some of the crystals could be removed. It was found that if crystals up to 17 mc could be used it would be possible to substitute one crystal for the five in the first crystal oscillator, and reduce the number of crystals in the second oscillator to 15. This represents a saving of 7 crystals, besides the elimination of two 32 position switches. With the new scheme one 4th order spurious is introduced, along with the possibility of 1st and 2nd crystal

oscillators beating together. This new frequency scheme permitted a greatly simplified mechanical construction, since the fixed first crystal oscillator was removed to the RF chassis.

 

Part 7: RF Coils

​

Experimentation was started on the 16-32 mc band RF coil to improve its tracking. A considerable improvement resulted from changing the length to diameter ratio and using a smaller diameter coil. As soon as the design was complete steps were taken to put this coil in production receivers. Later work showed that similar results could be obtained using the larger coil diameter with a ferrite core to replace the powdered iron slug. All the RF coils have since been redesigned to use the same ferrite core. These new coils also have a higher Q than those in the current R-390. The variable IF coils have also

been redesigned to use a common core. This core however differs from that used in RF coils, because a lower permeability higher Q core was needed. Coil frames have been redesigned for simpler assembly, and teflon terminals are used to maintain high circuit Q. Antenna links have also been redesigned to attain a more uniform antenna input impedance.

 

Part 8: Autotune Motor

​

An AC Autotune motor would be a real convenience on the R-391 receiver as well as saving the cost of the separate 28 volt DC power supply (such as the PP-629) required to run the Autotunes of the current production R-391 receivers. A suitable motor was obtained and tested in a model. Operation appears satisfactory although setup time has been increased to a maximum of 25 seconds.

 

Part 9: AF Filters

​

An investigation was made into the savings resulting from elimination of the 3500 cps low pass AF filter. Upon presentation of the cost information to the Signal Corps, authorization was granted for its removal on the final models. A twin-T feedback amplifier circuit was tested as a possible substitute for the 800 cps filter, but its performance was inferior to the filter, being narrower on the nose and wider on the skirts. By redesign of the one general 800 CPS filter for use at high impedance, it was possible to eliminate one interstage audio transformer while still retaining the filter's superior band-pass characteristics.

 

Part 10: DC Operation of Equipment

​

Dropping the requirement for 28 volt DC operation results in savings due to elimination of the requirement for the power receptacle J-104, and the 28 volt line fuse. It also permitted the use of a parallel tube filament line and 6 volt dial lamps.

 

Part 11: Main Frame

​

The cost analysis made on the main frame showed that the laminated type of construction was quite expensive. A new frame was built up using solid plates, similar to the R-392. This frame with the units mounted in place was tested at Evans Signal Lab with the cooperation of SCEL engineers. It was found to be generally quite satisfactory but it was apparent that the junction of floor plate and side brackets needed strengthening. Since then reinforcing strips have been added and no further trouble is expected. The new frame appears very much stronger and more rigid than the old.

 

 

Part 12: Hum Balance Control

​

Substitution of the L-C filter for the electronically regulated B+ supply eliminated the need for this control.

​

Part 13: VFO End Point Adjustment

​

By redesigning the output transformer can of the VFO and changing the VFO chassis slightly, the VFO end point adjustment was made more accessible. In the R-390A, holes were provided in the gear plate so that this adjustment could be made without removing the VFO or RF units. In the R-391A, the KC Autotune head gets in the way making necessary the removal of the VFO for end point adjustment.

 

Part 14: Gear Train

​

The gear train represents the most complicated mechanism in the receiver and much time was spent in an effort to simplify it. This was made possible to a large extent by the change in frequency scheme which eliminated the necessity for the movable high VIF table and the step cam, substituting in its place a continuously variable high VIF, driven at the same rate as the 8-16 mc slug rack. Placement of these coils behind the 8-16 mc rack, and the 3-2 VIF behind the 1-2 mc RF coils eliminated a considerable number of gears. In addition, by coupling both kc counter and kc drive to one gear on the kc shaft, several other gears were eliminated. Other rearrangements were made where possible to simplify and improve the performance of the gear train. Slug racks have individual slug holders to permit better slug centering in the coils.

 

Part 15: RF Unit

​

Tests were made to determine how much the 2nd RF stage with its coils was contributing to the performance of the receiver. Its advantages were that it afforded additional AGC control ahead of the mixers, and better image rejection above 8 mc. But because of the considerable saving resulting from the removal of this stage with its associated coils, SCEL representatives agreed it was worthwhile. In an attempt to regain some of the lost AGC control, the mixers were put on the AGC line. Attempts were made to restore some of the lost image rejection by improving coil Q and tracking, and by using better switch materials. Better antenna trimmer action on the 16-32 mc band was accomplished by mounting the trimmer closer to the high band antenna coil. This reduces stray C allowing the trimmer to be coupled tighter to the coil.

 

Field tests by SCEL engineers on current R-390 receivers revealed that cross modulation is a problem in the presence of strong interfering signals. Extensive tests were made on all available miniature RF pentodes to find a tube which best combined the characteristics of low noise and low cross modulation. Late in the investigation tube types 6DC6 and 6BZ6 became available. These had the most desirable characteristics of any tube tested. Because the 6DC6 affords slightly better AGC control, it was the one selected for use as the RF amplifier. Cross modulation tests were also run on various

mixers, and the 6C4 proved to be very good in this respect, actually improving slightly in performance as the bias was increased up to about 14 volts. On this basis, it was deemed safe to use AGC on the mixers. Late tests show that cross modulation is largely confined to the first IF tube at medium high levels (.05-.2 volts input) and the RF amplifier at levels above this. Investigation carried out by E. Read on the R-392 receiver

showed a substantial improvement in cross modulation at signal levels exceeding the normal operating bias of the 1st RF tube when a short-time constant circuit was placed in the 1st RF grid. Subsequent experimentation on the "A" model R-391 showed a small improvement particularly at antenna levels exceeding 2 volts. The results are not nearly as spectacular as on the R-392, however.

 

Part 16: Calibrator

​

A calibrator using a 200 kc crystal was built and tested. In comparison to the former model using a 1 mc crystal, reliability was improved and harmonic output increased. The divider control was found to be unnecessary and was eliminated. Some difficulty has been experienced in obtaining a suitable 85°C crystal but a solution appears in sight.

 

Part 17: VFO Stability

​

During the time of this contract, much trouble started to develop in production VFO's due to slight jumps in oscillator frequency. This was finally tracked down as being a result of driving the oscillator too hard. Reduction of screen voltage on the oscillator tube gave a great improvement but also reduced the oscillator output voltage. In cooperation with the oscillator department a lower loss output transformer was designed to restore the voltage to its original value. This change was immediately incorporated into the production receivers. In order to make the end point adjustment more accessible, the mounting can was made smaller on the cost reduction receiver. A few minor circuit changes were also required to adapt the VFO to the new B+ supply.

 

Part 18: Tube Types

​

In an effort to improve reliability all tube types were examined and where possible replaced by more rugged types. Thus the 6BJ6 tubes were replaced by type 5749/6BA6W. 12AU7 and 12AT7 tubes were replaced by the 5814A and 6AJ5's were replaced by type 5654/6AK5W. Because of the high cost of superior replacements, the 6C4 and 6AK6 tubes were retained.

 

Part 19: Antenna Relay

​

The poor attenuation characteristics of the antenna relay at high frequencies led to a search for a better unit. At first, it was hoped that an AC relay coil could be used. These relays were used on the "B" model and the two "A" model R-390 receivers but were not too satisfactory. Later, a DC operated relay was developed and installed in the "A" model R-391. A special selenium rectifier supplies the required DC power.

 

Part 20: Mechanical and Electrical Components and Construction

​

Besides the changes made in components for cost reduction purposes, a number of other changes were made on the basis of greater reliability or convenience rather than cost. Among these are included the smaller coax cable and fittings, substitution of a selenium rectifier for the copper oxide type, and the use of AC relays for the Autotune and break-in function. Changes in mechanical construction were also made, such as the closer attention paid to electrolytic corrosion, the new means for chassis mtg. and the radical power supply unit redesign. Finally a number of components were removed from the receiver entirely, such as the tube pullers and tube pin straightener.

​​

​

​

DESIGN AND EXPERIMENTATION 

​

With the completion of the planning phase, experimental work was begun. For the purpose of describing this work, the order of Phase A is followed more or less closely to show how the original plan was carried out.

 

Part 1: B+ Filter

​

The main problem involving the B+ filter was the excessive amount of heat generated by the series regulator tubes. This is aggravated by the location of these tubes on the bottom deck where it is difficult to get good convection cooling. Two methods of approach were tried for this problem. One method sought to reduce the cubic volume required by the electronic regulator so that it could possibly be mounted on the IF chassis. A single high power regulator, the Chatham 6336 was tested for this purpose. The slightly higher mu of this tube permitted use of a single OA2 instead of the two 5651 VR tubes. Although there was a considerable saving in space, and some in cost, power dissipation was about the same as the old circuit.

 

The second method was to substitute an L-C filter plus a VR tube for the electronic regulator. This filter was designed for choke input which has the advantage that current flow in each rectifier is maintained over a complete half cycle. This eliminates the high peak current impulses found in capacitor input circuits, resulting in lower rectifier plate dissipation and improved tube life. Another advantage is that B+ voltage variation is less with change of load. This is important when the receiver is switched to various functions, particularly to "standby" where the B+ load is quite light. Too great a change

in B+ voltage here would have an undesirable effect on oscillator stability. In order to conserve space, this design requires the use of electrolytic capacitors. Because of low temperature requirements, the highest possible capacitance was used consistent with space and working voltage requirements. These capacitors are easily replaceable plug-in units. Main filter capacitors are 45 uf and AF decoupling capacitors are 30 uf each. Use of such size decoupling capacitors in the audio circuits along with a few wiring changes,

virtually eliminates the remaining cross talk between line and local channels. The VR tube supplies current only to the screens of the VFO and crystal oscillator. Because of the wide line voltage tolerances, it was necessary to cut the regulated current as much as possible in order not to exceed the VR tube current rating and still maintain adequate starting voltage. Careful tests revealed that regulating the plate voltage of the oscillator had virtually no effect on oscillator stability. The BFO had been carefully designed for extremely low B+ voltage coefficients, and it was possible to operate it completely off the unregulated supply. With all these changes the receiver stability vs. line voltage has hardly changed over the voltage range 105-125. In contrast to the electronic regulator, the L-C filter permits receiver operation down to line voltages as low as 80 volts. This could prove to be a distinct advantage under some conditions. The power saving resulting from the L-C filter amounts to about 40 watts and the saving in cost is about

$8.00 This saving in cost was the secondary objective of revising the B+ filter.

 

 

Part 2: Elimination of Certain Functions

​

Because of the limited use of the remote control connector J-105, and the infrequent need for the squelch facility, consideration was given to the removal of these items. A cost study was made on the removal of the remote connector. Taking into account the savings in material and labor, the figure comes to nearly $8.00, a worthwhile saving considering the limited use made of this connector.

 

At the start of the contract it was considered desirable to drop the squelch requirement. An investigation was completed showing a cost saving of $14.00 by eliminating this function. Later on, the Signal Corps decided to make this an optional unit retaining  the cable wiring and function switch and leaving out the parts in the AF unit. With this method a cost reduction of about $10.00 is still possible. An optional kit of parts can be made available for field installation of the squelch facility.

 

Part 3: IF Unit

​

Redesign of the IF unit to employ electromechanical filters was the first project on which work started for this contract. Work had been completed on contract DA36-039-sc-56636 resulting in the development of 455 kc filters having bandwidths of 2 kc, 4 kc, 8 kc and 12 kc. These filters were incorporated into the first model IF units built at Burbank. These first units were strictly experimental, to show the superior selectivity characteristics of mechanical filters and to uncover the problems involved in adapting these filters to the R-390 receiver. The problems encountered in these first models were spurious responses, blocking on strong signals outside the IF passband, and feedthru around the filters. As the contract proceeded, further design work was done on these filters to improve their performance in these respects. At first it was thought that plug-in filters would be a great convenience, but tests showed that feedthru was aggravated by the close spacing between input and output pins. The IF strip built up using these plug-in filters was used in the "B" model receiver. Another IF strip was built up using single-ended filters but with solder lugs instead of plug-in pins. The better shielding improved the feedthru problem slightly but greater improvement was desirable. The next IF strip was built using double-ended filters and it proved to have about 20 db improvement in feedthru over the single-ended type. One "A" model R-390 and the "A" model R-391 were supplied with these fillers. These later filters are also superior to earlier models in the matter of spurious responses. Seven disc filters are now employed for optimum shape factor. The improvement in shape factor over the tuned circuit IF is especially great at

the narrower bandwidths. Besides the improvement in filters, the circuitry was also carefully examined for possible improvements-and simplifications. In order to improve the blocking characteristics, the first IF stage gain was greatly reduced by increasing its bias. This has two beneficial effects. First it reduces the strength of the signal at the input to the filter, and secondly the high bias improves the blocking and cross modulation characteristics of the first IF stage. Late in the program it became apparent that the Signal Corps required all tuning adjustments to be removed. This requires the use of very wideband transformers so that changes in tubes, temperature and humidity have no effect on circuit gain or tuning. New interstage coupling transformers were designed to give the required flat top over a wide frequency range. Tests were conducted to determine performance under conditions of capacitance change, temperature and humidity. Up to 2 uuf capacitance variation is permissible on

either primary or secondary. This is sufficient to accommodate maximum and minimum limits on tube capacitance variation. Performance of the transformers under limits of temperature and humidity was very good. The R-391 "A" model IF strip incorporates all these latest improvements. Parts were rearranged on this chassis to afford greater accessibility underneath the unit. Brass shields were changed to aluminum, and the B+ decoupling network was also eliminated on this unit. It appears that an initial adjustment is necessary to align each transformer after the IF unit is built. After this initial adjustment, no further alignment should ever be necessary. Figures 15 and 16 show the latest model IF unit as supplied with the R-391receiver.

 

Other changes made in the IF unit include a more positive BW switch detent which at the same time is easier to operate, fewer switches and a toroid IF filter inductor in the diode load circuit. The magnetic field of the toroid is closely confined and causes less trouble with BFO voltage induction in the IF output circuit than the old pie-wound inductor it replaces. Shields in the IF are made of aluminum and a simpler method of mounting them is used. As a cost reduction item, the capacitor mounting clips were removed. The leads of all the capacitors are adequate to support them, even under conditions of shock and vibration. A carbon potentiometer is used in the carrier level meter adjustment circuit. It offers less erratic operation and smoother control than

the wire wound control.

 

Part 4: Power Line Filter

​

The AC line filter was tested for effectiveness in preventing external signals from getting into the receiver through the power lines. At frequencies up to 15 mc the filter gave better than 50 db attenuation, falling off to 10 db attenuation at 30 mc. The filter also provides a considerable reduction to internal signals by not allowing them to be radiated out the power line. Attempts were made to get equivalent performance from a less expensive filter. The simplest type of filter, a Sprage Hypass feedthru capacitor, was placed in each side of the line and tested. Its performance was found to be considerably

worse, particularly at low frequencies, although at frequencies above 25 mc it was slightly better than the old filter. A single L-C pi filter was then tried. It was similar to the old filter but had fewer inductive elements. The old filter was designed for a cutoff below 100 kc, but there was no real need for such low cutoff. Eliminating four of the inductors raised the cutoff frequency, but above .3 mc the new filter had just as good or superior attenuation characteristics as the old filter. This new filter was placed into the production receivers as soon as possible. After repackaging and elimination of the 164-1 connector, it now serves for the cost reduction receiver. The line cord now fastens to the filter by means of screw terminals protected by a cover. The power plug connector was changed to a U-120/U conforming to MIL-C-3767. Total saving on this line filter and power cord is about $17.00. Figure 23 shows the new line filter.

 

 

Part 5: KC and MC Band Stops

​

Early attempts to solve the problem of preventing damage to the gear train and VFO through failure of the kc and mc end stops resulted in various clutch designs for the kc and mc knobs. The theory was that for normal operation the clutch would be strong enough to provide solid mechanical coupling between the knob and shaft. Upon reaching the end stop, the extra torque would cause the clutch to slip and prevent damage to the gear train and VFO. The simplest design consisted of a bronze clutch insert between the knob and shaft. Unfortunately it had a tendency after slipping several revolutions to gall up and grab the tuning shaft. A later design used a more complex clutch assembly in the knob. This provided more reliable operation due the closer tolerances on the value of the slip torque. However, the greater cost of this clutch knob led to an investigation of simpler means to provide the required protection. As a result of continuous difficulty with the end stops on the production R-390 receiver, this work was accelerated. A new ten-turn stop was developed having much greater strength than the old. Because of the urgency of the problem this new stop was immediately incorporated into the production receivers. This same stop is used virtually unchanged in the "A" and "B" model receivers.

 

Part 6: Crystal Oscillator

​

Simplification of the crystal oscillator depended largely on development of a new frequency scheme to reduce the number of crystals required to accomplish the desired frequency conversion. Substitution of a fixed first crystal oscillator would also eliminate two 32-position switches and four crystals. Investigation of possible frequency schemes showed that in order to avoid third order spurious responses a first oscillator would be required of no lower than 17 mc. This would allow one fourth order spurious response, but the total number of spurious responses up to 8 mc and below 9th order would be

reduced from 15 to 6 by the new frequency scheme. Use of crystals up to 17 mc in the second oscillator allows a further reduction in the number of crystals by three. The first oscillator built up used CR-l9/U crystals with no oven. Tests showed the stability over the required temperature range of -40 to +85°C was inadequate for Signal Corps requirements. All later models incorporated ovens to obtain the required stability.

Figures 18, 19 and 20 show the new crystal oscillator. In comparison to the oscillator in the production R-390 it has been greatly simplified. Since the first crystal oscillator is fixed in frequency it has been removed from this chassis and placed on the RF. This was found to be necessary also to prevent interaction between the two crystal oscillators. This interaction between oscillators takes the form of internal signals generated when both oscillators get into the first or second mixers and produce the VIF frequency. Removing the first crystal oscillator to the RF chassis greatly reduced the intensity of these internal signals. It has proven to be impossible to completely eliminate them, but they should all be lower than the 3.5 uv limit given for internal signals. The oven is easily removable with the crystal oscillator in place. Removal of the crystal oscillator itself requires dropping the front panel, but the RF unit need not be removed. However when taking the RF unit out it is most convenient to leave the crystal oscillators attached and remove both together. This is an advantage in that it allows the two to be tested together more conveniently. The frequency scheme as finally adopted is shown in table I. The number of crystals in both oscillators has been reduced from 23 to 16. Design of the oven proved to be a difficult problem. Since the crystal switch was removed from the oven, there was considerable heat loss through the crystal socket leads and at low temperatures the oven ran cold. Placing extra heater windings around the crystal mounting board counteracted this loss of heat and made much better temperature compensation possible over the range -40 to +65°C. A Fenwal thermostat is used in an effort to obtain more reliable operation. Several turns of wire wrapped around the thermostat reduces the temperature cycling to an average of about 1°C. This figure varies somewhat with the particular thermostat and ambient temperature, but is fairly

representative at room temperature.

 

 

Part 7: RF Coils

​

Since the 16-32 mc coils were causing considerable tracking difficulty in production, work was first begun on this band. The method of approach was to increase the length to diameter ratio of the coil by using a smaller core diameter. Reduction of the diameter from .25 to .20 inches accomplished the desired result. As soon as the design was completed, steps were taken to put this coil in production receivers. Later on as ferrite cores of improved characteristics became available, it was found that the same good tracking could be obtained with the larger diameter coil using the higher permeability

ferrite core. Finally, all RF coils were designed to use this same type of ferrite core. In addition to their higher permeability, these cores also have a higher Q, helping to overcome the loss of image response resulting from the elimination of one RF stage. The VIF coils have also been redesigned for use of a common core. This core differs from that used in RF coils however because this application called for a lower permeability, higher Q core. Coil frames have been designed for simpler assembly and teflon terminals are used to maintain higher circuit Q. The antenna links have been redesigned to attain a more uniform input impedance. The introduction of ferrite cores in the antenna coils raised the primary input impedance. This had a beneficial effect on

noise figure but reduced the selectivity. In order to return to an impedance near the nominal figure of 125 ohms, it was necessary in most cases to reduce the number of primary turns. This improved the image rejection and cross modulation characteristics of the receiver at a slight sacrifice in noise figure. Figures 9 and 10 show the new RF and antenna coils respectively.

 

Part 8: Autotune Motor

​

One of the most important factors in cost reducing the R-391 receiver is the AC Autotune motor. Although the AC motor itself is more expensive, elimination of the PP-629, 28 VDC power supply required for the 28 volt motor, results in a great overall savings. Besides the saving in cost, elimination of the PP-629 results in far greater convenience and versatility for the receiver, to say nothing of the saving in weight and space. A suitable motor was obtained and tested in the "B" model. Operation was satisfactory and more quiet than the current R-391 Autotunes. This is largely due to the lower speed at which the new Autotune motor runs. Being governor controlled, this motor does not exceed 5000 RPM. As a result the set up time is increased to a maximum of 25 seconds which is 5 seconds longer than that of the current R-391 receiver. However this lower speed should result in longer life to the gear train.

 

 

Part 9: AF Filters

​

The need for a low pass AF filter with a cutoff of 3500 cps in this receiver was always rather questionable. It does provide the minimum bandwidth for maximum intelligibility of speech, but it was somewhat doubtful if it was worth the extra cost. Now that the IF employs mechanical filters with their flatter tops and steeper skirts, the AF filter becomes even less necessary. If, for instance, the IF bandwidth is set to 8 kc for speech reception and there is strong interference from a signal 4 kc away, it only requires a

slight retuning to push the interfering signal off the edge of the passband while still maintaining the intelligibility of the desired signal. In extreme cases where interfering signals are found on both sides of the desired signal, the bandwidth may be set to 4 kc and the signal tuned for maximum intelligibility and rejection to interference. In many cases, this may result in tuning to one of the signal's sidebands for best reception. Under this condition the 4 kc filter provides a sufficient bandwidth for optimum speech

intelligibility. It appears then that under most conditions of interference, careful tuning and judicious use of the bandwidth control eliminates the need for the 3500 cps filter. An investigation was made of the cost saving resulting from elimination of the 3500 cps low pass AF filter. Upon presentation of the cost information to the Signal Corps, authorization was granted for its removal on the final models. A twin-T feedback amplifier circuit was tested as a possible substitute for the 800 cps filter. The shape of the response curve using such a circuit is very similar to that of a single tuned circuit. It is narrower at the nose and wider on the skirts than that obtainable with a band-pass filter. The bandpass filter was redesigned for use in the plate circuit of the first AF amplifier so that the interstage transformer could be eliminated. The high impedance

filter has slightly greater bandwidth so that the beat note can be varied a little more and still remain within the passband. Figure 34 shows a comparison of audio selectivities among the various bandwidths and the sharp AF filter.

 

 

​