An example of the early use of the 10.7 MHz IF in the UK was provided by the Mullard GFR520 of 1949, as described in the attached WW excerpts. This receiver covered 54 kHz to 110 MHz. A 10.7 MHz IF strip was used above 27 MHz. Below that frequency, a 455 kHz IF was used, which suggests that Mullard was following American practice, whereas UK practice, pre-Copenhagen Plan, favoured 465 kHz. Anyway, it shows that the 10.7 MHz number was known in the UK quite early on.
Cheers,
Steve
Another IF that would appear to justify inclusion in this list is 9.72 MHz. It appears to have been used in British and perhaps European marine and aviation VHF receivers in the late 1940s and into the 1950s. Examples were the STC STR.9-X (aviation) and “Receiver 62H” (marine). It was also found in the BBC HR/12 outside broadcast wireless microphone receiver, which was described as being based upon the EMI Type 1250 receiver.
It could be that 9.72 MHz predated 10.7 MHz for non-entertainment VHF receivers. But once established for domestic FM receivers, 10.7 MHz then became the choice for marine, etc., applications. The introduction in the late 1950s of narrow bandwidth 10.7 MHz crystal IF filters by Cathodeon and others either reflected an industry preference for this frequency, or pushed the industry to it. Cathodeon’s initial such filter offering included model suitable for 50, 25 and 12.5 kHz channelling.
Cheers,
That is an intriguing choice- not only for its precision but also its presence in the 31m broadcast band. Admittedly, just about any IF ever chosen happens to have some sort of spectrum activity nearby but it does seem odd to place it somewhere where there would have been world-wide high-power broadcasting, rather than sporadic and lower-power marine/aviation/commercial activity elsewhere in the HF spectrum. No doubt the mists of time hold an entirely cogent argument for it.
The first article in Synchrodyne's 10th Dec post is also interesting for the mention of the dual-conversion Eddystone 750, and the implication that previous similar attempts at multiple conversion ran into excess birdie-type problems- the suggestion being (and, as always, one needs to take manufacturer's claims with a pinch of salt, particularly those from the days of "optimistic" advertising claims) that the 750 was supposedly rather better in this respect. It certainly endeavours to quarantine 2nd mixer/oscillator in its own separate screened module but subsequent developments of the theme, such as the 910 and 830, took rather more care in isolation with quite complex screening and use of feed-through capacitor filtering. The 880 receiver seems to have been a (literally) heavyweight tour-de-force in screening and filtering complexity and other contemporaries, such as the Racal and Collins receivers, also took metalwork compartmentalisation and electronic isolation and decoupling to new extent. It all suggests that double/multiple conversion could create as many problems as it supposedly solved.
Yes, I agree that 9.72 MHz is a strange IF number, and one that does not lend itself to post-facto rationalization. But as you say, there was probably good reason for it at the time.
Something that has just occurred to me is that the “EMI 1250” that was the basis of the BBC HR/12 receiver (see: http://www.bbceng.info/ti/eqpt/HR_12.pdf ) might have been the same as the “HMV 1250” briefly mentioned in Wireless World 1949, pertinent pages attached. This was an early British VHF-FM receiver. Evidently there must have been one or two such developed before the introduction of UK FM broadcasting was delayed by that silly BREMA-inspired digression into an FM vs. AM comparison. Anyway, the tuning ranges for the BBC HR/12 (EMI 1250) and HMV 1250 were exactly the same, namely 87.5 to 94.5 MHz. This was more-or-less the initial part of European Band II (then 87.5 to 100 MHz) that was allocated to broadcasting in the UK. (Actually 88 to 94.6 MHz according to: http://www.bbceng.info/Technical%20Revi ... ground.pdf.) Coincidences of this nature being unlikely, I am inclined to the conclusion that the HMV 1250 (EMI 1250) was an example of a domestic FM receiver that used the 9.72 MHz IF, for which the working assumption is that it was developed for marine, aviation, etc., VHF receiver use. Familiarity may thus have been a factor in its choice for the 1250, even though the American move to 10.7 MHz was probably known.
An apparent outlier and a one-off though was the 14 MHz chosen for the Ambassador/BBC VHF-AM/FM comparator receiver, which dated from circa 1951. This tuned 87.5 to 95 MHz.
Compartmented and heavily screened construction for multiple-conversion receivers seems to have been the norm where high performance was required, and the practice was carried over into the solid-state era. As far as I know, receivers such as the Marconi H2900, Eddystone EC958, Racal RA1772, etc., conformed to this pattern in a serious way. I have heard anecdotally that some consumer-level receivers that were less-well screened, such as some of the Drake models, did suffer to some extent from birdies. Also, the dealer who sold me my JRC NRD-525 way back when commented that the optional VHF converter for this model had non-negligible birdie problems. If I have this right, the VHF converter board went into a vacant slot alongside the other boards, with no special screening.
Cheers,
Steve
Taking of double-conversion HF receivers, another strange IF number is the 4.034 MHz used for the 1st IF on the highest frequency band in both the Heathkit GR-78 (of late 1969) and the Kenwood/Trio QR-666 (of circa1974).
Both were six-band receivers covering from LW broadcast to 30 MHz. Both were single-conversion, to 455 kHz, on the five lower bands, but double-conversion on the 18 to 30 MHz band. Both used three dual-gate mosfets in the front end, as RF amplifier, mixer and 1st IF amplifier on the lower bands and as RF amplifier, 2st mixer and 2nd mixer on the highest band. Presumably this approach was a relatively simple way of getting adequate image rejection on the highest band without the need to go to a four-gang arrangement.
Possibly Kenwood/Trio simply copied what Heathkit had done, but even so, it surely would have put some thought into the matter, if only to check that Heathkit’s choice of 4.034 MHz as the 1st IF was the right one.
So what was special about 4.034 MHz, and in particular, why was such a precise number required. The precision implied by this number suggests that, for example, neither 4.033 nor 4.035 MHz would work as well. Given the diversity of 1st IFs found in double-conversion HF receivers, this implication might be possible, but it does not look to be probable.
Both Heathkit and Kenwood/Trio used an oscillator-low second conversion, with a crystal-controlled second oscillator operating at 3.579 MHz. Therein might lie the answer. That was in fact the colour subcarrier frequency for NTSC-M television (actually 3.579545 MHz). So crystals for that frequency would have been readily available as consumer product components, and with relatively attractive pricing. Perhaps the IF choice came down to being a case of “can we work with an NTSC-M standard crystal”. Then the outcome was that 4.034 MHz was found to be a satisfactory number both in terms of being in the right ball-park for the desired image rejection performance, and not causing any undue birdie or other interference problems.
Of course, that might not have been the case, but it does have a ring of plausibility as an explanation for an otherwise strange and apparently very precisely specified IF.
There might have been the possibility of mutual interference when these Heathkit and Kenwood/Trio HF receivers were operated close to NTSC-M TV receivers, but with TV receivers my understanding is that the intercarrier was more of a bad actor than the colour subcarrier when it came to radiation.
Cheers,
Steve
The NTSC crystal hypothesis seems very plausible- a traditional free-running LC 2nd conversion oscillator at around this frequency could be a liability as regards drift- even if the 1st LO (i.e. main tuning) was adjusted to compensate, the wanted signal would now be unfavourably placed in the 1st IF passband (assuming, as would be hoped, that this at least provided initial or "roofing" selectivity). Nowadays, we're accustomed to crystals being cheap and ubiquitous in everything from alarm clocks through radios to computers, churned out by the umpteen million in a bewildering array of frequencies but it would have been a different matter then. Could 3.579545 MHz crystals perhaps have been the first widely-available and reasonably cheap type? I gather that they were also used, divided down in various ratios, as the basis for tone dialling. Additionally, "osc low" for 2nd conversion will preserve the sense of USB/LSB BFO setting (or sideband filter selection, if used) compared to the single-conversion bands.
I am a little surprised that the dual conversion wasn't switched in, say, a band lower though- whilst a set that switches across an array of broadcast bands of relatively narrow span may not be too affected by misleading image reception, a general coverage set of 455kHz IF, presumably here with three-gang tuning, would be showing misleading "ghost" reception in the images of 16 and 19m bands at least.
I had also wondered why double conversion was not used on the 7.5 to 18 MHz band as well as on 18 to 30 MHz. Heathkit quoted image rejection numbers of ≥35 dB for the 18-30 and 3-7.5 MHz bands, but only ≥25 dB for the 7.5-18 MHz band. So the last-mentioned would have benefitted from double-conversion. Now that I have thought about it some more, I wonder if there were lowish harmonics of the crystal oscillator that might have been problematical in the 7.5-18 MHz band without additional and serious screening. From what I have read about TV IF selection, it seems that the 5th harmonic of anything potentially nasty is about the highest that one has to worry about. So 18 MHz just clears the 5th harmonic of the crystal oscillator.
Perhaps too a mitigating factor was that both the Heathkit GR-78 and the Kenwood/Trio QR-666 had bandpsread tuning (of the fiddly kind that first required precise setting of the main tuning). The bandpsread drums could be marked optionally with the broadcast or amateur bands. That facility would have helped with avoiding accidentally tuning into a ghost, something that I know is all too easily done with non-bandpsread shortwave receivers. Kenwood/Trio must have been happy enough with its product (in relationship to its price point), as essentially the same arrangement was used on its later R300 model.
Cheers,
Steve
One thing that strikes me about the generation of HF PLL synthesised receivers that were extant about 10-40 or so years ago (thus hopefully qualifying as "vintage"!) is the remarkable stopband selectivity achieved by VHF 1st IF crystal filters, thus enabling 2nd conversion straight to a low final IF of the 455 or 1400kHz region without an intervening and intermediate IF (tautology unintended!). 1st IF frequencies I am aware of are scattered between 35.4MHz (Racal RA1772) and 70.455MHz (JRC NRD505/515) and I'm sure higher ones were used in some sets. To put this in perspective, imagine this 1st IF as a single-channel VHF receiver with a very low subsequent IF- with classical LC filtering, this would be asking for almost unusably severe image problems. Before crystal 1st IF filters became mainstream, the usual 1st:2nd IF ratio seemed to settle in the region of 15-20:1, presumably representing optimum compromise between effective 1st and 2nd IF image rejection capabilities without unmanageably complex and expensive 1st IF filters. Marconi's Mercury receiver is a good example of an exception to this rule, with 1st IF of 4.5MHz and 2nd IF of 85kHz- here, the 1st IF filter was an optimally-coupled module with 4 tuned circuits and the 2nd LC conversion oscillator presumably took some care over stability and temperature compensation for the reasons discussed above re. NTSC crystal usage. Examples of this approximate ratio include Eddystone 750 (1.62MHz:85kHz), 910 (1.4MHz (centre):85kHz), 830 (1.35MHz (centre):100kHz) and STC R4187 (2.15MHz:100kHz)- this latter having 4 tuned circuits at 1st IF. The Mk3 version of the G2DAF receiver featured 5-5.5MHz 1st IF and 455kHz 2nd IF (i.e. approx 11:1 ratio)- even here, considerable emphasis was placed on the need for careful isolation and screening to ensure good stopband suppression. Racal's RA17 featured 3-2MHz 2nd IF and 100kHz 3rd- another example of a wide ratio, here, a four-gang "kHz" tuning capacitor was used, presumably in the interest of 3rd IF image suppression on what was pitched as a "reference" standard receiver.
That certainly provides food for thought.
In respect of the Racal RA17 and its final conversion, one might look askance at a single-conversion receiver that tuned 2 to 3 MHz and had a 100 kHz IF. A four-gang rather than a three-gang front end would confer some credibility, though.
Amongst upconversion HF receivers, the highest 1st IF of which I am aware is 81.4 MHz for the Rhode & Schwarz EK-070, which I think was a 1980s model. Its 2nd IF was 1.4 MHz, which gave a ratio of 58:1. The Marconi Oceanic (a rebranded Dansk M3000) of circa 1984 had a 75.0 MHz 1st IF. The Eddystone 1650 of about the same time had a relatively ordinary 46.205 MHz 1st IF, but in the very wide AM mode (14 kHz bandwidth), the 46.205 MHz crystal filter provided all of the selectivity, there being no corresponding 1.4 MHz filter. So definitely those crystal filters were very good in-situ. One imagines that impedance matching and screening of possible bypass paths needed to be near-immaculate.
Conversions to and from 10.7 MHz provided some high IF ratios, too. 10.7 MHz to 455 kHz was not too high, at 23.5:1. Still I suspect that the combination might have seen increased used for VHF R/T equipment after narrow bandwidth 10.7 MHz crystal filters became available towards the end of the 1950s. In some cases the 10.7 MHz filter might have been the primary selectivity element, with the 455 kHz filter playing a secondary role. 10.7 MHz to 100 kHz, 107:1, was done in the Plessey PR155 HF receiver, so the 10.7 MHz crystal filter must have been a very good one. Amongst VHF-UHF receivers, the ICOM R7000 converted from 778.7 to 10.7 MHz, and the Rhode & Schwarz ESM-1000 from 810.7 MHz, for a ratio of 76:1. Both I think had SAWF 1st IF filters. That in the R7000 was of wide enough bandwidth to accept an NTSC or CCIR TV channel.
Going back to the valve era, the Mullard GFR552 point-to-point ISB receiver had a 1st IF of 3.1 MHz and a 2nd IF of 100 kHz, for a ratio of 31:1. That would have been with LC filters at 3.1 MHz. The Marconi valve-era ISB receivers seem to have had 1st IFs in the range 1.6 to 2.9 MHz, although the solid-state Hydrus had 40 MHz, 5 MHz and 100 kHz, thus 50:1 for the final conversion. Probably this had a crystal filter at 5 MHz, though.
Some of the Drake receivers converted from a 1st IF of 5645 kHz down to 50 kHz, for a ratio of 113:1, but as far as I know these had crystal filters at 5645 kHz.
An extreme, albeit probably trivial example is provided by the small American weather band receivers, which covered 7 NBFM channels in the 162 MHz band. Some of those converted signal direct to 455 kHz, for a ratio of 356:1. But they served their purpose well enough.
Cheers,
Steve
I recently came across the attached item, from Wireless World 1958 January.
It confirms that BREMA did formally adopt the 10.7 MHz IF for Band II FM radio receivers.
By then 10.7 MHz was the de facto global standard, so for BREMA to have done otherwise could have been awkward for setmakers involved with exports, to say the least.
Nevertheless, Leak stayed with its own 12.5 MHz number, adopted for the original Troughline, through the Troughline II, Troughline III and Troughline Stereo iterations, all of which postdated the BREMA proclamation, not changing to 10.7 MHz until the solid-state Stereofetic of 1969.
On the other hand, Jason chose the 10.7 MHz IF for its JTV FM (Band II) and TV sound tuner (Bands I & III) of 1958, and stayed with this for the JTV2 version in 1960 and the Mercury II of the same year. There was American precedent for using 10.7 MHz for combined FM and TV sound tuners, at least where the apparent TV soudn coverage was Bands I & III.
Cheers,
Steve
Here is another variation on FM receiver IFs, though. I have come across one data point only that suggests that Russian FM receivers, which would have tuned the OIRT FM band of approximately 66 to 74 MHz, had a standard IF of 8.4 MHz, at least in earlier times.
This comes from a perusal of an article in Wireless World for 1961 August that describes a Soviet exhibition held in London. Pertinent parts of that article are attached.
In respect of Russian combined TV-FM receivers, it was noted that some use a second FM frequency changer to get from the 8.4 MHz FM IF to the 6.5 MHz TV intercarrier. At least that creates the impression that 8.4 MHz might have been the standard Russian FM receiver IF, which number appears to be reasonable for the transmission frequencies used. But one would need additional datapoints to confirm, so for now 8.4 MHz is a provisional number.
There are some other points in the article, more in the nature of obiter dictum in the context of this thread, but nonetheless interesting. It is mentioned that Russian TV receivers were often fitted with headphone and tape recorder outlets. Whether the latter were done “properly”, i.e. low-level direct from the demodulator and ahead of the AF stages, or by the “quick and dirty” method, i.e. by tapping the audio output transformer, is unknown, but I guess that there is some chance that the former was used at least in some cases. The picture tubes of the time evidently had a 4:3 aspect ratio, whereas I think that Western tubes were then still mostly 5:4, with the 4:3 type not arriving until the mid-1960s.
And the polar system for FM stereo broadcasting was mentioned. This does not seem to have had much mention in the West. Quite why the Russians adopted it rather than the GE-Zenith Pilot Tone system is unclear, although I recall reading somewhere that the latter was less suitable with the ±50 kHz deviation. The actual subcarrier frequency was 31.25 kHz, as far as I know arrived at because it was twice the 625-line TV line frequency, thus allowing the same system to be used for TV sound, although I am not sure it ever was.
Cheers,
Steve
The set under discussion in this thread;
www.forum.radios-tv.co.uk/viewtopic.php?f=3&t=11446
is interesting, not merely as an early example of double superhet technique, but also of the "tunable IF" system that proliferated in the 'fifties in professional (typically, but not always, spanning 1MHz) and subsequently amateur (often spanning around 500kHz) receivers. In this case, the use of the MW band as the tunable IF was dictated by "what's already budgeted for" considerations, being essentially an SW converter in front of a MW/LW receiver but could be viewed as asking for trouble with breakthrough of strong MW stations, even in the mid/late 'thirties. I wonder how much of a problem it was in practice (given that aerial advice in those days was frequently something like "as long and as high as possible"!) and whether it was found necessay to incorporate any high-pass filtering in the SW part front end, in addition to any normal input tuning?
If fixed 1st IF was always going to be something of a Hobson's choice as far as finding a quiet spot in the spectrum was concerned, tunable IF must have been rather worse, given that the usual LC filter technology of the time pretty much dictated choice of a span in the lower HF bands, frequently awash with strong RTTY/fax type transmissions even when away from the broadcast/amateur bands. For some of their models, Eddystone even used a narrow (100kHz (910) or 200kHz (other, both thermionic and semiconductor sets)) slice of the upper MW bands. Perhaps I'm envisaging ogres where only gnats exist, but the necessity for good screening/trapping/filtering seems to have become more prominent over the evolution of these sets, possibly users became less tolerant of spurios/breakthrough.
I wonder how much of a problem it was in practice (given that aerial advice in those days was frequently something like "as long and as high as possible"!) and whether it was found necessay to incorporate any high-pass filtering in the SW part front end, in addition to any normal input tuning?
The circuit is at 
for study but by this time, "as long and as high as possible" for aerials was already in its death throes; many manufacturers suggested limiting the aerial to just forty feet total by the time the A36 came out.
Thanks for that! So no added MW rejection, other than the LF-attenuating action of the tapped-down aerial coupling arrangement itself. Fixed gain on the SW front end too- I can see that the reflexed RF stage would need to be left optimally biased for low AF distortion but I'm surprised that there wasn't even potted down AGC on the mixer. A few other interesting points but I won't distract from the thread title...
Forty feet is quite short, really- a small garden, or not even two-thirds of a cricket pitch!
To be fair, there were quite a few car radio SW converters over the years that used the "MW IF" principle- as car radios had to be inherently well screened and filtered devices anyway, the principle here started on a sound practical footing and aerials were necessarily short. Always assuming that the installer was reasonably competent and scrupulous re. soundness of interlinking screening.
I have recently come across a couple of articles in “Radio News” magazine for 1947 which provide useful background information on American domestic receiver IF choices. This magazine is available at the excellent American Radio History site, at: http://www.americanradiohistory.com/Rad ... _Guide.htm.
The two articles are in fact part of a long series entitled “Practical Radio Course” by Alfred A. Ghirardi, and the two at interest here are: Part 53, in the 1947 May issue, and Part 56, in the 1947 November issue. Part 53 discusses IFs for AM, FM and TV receivers. Part 56 elaborates on FM IFs.
Looking at FM first, the original American FM band was 42 to 45 MHz, early receivers for which typically had an IF of 2.1 MHz. Later receivers for this band had an IF of 3.2 or 3.3 MHz. Then the FM band was extended to cover 42 to 50 MHz. For this, an IF of 4.3 MHz was used and recommended by the RMA. Apparently this number satisfied the criterion that it be high enough that all images were out-of-band, and was also adequately clear of the 3.5 to 4.0 MHz amateur band.
With the upward movement of the FM band to 88 to 108 MHz, as announced in 1945, the RMA recommended an IF of 10.7 MHz, with oscillator high, rather than oscillator low as had previously been the norm for FM receivers. 10.7 MHz ensured that all images were out-of-band, and oscillator-high ensured that the images were above the FM band, and not below it where they would be in the TV channel allocations. 10.7 MHz was in the aero band, 10.2 to 11.3 MHz, which was considered not likely to create an interference problem, particularly as 10.7 MHz would not itself be allocated to a transmitter.
Apparently IFs of 8.25, 8.3 and 8.6 MHz were also used for early receivers for the 88 to 108 MHz FM band. Possibly there was some delay before the RMA number of 10.7 MHz was promulgated, and during that time setmakers had to choose their own IFs. Be that as it may, it does look as if the 10.7 MHz number was arrived during 1946, and so very early in the history of Band II FM. In fact it probably predated the Band II terminology, which I think came out of the 1947 Atlantic City frequency allocation meeting.
In the AM case, there was a migration to higher IFs as improved valve and IF transformer technology allowed, and as better suited “all-wave” receivers. Apparently IFs in use included, but were not limited to 450, 455, 456, 465 and 470 kHz, although many car receivers used 260 kHz. Following this the RMA studied the subject and came up with the standard number of 455 kHz. Although not stated in the Radio News articles, I imagine that this happened in the later 1930s. It was noted that it was not possible to have a below-band IF that completely avoided in-band images, and that 455 kHz was in a marine band, so that interference from marine transmissions was possible.
The foregoing generally confirms what has gone before in this thread, with some elaboration.
I’ll also add a few comments to the TV IF thread based upon those Radio News articles.
Cheers,
Steve
A refreshingly straightforward explanation then for several of the FM IFs! (including 10.7MHz), i.e. that they were slightly more than half the span of the "currently mooted" band, allowing for the image to be out-of-band.
Synchrodyne's mention of the Heathkit Mohawk on another forum (re. its usage of the 6CS6 (EH90) as first and second mixer, rather than the typically used 6BE6 (EK90)) prompted a quick search, revealing that it used another curiously precise 1st IF: 1.682MHz. This doesn't ring any bells here as allowing for any immediately obvious 2nd LO crystals (50kHz 2nd IF) and, moreover, it used sideband selection by switching between 1.632 and 1.732MHz types for 2nd conversion- any ideas, anyone? 
There were a miscellany of receivers that resorted to IFs in the region of 1.6-1.7MHz, hoping to strike a compromise between image rejection and selectivity in the higher HF regions without having to resort to four- (or more) gang tuning with its expense and complexity and consequent receiver size and expense by using a relatively quiet spot between the top of the MW band and the commercial/amateur allocations (though there is an account by a member on another forum of a Star amateur bands rceiver with IF of 1650kHz being blighted by a nearby shore station on just this frequency). Eddystone's 640 was probably one of the earlier examples made in significant numbers and featured one RF stage and two 1.6MHz IF stages, probably representing a good performer for the keen SWL/amateur of the time. There were murmurs of "barn-door" selectivity, though. The subsequent 750 model effectively represents (at least electronically) a revisiting of the design with miniature valves and a re-appraisal of the weak points- a separate HF oscillator valve was now used, together with a shunt voltage regulator and what would have been the 6BA6 1st 1.62MHz IF amp was now an ECH42 converting to 2nd IF of 85kHz. Perhaps Eddystone were slightly stung by criticism of the 640's selectivity shortcoming and there was an element of "we'll show 'em" in choosing a second IF all the way down at 85kHz, hopefully now giving good selectivity with just two IFTs at this frequency (four tuned circuits). Certainly, Eddystone dispensed with the 1.6MHz crystal filter that had featured in the 640, claiming that the variable-coupling 85kHz IFTs in the 750 would be sufficient for CW reception.
This 85kHz is another IF that featured in an eclectic selection of receivers, the BC453 being perhaps one of the best-known, it also featured in a number of Marconi designs. Not sure of the origins of this choice- I'm aware of early superhets that used in the region of 110-130kHz, perhaps some of the pre-war gurus here might shed some light. Possibly, it was too low for "fidelity" bandwidths and was more of a "communications" design choice. As well as the Drake receiver mentioned upstream, the previously mentioned Mohawk featured 50kHz IF- again, this looks like an attempt to realise both extreme and relatively easily variable selectivity without excess expense or complexity. With respect to the 1.682MHz/50kHz ratio, perhaps it's unsurprising that the Mohawk featured two double-tuned IFTs at 1st IF to keep 2nd IF image suppression acceptable. A 6BA6 sat between them, operating at low gain with 1k cathode and 68k screen-grid resistors- presumably not much more than was necessary to overcome IFT loss and prevent the second 6CS6 multigrid from detracting significantly from noise performance.
Unlike the case of the Heathkit GR-78, for whose 4.034 MHz 1st IF one may derive a plausible rationale, the 1.682 MHz 1st IF of the Heathkit Mohawk defies post facto analysis, at least in its precision.
Similarly not obvious are the reasons for choosing 85 kHz as a final IF, as compared with say 50 or 100 kHz. A working hypothesis might be that a 2nd IF as low as reasonably possible was required, but that calculation of the possible spurs in a given situation indicated that around 85 kHz was the lower limit. Spurs aside, I am not sure what would have been the practicable lower limit for an IF strip back in the days of valve technology, but I imagine that 50 kHz was getting close to it.
An interesting commentary on the relative merits of the circa 1.6 MHz IF are provided in Amalgamated Wireless Valve (AWV) “Radiotronics” #145 of 1950, October. This is available at: http://frank.pocnet.net/other/AWV_Radiotronics/ . It does not though provide precise IF recommendations within the 1.6 MHz range.
AWV Radiotronics #119 of 1946 May helps establish a timeline for the American FM receiver IF of 10.7 MHz. On page 54, it is noted: “It has been proposed by the R.M.A. Standards Committee that an intermediate frequency of 10.7 Mc/s be adopted for use with v-h-f receivers.” Allowing that this happened a few months ahead of publication date, then the RMA worked fairly fast, as the decision to move FM up to the 88 to 108 MHz band was announced mid-1945.
Also from AWV Radiotronics, #117 of 1946 January, records that in Australia, 455 kHz had been kept as a clear channel for use as a radio receiver intermediate frequency. Evidently Australia had followed US practice, not so surprising as at the time, it also used 10 kHz channelling for the MF broadcast band.
Cheers,
Steve
Also worth recording here is the renaissance of the 450 kHz for domestic AM (MF and MF/LF) receivers, including the AM sections of hi-fi tuners. I am not sure exactly when or why this happened, but I think that it was in connection with the introduction of electronic tuning, particularly for car radios. So that would have made the transition period late 1970s and early 1980s. Given that 450 kHz, being a multiple of 10 kHz, was probably a less desirable IF than 455 kHz in those areas that had 10 kHz MF channelling, one assumes that it was sufficiently a preferable number for use with electronic tuning systems (at least in their early incarnations) that the change was justified. Thus the filter industry made available an appropriate range of ceramic 450 kHz IF bandpass filters and the like.
Of course, 450 kHz as an AM receiver IF was not new. As noted upthread, Eddystone had used it for some of its valve-era single-conversion HF receivers, although it had also used the standard 455 kHz for some of its later solid-state single-conversion designs. And 450 kHz was one of the several IFs that had been in use in North America prior to the standardization on 455 kHz.
The widespread introduction of 450 kHz as an AM receiver IF did not seem to carryover to HF receivers though, where 455 kHz remained the norm, at least where a number in that general vicinity was required. Two mid-1980s Sony portable receivers provide examples. The SRF-A100 was a small FM stereo/AM stereo model, and had a 450 kHz AM IF, even though it was conventionally tuned not electronically tuned. The well-known ICF-2001D/2010 HF receiver was double-conversion on AM, with a 455 kHz second IF. That was notwithstanding the fact that for AM synchronous demodulation, it used the same stereo AM IC set as was deployed in the SRF-A100. Thus for the included PLL, which operated at 8x IF, the SRF-A100 had a 3.6 MHz oscillator crystal, whereas the ICF-2001D/2010 had a 3.64 MHz crystal.
450 kHz may have been seen as something of a norm by the time AM stereo arrived in the USA and elsewhere. The Motorola MC13020 was a very early C-QUAM stereo decoding IC, and the datasheet quotes 450 kHz as a typical IF, which in turn required a 3.6 MHz PLL oscillator crystal.
Interestingly, the datasheet for the Motorola MC13030 AM dual-conversion IC mentioned upthread referred to both 450 and 455 kHz for the 2nd IF. Other AM receiver IC datasheets provide some additional clues about the transition to 450 kHz. That for the well-known Hitachi HA1197 refers to a 455 kHz IF. The Sanyo clone of the HA1197 was the LA1240. The Sanyo LA1245, presumably a later development, accommodated electronic tuning, but still referred to a 455 kHz IF. But the LA1247 data sheet referred to a 450 kHz IF. National had also entered the electronically-tuned AM radio IC field in the early 1980s with its LM1863, and the corresponding data sheet mentioned a 450 kHz IF.
Cheers,
Steve
I had assumed (and I'll readily admit that I haven't exhaustively researched the subject, either in its practical implementation or its underlying maths) that, post-WARC '78 (?), 450kHz represented a convenient common multiple in both 9kHz and 10kHz channeliised broadcast receivers with 900kHz and 1350kHz IF harmonics zero-beating (or at least, given production tolerances and drift, not whistling too obtrusively) with allocated channels in both systems.. I've also encountered 459kHz and 468kHz ceramic filters in synthesized car radios and domestic tuners, sometimes the instructions will make reference to this being the IF for "European" market version, evidently with 9kHz spacing in mind. Presumably, these offered sufficient advantage over 450kHz filters for the MW/LW spectrum as it existed in this region to make it worth producing non-"universal" front end/IF strip manufacturing lines.
As for the continuing usage of 455kHz in HF receivers, maybe 5kHz channel spacing and its IF harmonic implications/other possible spur products were in mind and the existing 455kHz product array was "good enough". Or could filter manufacturers have been reluctant to produce a 450kHz line of high-quality multi-element communications-spec filters when there was already a range of such items long established at 455kHz? Such a market at any frequency is probably much smaller than that for simpler filters for less demanding mass-market consumer broadcast receivers and choosing/sticking to 455kHz perhaps made it easier to offer selectivity options at "enthusiast" level spec and above- and anyone who more than casually listened to HF broadcasting soon becomes "enthusiast" at least as far as an awareness of the need for half-decent selectivity is concerned!
As for Eddystone's choice of 450kHz, it is a bit off the by-then-accepted "beaten track" of 455kHz. One thing that occurred to me is that they typically liked to offer MW coverage down to 480kHz, rather than the usual "broadcast" 520-530 or so kHz of others. (Not sure why, possibly they liked to pitch their receivers at professional, as well as well-heeled domestic usage, and this would have offered watch capability at 500kHz and other marine frequencies.) Tuning IFTs down to 450kHz might have offered a bit more safeguarding against IF breakthrough or other spurs resulting from input RF and high level IF getting too close in frequency. I know from my own tinkering with a CR100 (IF 465kHz, input RF coverage gap between 420kHz and 500kHz) that slight mis-tweaking of oscillator frequency can produce a shriek in this region.
