The Collins model 75A-4 is a 22 tube single/double conversion (original schematic of THIS UNIT, date mfg July 12, 1956) communications superhet receiver designed on the converter principle, i.e. the LO operation, and therefore tuning, is achieved at the lowest frequency at which the receiver operates, this ensures maximum stability and very low drift on all bands (almost negligible). The frequencies above the lowest frequency at which the receiver operates are transferred through Xtal converters to be tuned into that lower frequency range (that's why drift is negligible in all bands); this was the basis of Collins' "magic".

The manufacturing period of the 75A-4 model can be divided into 3 editions:

First, 1955: Q-Mult mod and 4:1 tuning reduction.

Middle, 1956: Final Noise Limiter mod and some undocumented circuit changes.

Last, 1957: Final S-Meter mod and the undocumented change of the 15 meter band.

The design of this receiver is simpler than it seems, the use of a simplified front-end and mechanical filters have a lot to do with it, in fact the 75A-4 is a 1.5 to 2.5 Mcs Medium Wave single conversion receiver that has a RF front-end line "hacked" (in 'modern' speaking, see the align page) to provide the rest of the frequencies. In this design it is very important that the bandwidth of 1 Mc (2.5 - 1.5) of the input area to the mixer (V-5) has a response curve as flat as possible because it needs to emulate the frequencies as would be received by an antenna, any irregularity of this area will affect all bands. The “hack” of putting a converter system in front of a mixing system was a very popular resource in the 50s, practically all amateurs who wanted to expand reception ranges did it (remember that this was a time when many Hams were electronics enthusiasts, so the level of self-construction or intervention in the device was relatively high).

On single conversion, the 75A-4 is a 1.5 to 2.5 Mcs Medium Wave receiver with comm-tech specialties. This 1 MHz radio can be aligned via the usual methods used to align 5-tube radios, setting the IF and low RF via coil and the high RF via trimmer. The front-end circuit begins with an RF transformer (T1) followed by an RF amplifier (V-2), which tunes the entire bandwidth of 1 megacycle using the secondary coil of the the RF transformer (this variable inductance is equivalent to a section of the tandem capacitor in mixer-style receivers) to the grid of V-2. The V-2 plate gives output to the signal in broadband operation (untuned) to V-3. On 160 V-3 acts as an RF amplifier using only the high side of the 6BA7 pentagrid converter tube. The output of the V-3 has tuned plate output (L18) and tuned grid input (L22) to V-4 (2nd mixer) separated by a 5.7 Mc trap (L18 and R22 are also parts of a "tandem").

On double conversion a 7-band converter circuit is inserted in the RF line, the single conversion design is 'hacked' (to make it understandable in a current way). The converter-mixer design was very popular among DIY radio amateurs at that time, mainly on high frequencies. This 'hack' consists in an HF converter that operates before the LF mixer. The RF output from V-2 (6DC6) includes a set of coils on the input grid of the pentagrid 6BA7 converter (pin 7 ), which now acts for what it was designed for (converter tube -xtal mixer-), and to complete this its control grid (pin 2) receives a xtal controlled RF signal from a local oscillator (V-4, 12AT7). On double conversion the 75A-4 inverts side bands, but the panel is marked taking this inversion into account; therefore the side band indications on the panel will only be erroneous for the 160 meter band.

455 IF section: The output from the Mixer stage (2nd mixer) is taken via a set of mechanical filters, followed by the first IF amplifier, a Q-multiplier circuit that has only null (refection) effect drives the signal to a second and after a third IF amplifier. In parallel to the last IF amplifier is the AVC amplifier (AGC) which provides the AVC bias voltage. 6BA6 vari-mu RF pentodes are employed in these IF stages, the three in sequence are controlled from the AGC line. Following these stages are the detectors, an AM-MCW detector, an SSB-CW detector, and an AVC (AGC) detector.

AF and PS section: The output from a signal detector circuit is applied to an optional Noise Limiter, which feeds the 1st triode portion of a 12AT7, which is resistance capacity coupled to the 2nd portion, and this feeds a 6AQ5 tube that loads the output transformer. The power supply circuitry has a power transformer of 115/300+300 with one section tapped at 100 volts to provide voltage to the bias rectifier, B+ is provided by the usual full wave rectifier with filtered output and parallel heather chain; the noise limiter tube filament has a 10 ohm resistor in series to limit hum from the filament.

The mechanical filter, the 'heart' of the 75A-4. Without this component, all its special features disappear.

Service Bulletin 2A S-Meter circuit implementation, but with the original overchassis mounting design.

This applies to units below SN 3423. Mouse over to see actual S-Meter wiring

No internal mounting plate required for potentiometers, no overcrowding an already crowded circuit area (mouse over the graph). Only two more phenolic spacers are needed; I have used a couple of these, but they don't need to be phenolic, can be metallic.

Collins oddities and 75A-4 thoughts:

RF GAIN is a control that is located on the panel because it serves to regulate the sensitivity of the receiver depending on the actual propagation conditions of the band in use. It should not be set fully clockwise by default relying on the AGC to automatically correct level differences (to set it this manner it would not be necessary to place it on the panel). An S-Meter is an indicator of the strength of the received stations which reflects the sensitivity of the receiver and the level of propagation. When the sensitivity of the 75A-4 is reduced with the GAIN control, the S-Meter should act accordingly, it should decrease ... not increase as happens in 75A-4. That's the surprising part on many Collins receivers and others who have copied the "system" (e.g. Heathkit). This "system" has become so normalized that it is believed that this is the correct operation, and it is not; what the S-Meter reflects is the increase in the AGC bias voltage. This is a cheap system to control the sensitivity of a receiver, in fact there is no a true sensitivity control because if the gain is reduced the S-meter no longer provides S units of incoming signal, but rather an internal voltage level; it's absurd. See the difference when operating a Hammarlund receiver: AGC line and GAIN control are separated, thus more gain, more S-Meter indication, less gain, less S-Meter indication; the S-Meter always provides S units of incoming signal, which can be attenuated by operating the gain control, that's the right operation. For SSB reception, the best performance is achieved by polarizing the AGC line a little, to do this RF GAIN must be set to 9 instead of 10 (S1 on the S-Meter). Collins advises on the AVC behavior in Section III, Tuning SSB signals, paragraph d.

75A-4 is an "squelched radio" when Noise Limiter is ON. This is more a feature than a bug (a voltage threshold is needed for the NL). When the noise is very low, the NL is not polarized and does not open. I don't know if the same thing continues to happen with the SB-2 version of the noise limiter. How to check this feature? The easiest way to check it is to disconnect the antenna, select AM, NL ON: High decrease of background noise. This is because the NL aperture level depends entirely on the RF background noise, not on setting a minimum bias voltage. The SSB position injects the required bias. Don't be surprised if the receiver goes silent after switching to AM with the Noise Limiter in the same advanced position that worked OK on SSB.

75A-4 is not as functionally sensitive(*¹) as other similar receivers, but it is a common feature of the converter design due wideband implies less Q. The lack of functional sensitivity is due to the effect of noise on the signal. The converter system needs to compensate for all these drawbacks trying to make the front-end stage as selective as possible, and also increasing Q in the mixing system that follows, but here Collins designed a receiver with very limited IF dynamics, with a "low impedance AVC line to minimize blocking on strong signals", in his own words. But this must be well understood, it means that instead of boosting the AVC so that it can be used with signals of different levels, what it does is compact the level of the signals to adapt them to the AVC line, this makes life less complicated but it is a wrong decision.

75A-4 tends to get noisy and has a very questionable signal-to-noise ratio from 15 meters onwards, but the tendency to be noisy is also a common feature of converter design because it generates a wide area (broadband) at the output so that a mixing stage can select frequencies with a variable local oscillator operating at low frequencies. The noise effectively perceived depends entirely on the bandwidth of the mechanical filter inserted; more empty space around the signal, more perceived noise; less empty space, less perceived noise. Conclusion: the mechanical filter must be exact to the signal amplitude, the "secret" to 75A-4's efficiency.

Why do Collins receivers have significantly less drift than other receivers? → Mixer design versus Converter design. Collins explains it, but omits some aspects on the matter (QST ad related to 75A-4, Oct-1958).

75A-4 relies too much on mechanical filter, depending on one or the other a different receiver will be obtained. This high dependency is very criticizable because there is almost nothing that allows to customize bandwidth and Q upon receiving except mechanical filters. I say "almost" because REJECTION TUNING can be used to cut off a noisy side area of the received signal, resulting in an improvement in the signal-to-noise ratio.

75A-4 has a questionable AVC starting point, this is a design consequence. IMO excessively delayed with the excuse that it does not act on the noise, but the nature and level of noise are diverse, in fact noise is a "special type" of signal (*²), so what was said in this note on the subject does not make sense, the real reason has to be something else. 75A-4 needs to increase signal-to-noise ratio, but it should be clear that increasing the S-N ratio will make the current AVC design insufficient to handle a signal with greater dynamics, and will almost certainly require manual action. Although in an unmodified 75A-4 the AVC/SSB system works acceptably good, when receiving strong signals it is advisable to set RF GAIN to 9. There is controversy regarding this, many consider that the AGC (AVC) should cover all signal circumstances, but it is unrealistic to expect an AGC system to cover from a very powerful nearby station to a very weak distant station. Collins instructs on RF gain use in Section III of the manual, Tuning SSB signals, paragraph d (as already stated, this control is intended to be used).

This 75A-4 has been tested in difficult conditions to see if the reputation of the brand brought something new, but the result is a bit disappointing, in reality 75A-4 only performs very well in easy conditions. Connected to an "antenna" made up of a wire of about 2 meters (6.5 feet), when compared with Hammarlund HQ-160 (modified version, see HQ-160 AGC page), HQ-160 wins in sensitivity and signal-to-noise ratio, although 75A-4 wins in frequency stability of course; absence of drift and dial accuracy were the most powerful features of the Collins equipment. The 75A-4 front-end is a simplified broadband design and the IF chain has its Q attenuated by shunt resistors, making the 75A-4 appear to be a "solid brick" with little adaptability to different scenarios. According to Collins, these are design features to achieve flat AVC, but IMO flat AVC is the adaptation to these "features" that cause poor signal-to-noise ratio. To compensate for the broadband characteristic at the output of the converter, high front-end selectivity is needed as the converter heterodyne system only tunes through the front-end. The 75A-4 front end is an economical design that uses coils in parallel from 40 meters onwards (see alignment page, table at bottom), this is a low Q design that has even less Q as frequency increases. When a wideband feature is present, selectivity and dynamic range are automatically affected, both are usually corrected in later stages, but this is not the case here because Q is attenuated again in the 455 Kc IF chain.

IMO the 75A-4 is the last radio Collins made for old school (technical based) radio amateurs, but to their surprise, many of those amateurs were not as accommodating as they thought, and the 75A-4 was subject to a considerable number of Ham mods; so something was wrong with the 75A-4. Today this receiver has very good opinions and blessings, but it's probably more due to the 'Collins myth' and checking the 40 meter band than a complete user experience, being that many 75A-4s end up as an ornament in a Ham shack or keeping company with other equipment in a collection. To preserve something it is mandatory to make it useful, devices like the 75A-4 deserve to be used, and whoever wants to preserve them better think about putting them to use. Nowadays a 75A-4 can be used as long as the issue of paper/mica capacitors has been reviewed, and as long as the user knows what a vintage communications receiver is and what it can do. The S line (1959-1975) was advertised using broad audience marketing principles, it seems that this line was aimed at consumers of high-end appliances, not at clients or technically minded users. The S line anticipated what amateur radio would be in the future, but whether that was for better or worse is debatable.

Conclusion: This is a receiver whose excellent reputation precedes it, but I think is a reputation based more on the myth of the brand than on an objective analysis, in fact it is very difficult to find out precise opinions on the 75A-4, except, of course, the inevitable references to audio quality (sometimes negative), and the usual admiration for the manufacturing craftsmanship (always positive). The audio quality from a properly functioning 75A-4 is very adequate for communication service, which is what the 75A-4 is designed for. Collins took great care in the craftsmanship of his devices, but what works in electronic equipment is the circuit diagram, not its appearance. What was undoubtedly impressive at its time is dial reading accuracy, frequency stability (both common in all the Collins gear by design), the essential contribution of the mechanical filter, and the peculiar PASSBAND TUNING action to 'throw off' an interfering side signal by letting it fall into the side abyss of the mechanical filter, but 75A-4 performance could be better, especially regarding to signal-to-noise ratio; this is very evident when comparing a 75A-4 with a mixer+converter design receiver. If the S-N ratio issue is taken into account by selecting AT LEAST mid-level signals, and it is considered that the RF GAIN control is on the panel to be used, 75A-4 is an excellent receiver perfectly usable nowadays, but no one should be surprised that a weak signal that can be heard with an average communications receiver that uses the mixing and tuning system on high-frequency (e.g. Hammarlunds, frequency-dependent drift, but very good S-N ratio -excellent when compared with Collins receivers-), cannot be heard with a 75A-4 because it is covered by noise. The excellence that a 75A-4 can show depends on how noisy the band in use is, this happens with all communications receivers of course, but in this regard, it seems that inside a 75A-4 two different receivers coexist in one. When there is little noise in the band, a 75A-4 will offer excellent performance, but in the presence of a lot of noise, a 75A-4 will offer very questionable overall performance, too questionable for the existing noise; that is due to the low Q design of the front-end combined with the flat broadband design of the IF chain. Fortunately, the most modern TV IF tubes with transconductances greater than 10 mA/V help improve the effective Q of a 75A-4 RF amplifier. These tubes did not exist when the 75A-4 was designed (1955), they were designed to be used in very wideband circuits, when used in narrower band circuits they either make no difference or cause problems. Replacing 6DC6 with 6GM6 (1960), or any other very high transconductance tube, substantially improves the signal-to-noise ratio of a 75A-4, and this evidences that the 75A-4's front-end is a wideband design with low Q.