Notifications
Clear all
Radio
169
Posts
10
Users
39
Reactions
72.3 K
Views
Topic starter
In the preceding post I said: “In fact, electronic tuning had been used for AM receivers since the early 1980s at least,…”
I thought that I had better recheck the timing of this change, and in the process of so doing I found what was a significant marker in respect of car radio receivers. The three major US car makers of the period, Ford, GM and Chrysler all displayed electronically tuned AM-FM car radios at the 1978 February-March SAE Congress meeting in Detroit, MI. And all three presented subject papers at that meeting (*).
These papers collectively give some insights into the development of AM IFs.
The Ford paper presented a detailed comparison, weight, size, complexity, etc., of mechanical and electronic AM tuners. In respect of the mechanical tuner, it was said “The oscillator tank circuit…is tuned to a frequency 262.5 kHz (or 455 kHz) higher than the signal frequency”. This aligned with the then standard AM IFs used in the USA, the 262.5 kHz number mostly applicable to car radios, and the 455 kHz number for general use.
Ford did not quote numbers for the electronic AM tuner case, but as the was no prescaler between the local oscillator and the synthesizer programmable divider, it presumably followed the same pattern as the GM (Delco) and Chrysler units.
Both GM and Chrysler papers showed electronic AM tuners that had a 260 kHz IF. Both had a local oscillator range of 800 to 1860 kHz, in 10 kHz steps, corresponding to a tuning range of 540 to 1600 kHz. The variable divider had a divisor range from 80 to 186, to provide a 10 kHz input to the PLL. Chrysler made the comment: “Note that if the conventional AM IF of 262.5 kHz had been used, the system would have had 100 times more phase noise and additional cost because of the added complexity of the extra digits.”
Thus the advent of electronic tuning, at least in its early form within car radio economics, effectively forced adjustment of the IF to be a multiple of the channel spacing, 10 kHz in the US case. Had the target IF been 455 kHz rather than 262.5 kHz, then one may see that it would have to have been adjusted to 450 kHz. The high stability conferred by synthesized tuning meant that IF 2nd harmonic interference was unlikely to be a problem.
With FM, where the LO prescalers (dividers) were used ahead of the variable divider, there was no problem with retaining the standard 10.7 MHz IF.
In summary, a step change in tuning technology was one driver of a change in AM IFs to be integral multiples of the channel spacing, 260 and 450 kHz in the US case. Then as synthesizer technology improved, smaller tuning steps, much smaller than the channel spacing, became possible without significant noise or cost penalties.
That also ties in with the 1978 European change to MW channels that were integral multiples of 9 kHz, which would have made it easier to deploy synthesized tuning than with the previous set, which were non-integral multiples and had a mix of 9 and 8 kHz spacing.
(*) The three papers were:
SAE 780004 “The Future of Digitally Tuned Automotive Radios:, by R.L. Summerville and D.G. Ballentine, Ford Motor Company.
SAE 780005 “Signal Seeking, Digital Display, Clock Radio/Tape with Electronic Tuner”, by R.W. Shimanek, Delco Electronics Division of GMC.
SAE 780006 “Development of an Electronic Search Tune Radio”, by John M. Klobuchar, Chrysler Corporation.
Cheers,
Steve
Posted : 07/04/2023 8:58 am
Nuvistor reacted
Topic starter
Continuing from the preceding post, this diagram of the GM-Delco PLL shows the relatively simplicity, mathematically speaking, conferred by the use of the 260 kHz IF.
For the 450 kHz IF case, the variable divider would need to have a divisor range of 99 to 205, which would appear to have been quite an easy proposition. But for 455 kHz, and even more so for 262.5 kHz, the required “mathematics” gets more complicated.
For 9 kHz channelling, with frequencies from 531 to 1602 kHz, the 450 kHz IF could have been accommodated with a divisor range of 109 to 228 and a 9 kHz reference frequency. Possibly a single PLL system, with a 450 kHz IF, could have been switched between 9 and 10 kHz channelling by suitably programming the variable divider and using a 3.6 MHz crystal, which could be divided down to either 9 kHz (divisor 400) or 10 kHz (divisor 360).
One may also observe that the delay in shifting the European LW channels to integral multiples of 9 kHz may have slowed the adoption of electronic tuning there, or caused some makers to ignore the LW band. Of course, the steady improvement in synthesizer performance/cost ratio meant that it was probably not too long before say a 1 Hz reference became feasible for domestic equipment.
The Ford SAE paper also indicated that the introduction of synthesized tuning for car and domestic AM receivers was determined not so much by the availability of suitable PLL devices and circuits, but awaited the availability of satisfactory varactor diodes with a 3-to-1 tuning ratio.
That also required changes in car radio AM front end design. The first varactor tuned stage needed to be buffered by a wideband, agc’d RF amplifier, both to prevent large signals from affecting varactor tuning and to avoid any severe mismatching over the band. Variations on the jfet/bipolar cascode were typically used for this purpose, and also sometimes used in domestic receivers. And ICs, such as the National LM1863, were developed to work with such front ends. So there were multiple changes concurrent with the adoption of the 260 and 450 kHz IFs, the latter being more of an effect than a cause.
In contrast, with FM, with its relatively small 1.23 tuning ratio across the band, there was no problem in having the aerial feed directly a varactor tuned circuit, so existing front end techniques could be retained with the introduction of synthesized tuning.
Cheers,
Steve
Posted : 08/04/2023 10:42 pm
Topic starter
The ICOM R-7000 VHF-UHF communications receiver has been mentioned previously in this thread, but I have since taken a closer look at its IF structure.
It had frequency coverage 25 to 1000 MHz, with an onboard converter to provide the 1025 to 2000 MHz range.
To recap, and neglecting for the time being the converter, it was a double/triple conversion superhet with the following IF structure:
1st IF for the range 25 to 520 MHz was 778.7 MHz
1st IF for the range 512 to 1000 MHz was 266.7 MHz.
2nd IF was 10.7 MHz, and this was the final IF for wideband FM.
3rd IF was 455 kHz, and this was used for AM, SSB and NBFM.
The 2nd and 3rd IFs were industry standard numbers, so do not require further detailed analysis. The conversion from 10.7 MHz to 455 kHz was infradyne with an oscillator frequency of 10.245 MHz, which I think was the modal choice for this particular frequency change.
Returning to the first conversion, the local oscillator frequencies were as follows:
25 to 520 MHz band: 803.7 to 1290.7 MHz
512 to 1000 MHz band 778.7 to 1266.7 MHz
Clearly, the 1st IF pair was chosen to allow essentially the same local oscillator range to be used for each, presumably somewhat simplifying the synthesizer design. This resulted in an initial upconversion for the lower band, and downconversion for the upper band.
Preferably, each band would be about the same size in frequency span terms, in order to make best use of the oscillator range. Then the separation between the two IFs would correspond numerically more-or-less with the dividing point between the two bands, Here the separation was 512 MHz.
We may look at an arbitrary example. Let’s take tuning ranges of 0 to 500 MHz and 500 to 1000 MHz, with an “upper” IF of 750 MHz. Then the oscillator range for the lower band is 750 to 1250 MHz. Applied to the upper band, this gives a lower IF of 250 MHz. The two IFs are separated by 500 MHz. The boundary for IF selection here is probably the lower IF, which would need to be high enough to provide adequate image rejection got the upper range. Also, it would probably be preferable to keep it away from possible strong interfering signals, which would indicate that it should be above the top of TV Band III (variously 216 to 254 MHz according to geography).
Introducing some asymmetry in respect of the two receiving bands gives a limited amount of flexibility. If the lower range were 0 to 520 MHz, then for a 750 MHz upper IF, the oscillator range would be 750 to 1270 MHz. With the upper range staying the same at 500 to 1000 MHz, then the oscillator range for that band could be anywhere between 750 to 1250 MHz and 770 to 1270 MHz, giving an IF range of 250 to 270 MHz. So there was some margin for adjusting the relationship between the lower and upper IFs.
In the ICOM R7000 case, the two bands at interest had slightly different widths, 495 MHz lower and 488 MHz higher, and their respective oscillator ranges were offset, that for the lower band starting 5 MHz higher than that for the higher band. That might have been done to get the lower and higher IFs exactly where ICOM wanted them to be.
It is evident that the actual IFs that ICOM used, 266.7 and 778.7 MHz, had an upward 10.7 MHz offset built-in, to allow infradyne conversion to 10.7 MHz with integer frequencies. It would be easy to infer that that choice was made because integer frequencies were easier from the viewpoint of synthesizer design. The actual 2nd oscillator frequencies were 256 and 768 MHz. The latter was exactly three times the latter, unlikely to be coincidental, which further supports the ease-of-synthesizer-design notion. One may then deduce that ICOM “juggled” the range offsets relative to the whole 1st local oscillator span to ensure that it obtained an integer ratio of the two 2nd oscillator frequencies whilst obtaining sane lower and upper IF numbers.
A look into the block schematic shows that both of these frequencies were obtained from a 51.2 MHz crystal oscillator. An initial x5 multiplication gave 256 MHz, and then x3 gave 768 MHz. One might have expected a start from say 32 or 64 MHz. But the 51.2 MHz number was also used in three different ways as inputs to the main synthesizer. That latter had a 12 MHz main crystal oscillator.
So we could infer that the 51.2 MHz number was chosen more-or-less to suit the main synthesizer, but as it had to multiply up to the required second conversion integer frequencies, it was somewhat constrained. And perhaps some details of the synthesizer had to be adjusted accordingly.
The previously mentioned converter, not connected to the main synthesizer, simply used a 1000 MHz local oscillator to convert the 1025 to 2000 MHz range to 25 to 1000 MHz for handling by the receiver proper.
As always with this kind of deductive exercise, there is no assurance that one has reached the correct conclusions. But if nothing else, there is a reasonably plausible rationale for those 266.7 and 778.7 MHz 1st IF numbers.
Cheers,
Steve
Posted : 07/12/2023 4:56 am
Doz and turretslug reacted
Topic starter
Upthread (1), in discussing the Rohde & Schwarz (R&S) EK-070 (VLF-HF) and ESM-1000 (VHF-UHF) receivers, I said:
>>>>>>>>>>
Having postulated the “2.1 x” rule [as applied to the EK-070], R&S then had to explain why they did not adhere to it in their ESM-1000 VHF-UHF receiver,
“The IF selection in this receiver does not follow the rule of thumb given for the HF receiver. A first IF of 810.7 MHz is used because of filter availability. At this frequency, a surface acoustic wave (SAW) filter can be obtained with low insertion loss and high selectivity (about 2.0 MHz). The ultimate rejection of such a filter, with 4-dB insertion loss and less than 1-dB ripple in the passband is about 80 dB. The preferred choice of a first IF, above 2500 MHz, would have resulted in a filter with too wide a bandwidth and excessive insertion loss. The second IF is 10.7 MHz. A drawback of the first IF frequency selection is that at IF/2 (405.35 MHz) there is a frequency window where the balanced mixer shows poor image rejection. This band is about ± 500 kHz wide since the first IF bandwidth has about 2.0-MHz bandwidth. This design indicates that the availability of components sometimes limits design so that general rules must be abandoned and a performance compromise accepted.”
This receiver tuned the range 20 to 1000 MHz. Not explained though were the consequences of having an in-band 1st IF, namely 810.7 MHz. I think it may be assumed that the 1st mixer had excellent isolation between its input and output ports.
>>>>>>>>>>
The above quote was taken from the book: “Communications Receivers Principles and Design”, by Ulrich L. Rohde and T.T.N. Bucher; 1988 edition.
More recently I have found some additional information (R&S service data) on the ESM-1000 series receivers that suggests that the foregoing was not the complete story.
In fact the ESM-1000 series models had two separate front ends. One covered the range 20 to 500 MHz, and produced the previously stated 810.7 MHz 1st IF using a local oscillator frequency range of 830.7 to 1310.7 MHz.
The other front end covered the range 500 to 1000 MHz. In this case the local oscillator range was 810.7 to 1310.7 MHz, which produced a 1st IF of 310.7 MHz.
Essentially the same oscillator range was used for both front ends, except that the initial 20 MHz was not used for the 20 to 500 MHz range.
With a 1310.7 MHz local oscillator frequency used for upconversion of 500 MHz and downconversion of 1000 MHz, then the two 1st IFs were 500 MHz apart.
Thus, the R&S first conversion approach was, in frequency terms, broadly similar to that used by ICOM in its R-7000.
And contrary to what I previously said, the ESM-1000 did not have an in-band IF. When tuned to 810.7 MHz, the 1st IF was 310.7 MHz.
The 2nd IF of 10.7 MHz was obtained by infradyne conversion using synthesized local oscillator frequencies of 300 MHz and 800 MHz.
The 2nd IF of 10.7 MHz was also the final IF for all modes, including NBFM, AM and SSB. In the professional bracket, excellent 10.7 MHz filters were available for all purposes including relatively narrow band modes, so that another conversion down to say 455 kHz was not necessary, perhaps unlike the consumer level case, where cost considerations may have indicated the use of lower cost 455 kHz filters for the narrowest band modes, as in the ICOM R-7000.
Cheers,
Steve
Posted : 09/12/2023 3:52 am
Nuvistor reacted
Topic starter
The ICOM IC-R9000 was more-or-less the successor to the IC-R7000. A major change was that it added HF band coverage, from 0.1 to 30 MHz, as well as providing VHF-UHF coverage from 30 to 2000 MHz. That made it unusual, and so worth a “dissection” of its IF structure, including comparison with the R7000.
The R9000 had separate HF and VHF-UHF front ends.
Looking first at the VHF-UHF side, it had a broadly similar IF architecture to the R7000, but there were some differences.
As with the R7000, the R9000 had a converter to cover the range above 1000 MHz. But whereas the R7000 used a single-frequency1000 MHz oscillator to translate the 1025 to 2000 MHz band down to 25 to 1000 MHz, leaving a 1000 to 1025 MHz gap in coverage, in contrast, the ICOM R9000 had a two-frequency converter oscillator. One frequency, 900 MHz, was used to convert 1000 to 1150 MHz to 100 to 250 MHz. The other, 1000 MHz, was used to convert 1150 to 2000 MHz to 150 to 1000 MHz. Thus there was full coverage from 1000 to 2000 MHz.
Given the existence of the HF section covering 0.1 to 30 MHz, VHF-UHF coverage was from 30 to 1000 MHz as compared with 25 to 1000 MHz for the R7000. As with the latter, the same approach was used, with initial upconversion for the lower “half” of that band, and initial downconversion for the upper “half”.
The R7000 tuning stepped in 100 Hz increments, provide by the 1st LO synthesizer. The R9000 though tuned in 10 Hz increments across the whole range, including the VHF-UHF band. For VHF-UHF, this was achieved with a 100 kHz stepping of the 1st LO, and using the 2nd LO to interpolate the 10 Hz steps. Thus the 1st IFs were not single frequencies, but varied very slightly over a nominal 100 kHz range. In turn the 2nd LOs varied over the same range. Thus unlike the R7000 case, there was no possibility of having the two 2nd LOs share an integer common factor. In turn there was no need to juggle the two tuned frequency ranges and their respective IFs to obtain that relationship. That being the case, the lower and upper bands were simply 30 to 500 MHz and 500 to 1000 MHz respectively, and the two IFs were separated by 500 MHz. The lower band 1st IF was 778.7 MHz nominal, as was used in the R7000, whereas the upper band 1st IF was (778.7 – 500) = 278.7 MHz nominal, as compared with 266.7 MHz in the R7000.
Taking account of the 100 kHz variation, the two 1st IF ranges were 778.6 to 778.7 MHz and 278.6 to 278.7 MHz. Both of these1st IFs were downconverted to 10.7 MHz, using 2nd LO frequencies of 767.9 to 768.0 and 267.9 to 268.0 MHz.
One may speculate that once the 10 Hz tuning step became a requirement, it was necessary to use 2nd conversion interpolation in order to keep the 1st oscillator PLL complexity and cost within bounds. And once that pathway had been chosen, it was logical to use a coarser step for the 1st PLL, which would likely have facilitated obtaining better performance, including lower phase noise, at a given cost.
The 10.7 MHz 2nd IF was the same as for the R7000. This was also the meeting point of the HF and VHF-UHF sections. So before considering that, it is necessary to look at the HF section.
Here there was initial upconversion to a nominal 1st IF of 48.8 MHz. This was in and of itself an unremarkable number in the field of upconversion HF receivers, but it was different to the 70.4515 MHz that ICOM had used for its then-current R71 and earlier R70 HF receivers. In those cases, it is reasonable to assume that ICOM had selected the 70 MHz vicinity as an optimum choice, with the actual 70.4515 MHz number determined by synthesizer and subsequent conversion considerations. Why then was 48.8 MHz used in the R9000?
The answer appears to be that ICOM used the same synthesizer for HF and VHF-UHF. Considering that the local oscillator swing for VHF was 500 MHz, and that a 30 MHz swing was required for HF, then it may be seen that the required HF swing was about one sixteenth of the VHF-UHF swing. The latter was 778.7 to 1278.7 MHz. One sixteenth of that was 46.67 to 79.92 MHz, whose swing of 33.25 MHz was more than was needed. But a 30 MHz (or really 29.9 MHz) block nested within that range would occupy somewhere between 46.67 to 76.57 MHz and 50.02 to 79.92 MHz. The chosen 48.8 MHz was approximately in the centre of that range.
A look at the ICOM R9000 1st LO synthesizer shows that the HF and VHF-UHF feeds were split after the penultimate stage. The VHF branch was multiplied by 2 to provide a frequency range of 778.7 to 1278.7 MHz. The HF branch was divided by 8 to provide a 48.89375 to 78.8 MHz range. Thus there was a 16:1 ratio between the VHF and HF 1st LO frequencies,
As the VHF-UHF LO moved in 100 kHz steps, then consequently the HF LO moved in 6.25 kHz steps (one sixteenth of 100 kHz). Interpolation to 10 Hz steps was done by adjustment of the 2nd LO over a narrow band, as in the VHF-UHF case. The HF 2nd LO range was 38.09376 to 38.1 MHz. The nominal 38.1 MHz value converted the 48.8 MHz 1st IF to the 10.7 MHz 2nd IF.
The 2nd LO frequencies for both the VHF-UHF and HF cases were derived from the second PLL, which also provided the 900 and 1000 MHz LO frequencies for the UHF converter.
The 10.7 MHz 2nd IF section branched two ways. Final processing of WFM (wide FM, as in broadcast) was done at this frequency, as it was in the R7000. And a feed at 10.7 MHz was provided to the TV section, again as it was in the R7000.
The second 10.7 MHz branch served all other modes, namely AM, SSB, FSK, CW and NBFM. ICOM identified three bandwidth variations for NBFM, namely FM-n (narrow), FM-m (middle) and FM-w (wide). For all of these except FM-w, there was firstly a noise blanker circuit, operated by a 10.7 MHz sidechain. This though was bypassed for FM-w.
Next came conversion to the 3rd IF of 455 kHz. The primary bandpass filtering for the various modes was done at this frequency, as was final processing of NBFM. For the other modes though, there was a further conversion to a 4th IF of 10.7 MHz using the same nominally 10.245 MHz oscillator (referred to as the shift oscillator) as was used to downconvert from 10.7 MHz. This was done to provide a passband shift facility by varying the shift oscillator above and below its nominal frequency.
Final processing of AM, SSB, FSK and CW was done at the 10.7 MHz 4th IF frequency, which section also included a variable notch filter.
Thus it may be seen that the R9000 differed from the R7000 in having a 4th IF to facilitate passband shifting, and in doing the final processing for AM, SSB, FSK and CW at that frequency (10.7 MHz) rather than at 455 kHz. Nonetheless, the main bandpass filtering for each mode was done at 455 kHz, consistent with consumer-level receiver general practice, and as was done in the R7000.
In this regard, the 2nd/3rd/4th IF scheme generally followed ICOM’s established practice for HF receivers. In that R71, the 2nd and 4th IFs were 9.0115 MHz, the 3rd being 455 kHz. NBFM final processing was done at 455 kHz, with AM, SSB, etc, done at 9.0115 MHz. For the R9000 case, and evidently in deference to established VHF-UHF receiver practice, ICOM used 10.7 MHz rather than 9.0115 MHz as the 2nd and 4th IFs.
The TV unit was broadly similar to that in the R7000, although it was an integral onboard unit rather than an attachable outboard unit. The input signal was 10.7 MHz SIF, and either 15.2 MHz VIF for systems M/N, or 16.2 MHz VIF for systems B/G/H. The circuitry was configured to handle just one of those two choices, according to the individual receiver intended geographic destination. Within the TV unit, the 10.7 MHz SIF was suppressed, and the VIF demodulated in a conventional quasi-synchronous processor (Sanyo LA7530 IC), albeit that the intercarrier part of that IC was note used. Any residual intercarrier (4.5 MHz for systems M/N, 5.5 MHz for systems B/G/H) in the video was suppressed. Thus the unit provided only a video output for external use or internal use by the LCD display. Sound came from the 10.7 MHz WFM channel to the line output. Thus the R9000 necessarily operated in split sound mode.
In contrast, the R7000 TV processor (TVR-7000) accepted SIF (10.7 MHz) as well as VIF. Processing was also done in an LA7530, but in this case the intercarrier was used and demodulated to provide an audio feed. Thus the TVR-7000 had both video and audio outputs, with the sound being of the intercarrier type. But split sound was also available (from the 10.7 MHz WFM IF subsection) from the line output on the R7000. It is possible that the duplication in this case came about because it was thought that the TVR-7000, as a separate unit, should have both video and audio outputs. That the R7000/TVR-7000 combination thus offered a choice of split and intercarrier sound (in the manner of US market component TV tuners of the time) was probably incidental rather than intentional.
In summary, the R9000 used a complex of IFs. The three 1st IFs were ad hoc. But the two for VHF-UHF, although differing in detail, conformed to the general pattern that had prior use in the R&S ESM-1000 and ICOM R7000, and possibly elsewhere. In the upconversion HF receiver case, as previously noted, 1st IF selections appear to be quite random, although sometimes it is possible to impute some logic. Here it was simply that the HF 1st LO be an appropriate submultiple of the VHF-UHF 1st LO.
10.7 MHz and 455 kHz were of course very widely used standard numbers. The upconversion from 455 kHz to 10.7 MHz, to facilitate passband tuning, was probably quite rare. More often, with HF receivers, 455 kHz was associated with a different number, such as around 9 MHz, for this purpose. Although by the time the R9000 was released, AM receivers had generally moved across to the 450 kHz IF, the long legacy of and wide filter availability for 455 kHz ensured its continued use in communications receivers. Notwithstanding the final upconversion to 10.7 MHz, the main bandpass filtering for the “narrowband” modes (except NBFM) was done at 455 kHz.
The use of 10.7 MHz for TV sound, as done in the R9000 and R7000 also had precedent that went back well before the R7000. It was used as a 1st IF for TV sound tuners in the 1950s, then as 2nd IF both for double-conversion TV sound tuners and the split sound side of component TV tuners.
Cheers,
Steve
Posted : 17/12/2023 10:04 pm
Topic starter
Much earlier in this thread (1), I said this about the Yaesu FRG-9600 VHF-UHF communications receiver:
>>>>>>>>>>
The Yaesu FRG-9600 tuned continuously from 60 to 905 MHz. It had a 1st IF of 45.754 MHz. This was the American standard television (System M) vision IF of 45.75 MHz, with a small (4 kHz) offset. Thereafter the narrow FM/AM/SSB channel saw conversion to a 2nd IF of 10.7 MHz, with conversion to a 3rd IF of 455 kHz taking place within an MC3357 IC, which also provided narrow FM IF amplification and demodulation. The wide FM channel had conversion to a 2nd IF of 10.7 MHz within an MC3356 IC, which also handled IF amplification and demodulation. The MC3356 was said to be able to handle input frequencies up to 200 MHz, and a 2nd IF up to 50 MHz, although with an expected value of 10.7 MHz. That Yaesu chose an MC3356 for wide FM (at 10.7 MHz) when it used an MC3357 for narrow FM (at 455 kHz) lends some credence to the notion that the MC3357 would not handle 10.7 MHz.
The FRG-9600 had an optional (System M) internal video board that received the 1st IF and demodulated it to baseband video. It differed from the ICOM R7000 in that the video board did not also deliver accompanying audio; presumably the wide FM channel was used for that. Thus in IF terms, the sound carrier of a tuned-in TV channel would have been at 45.754 MHz, with the vision carrier thus at 50.254 MHz, making it non-standard. Hence it would appear that the 45.754 MHz 1st IF was not chosen because of the TV reception option, but for other reasons.
>>>>>>>>>>
Since I wrote that, I have found additional information that nullifies my earlier closing statement, above.
Evidently, the front end of the Yaesu FRG-9600 was based upon a standard VHF-UHF TV tuner.
The tuning ranges were:
VHF-1 60 to 107 MHz
VHF-2 107 to 230 MHz
VHF-3 230 to 460 MHz
UHF 460 to 905 MHz
These numbers suggest that it was an American market, “cable-ready” tuner with three VHF bands. The standard IF (vision) for that type of tuner was 45.75 MHz. The cable TV channels, as well as overlapping the broadcast channels, by the mid-1980s fully occupied the intervals between the designated VHF and UHF bands, hence the continuous coverage.
Thus one could say that with the approach it adopted, Yaesu was in fact tied to the 45.75 MHz number, or something close thereto, unless it undertook modification of the TV tuner that it used. As noted, it included a slight offset, to 45.754 MHz. Possibly this was done to suit the synthesizer that it used.
Obviously, incoming radio signals were tuned so that they were converted to the 45.754 MHz. Incoming TV signals were also tuned to align the sound carrier, rather than the vision carrier, with 45.754 MHz, by which means the sound channel could be accessed by the “wide FM’ circuitry. That put the vision carrier (systems M/N) at 50.254 MHz, which meant that the TV IF channel was non-standard, being 4.54 MHz higher than usual, nominally occupying 45.504 to 51.504 MHz.
The bottom end of the VHF tuning range was truncated to 60 MHz, rather than the 54 MHz that the tuner itself would almost certainly have allowed, the American channel A2 occupying 54 to 60 MHz. Thus there was some clearance between channel A2 and the standard 41 to 47 MHz IF channel. Perhaps the 45.504 to 51.504 MHz IF channel actually used was too close to 54 MHz, and the lowest tunable frequency was moved up to 60 MHz, the bottom edge of TV channel A3, as a consequence. There was probably little point in stopping somewhere between 54 and 60 MHz.
Anyway, it may be said with certainty that the Yaesu FRG-9600 1st IF of 45.754 MHz was directly derived from the American standard TV VIF of 45.75 MHz. Thus it was a minor variation of a standard number, probably within the range of normal variation for standard IF filter systems.
TV vision was processed at 50.254 MHz in a standard quasi-synchronous IC (Sanyo LA7507), with a trap on the output side to suppress any residual 4.5 MHz intercarrier. TV sound was processed via the 45.754-to-10.7 MHz wide FM pathway. Thus the FRG-9600 operated in the split sound mode.
I have not seen anything to suggests that the FRG-9600 was offered with an alternative TV module for European use, unlike say the ICOM R7000 and R9000 cases where an alternative TV unit was available for systems B/G/H. If it were done, then the 45.754 SIF would imply a 51.254 VIF for those systems. The 60 MHz bottom edge of the tuning range would have excluded European channels E3 (47-54 MHz) and E3 (54-61 MHz), although the E3 sound carrier at 60.75 MHz would have been accessible.
Be that as it may, it may be seen that vision aside, the three IFs used in the FRG-9600, namely 45.754 MHz, 10.7 MHz and 455 kHz were all standard numbers.
Cheers,
Steve
Posted : 19/12/2023 12:53 am
Topic starter
Upthread (1), I mentioned that there were Japanese domestic receivers and tuner-amplifiers (and possibly tuners as well) that had two AM sections in order to receive the experimental two-transmitter AM stereo broadcasts. Apparently such broadcasts were regular Weekly events over about a decade, and dual-AM section receivers were a standard production item in Japan. For example, Sansui made several such models, culminating in the SM-80.
Given the standard production status, it seems more probable than possible that there was a standard, approach to selecting the IFs for such units, whether formalized by, for example JIS, or simply an informal customary practice. Thus far, I have not been able to find any information on the IFs used. For example, the findable Sansui schematics do not note the IFs.
The typical configuration seems to be that the primary AM tuner was two-band (MW and SW) and shared its IF valve with the FM tuner. Presumably this would have used the then-standard 455 kHz IF. The secondary AM tuner was single-band (MW) and standalone, and presumably had a different (probably lower) IF. Both AM tuners were of the variable bandwidth type, allowing wideband reception in suitable conditions.
Thus, the Japanese second AM tuner IF is another unknown in this sequence, and not simply a “one-off” curiosity, but likely part of a pattern, albeit a sidebar, in the AM IF history. As said in the earlier post, 260 Hz or thereabouts seems to be a reasonable possibility, but that remains as an unsupported guess.
Cheers,
Steve
Posted : 19/12/2023 1:21 am
Topic starter
As mentioned upthread, in respect of professional HF receivers, the synthesized upconversion topology largely displaced the earlier Wadley Loop upconversion type in the late 1960s and early 1970s for new designs. The Racal RA1772 of c.1972 was an early and important marker in this process. The newer technology was also entering the upper end of consumer-type HF receivers by 1976-77. Nonetheless, Yaesu’s 1976 entry to the general coverage receiver market was with the FRG-7, which used a Wadley Loop. Presumably this gave the best performance vs. cost tradeoff at the Yaesu’s target price point.
With the FRG-7, Yaesu followed the established Wadley Loop pattern, but used a significantly higher 1st IF, 55 MHz centre, 54.5 to 55.5 MHz range, than the 40 MHz (centre) used for the Racal RA-17. This probably eased design in what was a relatively economical receiver. The 1st IF was further away from the 30 MHz top of the receiving range, and from the 3 to 32 MHz comb. Also, the 1st IF bandwidth of nominally 1 MHz was a smaller fraction (1.8%) of the centre frequency than was the case at 40 MHz (2.5%), thus easing the design of the 1st IF bandpass filter. With modern solid-state devices, such as dual-gate mosfets, obtaining the requisite gain at 55 MHz was less of an issue than it would have been in the valve and early semiconductor eras. It may be noted that the upward movement in 1st IF had occurred earlier with the Barlow Wadley XCR-30 portable of c.1970. There, the 1st IF was 45 MHz (centre).
The 2nd IF was the same 2 to 3 MHz range (2.5 MHz centre) as had been used for the RA17. In the case of the FRG-7, this required a 52.5 MHz 2nd LO frequency extracted from the mixing of the 1st LO and the comb generator, and subject to narrow band filtering. This one assumes would be more difficult than was the case with the 37.5 MHz used in the RA17, or the 42.5 MHz used in the XCR-30. Yaesu used L-C filters to clean up the 52.5 MHz 2nd LO.
The 1st IF and 2nd LO bandpasses are shown in this semiquantitative diagram.
Although the passbands ranges are shown, the stopband ranges are not.
The 3rd IF was 455 kHz, rather than the 100 kHz used in the RA17, with a 3rd LO of 2455 to 3455 kHz. By this means standard consumer-type IF filters for the various modes could be used. Also, potential images within the 2 to 3 MHz band were further removed from the wanted signal, easing design in that respect. 455 kHz had also been used as the 3rd IF in the XCR-30. The 3rd LO was 2455 to 3455 kHz.
The FRG-7 was joined by the FRG-7700 in 1977. This retained the Wadley Loop form with the same set of IF and LO frequencies, but added a digital frequency readout. However, there was a significant difference in the way the 52.5 MHz 2nd LO signal was processed. After generation in the premixer, it was first downconverted to 10.7 MHz by mixing with a 63.2 MHz signal generated in a crystal oscillator. Then it was passed through a 10.7 MHz ceramic filter, and after that upconverted back to 52.5 MHz using the same 63.2 MHz signal as used to downconvert, thus cancelling any oscillator drift. The ceramic filter was an SFG-10.7MA, evidently a standard FM receiver IF filter with a 280 kHz bandwidth, and a stated selectivity of 70 dB, presumably meaning that the stopband was 70 dB down. A reasonable conclusion is that Yaesu wanted more selectivity in the 52.5 MHz path than it had obtained with the L-C filter in the FRG-7, and had found it more effective and possibly more economical to do the double frequency translation and use an off-the-shelf FM 10.7 MHz than to develop (or buy-in) a suitable high performance 52.5 MHz filter unit. That said, the FRG-7000 was a somewhat higher -priced unit than the FRG-7, so may well have justified a costlier 52.5 MHz filter system. In any event, it was an unusual variation on the Wadley Loop theme, and an unconventional use of the standard 10.7 MHz number, in this case though not as a received signal transport.
In 1981 the FRG-7 and FRG-7000 gave way to the FRG-7700, which was of the synthesized upconversion, double conversion type. By then it was likely that the downward cost and upward performance trends of synthesizer technology had combined to be the optimum for the performance and price point at which Yaesu was aiming.
Cheers,
Steve
Posted : 19/12/2023 7:29 am
turretslug reacted
Topic starter
The FRG-7 was joined by the FRG-7700 in 1977.
That should have read:
"The FRG-7 was joined by the FRG-7000 in 1977.
Cheers,
Steve
Posted : 19/12/2023 8:43 am
turretslug reacted
I gather that the Wadley converter architecture is somewhat prone to increasing spurios with decreasing input signal frequency, so perhaps Yaesu found it necessary to clean up the second injection strip beyond the FRG7's capability- an off-the-shelf consumer FM IF filter would have given good enough out-of-band rejection whilst having sufficient bandwidth to accomodate 1st LO drift at low cost. The RA17's coverage was only down to 500kHz, and it had famously/notoriously complex filters in 1st IF and second injection strips in order to minimize spurii. Racal's solution to LF coverage was an outboard 15-980kHz converter that switched off preselector and Wadley converter HT and used the final 2-3MHz IF with a reversed tuning scale. With the FRG7000 amounting to a "Gran Turismo" edition of the FRG7, perhaps the designers felt that extended coverage was necessary as an additional bullet-point as well as the digital frequency and clock/timer displays, requiring additional converter filtering.
Posted : 19/12/2023 12:42 pm
Racal's solution to LF coverage was an outboard 15-980kHz converter that switched off preselector and Wadley converter HT and used the final 2-3MHz IF with a reversed tuning scale.
That was a bit misleading- the 2-3MHz IF stage of the RA17 isn't the final IF, rather it's the "kHz tune" 2nd IF prior to the 100kHz 3rd and final IF.
Posted : 19/12/2023 8:55 pm
Cathovisor reacted
Topic starter
Thanks for that!
I wonder if the FRG-7000 was the only Wadley Loop receiver that tuned below 500 kHz?
A pertinent question is whether the nominal lower limit of 250 kHz was determined by the drop off in performance as the frequency went down, or by what the preselector could do. Perhaps the latter? Whereas the FRG-7 had four preselector bands and a two-section variable capacitor, the FRG-7000 had five bands and a three-section capacitor. As best I can work out from the schematic, all three sections were used in parallel for the lowest band, so perhaps the maximum available capacitance was the determinant? Otherwise, one might have expected 150 kHz, the lower end of the LW broadcast band, to have been a reasonable target.
Cheers,
Steve
Posted : 20/12/2023 5:06 am
ISTR that Plessey's PR155, a solid-state Wadley converter receiver, was described as having 60kHz to 30MHz coverage (I have also seen other figures such as 100kHz quoted, this might reflect the use of a broadband or low-pass below some frequency front end where minimum useful frequency can become something of a moveable feast). Plessey professional receivers seem to have been an unassuming but respected bunch, so it would be interesting to search the circuit and see what was going on.
The additional 250-500kHz preselector range of the FRG7000 over the FRG7 doesn't immediately come across as a particularly mouth-watering bonus, missing most of the LW broadcasting band. Admittedly, most sales would have been outside Region 1 anyway, but the aeronautical NDBs and limited maritime traffic covered by this additional preselector band must have had restricted, even niche appeal- I think the era of the FRG7000 was before the relevant amateur LF allocation, though I stand to be corrected.
Posted : 20/12/2023 11:25 pm
Topic starter
The Plessey PR155 was advertised as having a 60 kHz to 30 MHz tuning range:
It was of an unusual architecture, perhaps something of a link between the true Wadley loop and the fully synthesized designs. It followed the latter in doing all of the tuning in the first conversion, the first mixer being followed by a 12 kHz bandwidth filter. The first local oscillator was phase locked to a mix of the outputs of a VFO with a 1 MHz range and a suitable 1 MHz harmonic chosen from a comb generated from the output of a 1 MHz master oscillator. The second and third LO signals were also generated from this comb, directly in the second LO case, and by dividing down in the third LO case. The use of a 1 MHz comb followed the Wadley loop design. A more detailed look was provided at: https://www.radios-tv.co.uk/community/postid/107200/.
Perhaps the PR155 could be described as being of the premixer Wadley Loop type?
I imagine that with the phase-locked first oscillator providing a buffer between the comb and the mixer, and the early use of a narrowband IF filter, the PR155 was a lot “cleaner” than a conventional Wadley Loop design, enabling to the downward coverage extension to 60 kHz. The comb output was used directly for the second conversion, and via frequency dividers for the third conversions, but being fixed, as it were, it was possible to use fairly tight bandpass filters to select the desired harmonics.
Cheers,
Steve
Posted : 21/12/2023 1:10 am
Topic starter
This chart shows the IF structures of a (fairly random) selection of HF receivers, both professional and consumer, of the synthesized, upconversion type.
The IF numbers shown are nominal, and in some cases they are subject to minor variation ranges, not shown. The chart could be expanded to show those, as well as tuning ranges and tuning increments for each receiver – but that is a “work in progress”.
Where individual models have been the subject of prior postings, I have included the thread reference.
The wide range of first IFs used is readily apparent. Seldom did two receiver makers use the same number. And some makers used different number sets for different receivers.
Cheers,
Steve
Posted : 23/12/2023 1:12 am
Topic starter
The IF structure of the ICOM IC-R70 and IC-R71 HF receivers attracts attention because of the level of precision associated with the 1st and 2nd IFs:
IF1 70.4515 MHz
IF2 9.0115 MHz
IF3 455 kHz
IF4 9.0115 MHz
The 1st IF of 70.4515 MHz looks as if it could have been the result of a 70 MHz target with an ad hoc offset to suit the synthesizer.
The 2nd IF of 9.0115 MHz looks as if it was a 9 MHz target again with an ad hoc offset to suit the synthesizer.
The conversions to the 3rd IF of 455 kHz and thence to the 4th IF of 9.0115 MHz were done independently of the synthesizer, using the same variable crystal oscillator for both conversions. The variability effected passband tuning by offsetting the 455 kHz IF from the centre of its nominal channel.
Given that the precise values of the 1st and 2nd IFs appear to derive from the synthesizer used, here is a block schematic of that for the R70:
IC2 and IC3 were both of the Toshiba TC9123 type, originally intended for use in AM-FM broadcast receivers, but here pressed into HF receiver service.
As the TC9123 incorporated a 1/512 divider, then it followed that the reference oscillator would be a multiple of 512, and 20.84 MHz was the chosen frequency, divided down to 10.24 MHz for use by the 3rd PLL and again to 5.12 MHz for use by the two TC9123 ICs.
The 20.48 MHz was also multiplied x3 to obtain the 61.44 MHz 2nd LO frequency. The conversion from c.70 MHz to c.9 MHz, infradyne to avoid any conflict with the 1st LO range, required an 2nd LO of c.61 MHz. Thus 61.44 MHz was a reasonable fit, and it determined the exact position of the 2nd IF relative to the 1st.
The main PLL was of the mixdown and divide type, itself providing 100 kHz steps in LO1 frequency. In the mixdown were added both the receiver incremental tuning (RIT) variation range, generated in the 2nd PLL, and the 10 Hz tuning steps, generated in the 3rd PLL and added into the 2nd PLL at the mixdown stage.
The RIT facility was based upon a 10.24 MHz variable crystal oscillator. This frequency looks as if it were chosen to be consistent with the existing range of reference-derived frequencies used in the synthesizer. The oscillator output was the third harmonic (with a variation range) at 30.72 MHz. This suggests that the desired 1st PLL mixdown was in the vicinity of 30 MHz, perhaps just enough above that number to ensure that the 1st LO and mixdown ranges did not overlap. Setting the mixdown range as 39.6 to 69.5 MHz might have been done for that reason. An exact 70 MHz 1st IF would then have implied a 30.5 MHz mixdown frequency. The use of 30.72 MHz would have shifted the 1st IF up slightly to 70.22 MHz.
The mixing in of the 10 Hz increments from the 3rd PLL, after dividing down by 100, provided a small upward offset to the mixdown frequency fed to the 1st PLL. This pushed the 1st IF further upwards to 70.4515 MHz. And from there, subtraction of 61.44 MHz gave the exact 2nd IF of 9.0115 MHz.
Something I have not attempted to rationalize is the choice of frequency numbers used in the 3rd PLL to generate the 10 Hz increments.
Whilst the foregoing has an element of plausibility, I could well be missing some key issues within the “black art” of PLL frequency synthesis.
The ICOM R71, as successor to the R-70, used the exact same set of IFs, albeit without the RIT facility, but with a different PLL system. The lack of the RIT facility allowed the use of just two PLLs. The main PLL used the Mitsubishi M54929P IC, designed for amateur radio applications. This had an internal divide by 1024 facility. The required 10.24 MHz feed was obtained by division by 3 of a 30.72 MHz master oscillator output. The same was doubled to obtain the 61.44 MHz 2nd LO. And the 30.72 MHz was mixed with the 10 Hz increment feed from the 2nd PLL to provide the mixdown input to the 1st PLL. The 2nd PLL had a 5.12 MHz reference input divided down from 10.24 MHz.
From the complex of reference frequencies required, it may be seen that for the R71, 30.72 MHz was a more logical master than the 20.48 MHz of the R70.
An interesting point is that much has been written about the ICOM R71 in particular. It was often mentioned along with the JRC NRD-525 and Kenwood R-5000, which formed an “I-J-K” trio of broadly comparable (although with somewhat differing price points) general coverage HF receivers at the upper end of the consumer market, in the second half of the 1980s, which period seems to have been “halcyon days” for this type of unit. But whilst the IF structures have certainly been mentioned, I have not come across any delving into the whys and wherefores of the exact numbers, even though a value such as 9.0115 MHz surely invites investigation. In respect of IF analysis, the NRD-525 is relatively straightforward, the R71 seems to yield somewhat, as shown above, but the R-5000 remains impenetrable.
Cheers,
Steve
Posted : 23/12/2023 1:24 am
turretslug and Nuvistor reacted
Thanks for that explanation (and reminder of the previous post) regarding the Plessey PR155, it sounds to have taken quite a sophisticated and forward-looking approach for the time. I often hear of people talking of double- (and multiple-) conversion as if it were an unequivocally good thing, but I've long thought that it brings a list of considerable pitfalls with it, including having the main selectivity further down the chain of active devices and the presence of additional oscillators and mixing processes multiplying the odds of spurious responses. Looking at the block diagram of the RA17 provokes the thought "They were lucky to get away with that...." but I gather that considerable development was involved, including the afore-mentioned filters and a strategic saw-cut across a prototype chassis to suppress a prominent spurious response. The Plessey designers evidently felt that keeping too much mixing out of the signal path was wise, commensurate with getting as much selectivity as possible as early as possible.
Yes, the choice of IFs in broad-coverage receivers must be the source of a fair amount of head-scratching and choice of best-fit compromises, with market-position budget also looming. Sometime, I must sit down with pencil, paper and clear head and pan out the maths involved in the relatively modest case of the FRG-100 here, which none-the-less offers considerable synthesis flexibility.
Colin
Posted : 23/12/2023 9:39 pm
Topic starter
Possibly the significance of the Plessey PR155 has been underappreciated. It appears to have been about the first fully tunable HF receiver to combine initial upconversion followed immediately by a relatively narrow IF filter, a principle which was thereafter followed by many manufacturers.
In respect of its RA1772 model, Racal said: “Although it is sometimes beneficial to frequency selection it is never advantageous to the receiver performance if the first i.f. bandwidth is wider than the final output bandwidth”. With its 12 kHz bandwidth initial IF filter, Plessey had hit that nail square upon the head around six years before Racal. That said, a key feature of the RA1772 was its bomb-proof Rafuse mixer, which pretty much did away with the need for RF preselection. Also, it had fully synthesized tuning down to 10 Hz steps, whereas the PR155 was synthesized at 1 MHz intervals with VFO interpolation, The Redifon R550/551, a couple of years later than the PR155, was synthesized at 100 kHz intervals, again with VFO interpolation.
In that period, synthesizer development was on a steep upward curve, with the realizable frequency steps getting smaller and the first IF possibilities getting higher. The Racal R1792 of 1979, reputedly aimed at a noticeably lower price point than the RA1772, was synthesized in 1 Hz steps with a much simpler synthesizer than was used in the RA1772. Possibly there was a trade-off between step size and 1st IF, which might explain why the 40.455 MHz 1st IF number for the RA1792 was not significantly higher than the 35.4 MHz of the RA1772. Plessey had gone to 65.0 MHz with a 10 Hz step size in 1977 with the PR2250, which was probably sold into the same markets as the Racal models. In 1976, Marconi had combined 68.6 MHz with 1 Hz stepping in the H2540, but as far as I know, this was a significantly more expensive unit primarily intended to do most of the traffic tasks in point-to-point receiving stations.
Re the RA17, as you say, Racal put in the effort to make its own luck in a potentially difficult situation, as it were, but it was a finely balanced act. It stayed with the same theme for the solid-state RA217, and there perhaps the “luck” somewhat ran out. The Plessey PR155 may have surprised it, but evidently gave it food for thought, hence the RA1772 which was developed along essentially the same vector, but with significant improvements facilitated by the rapid development of devices and techniques meanwhile.
The ICOM R-70/R-71 IF analysis case certainly involved much head-scratching. Hitherto I had stayed away from looking at what goes inside IC-based synthesizers, treating them as “black boxes”, simply providing LO signals as required for the chosen IF structures. That did not work in this case, so I had to delve into their workings. In the end, after quite a few false starts, I saw that the situation was somewhat like that for the Redifon R550/551, where the mixdown range and mixdown signal were key elements in determining the actual placement of the 1st IF within its target range. Sometimes you think that you have arrived at a neat numerical solution, only to find that it has a logical discontinuity when you come to write out the narrative!
Cheers,
Steve
Posted : 24/12/2023 7:45 am
Topic starter
Having made that comment about narratives in the immediately preceding post, it occurred to me that the narrative for the ICOM IC-R70 IF case could be improved. What follows is an attempt to do just that.
From inspection, the starting givens are assumed to be:
The required tuning range is 0.1 to 30 MHz in 10 Hz steps.
The desired IF1 is in the vicinity of 70 MHz.
The desired IF2 is in the vicinity of 9 MHz, with infradyne conversion from IF1.
The chosen IC used for PLL1 and PLL3 has an internal divide by 512 function in its reference channel.
The PLL mixdown signals and LO2 are to be simply related to the master oscillator frequency for minimum deleterious spur generation.
Although it is difficult to be certain about cause and effect in this case, a convenient starting point for analysis is the LO1 output fed back into PLL1. Note that for the time being, it is neglected that LO1 comprises four VCOs, each covering just a part of the tuning range. Rather it will be treated as if it were a single VFO.
The LO1 output from VCO1 is controlled by PLL1. For feeding into this PLL, it is first mixed down to the range 39.6 to 69.5 MHz using a mixdown signal derived from VCO2, about which more below. Then it [the mixed down VCO1 output] is divided by 10 to obtain the range 3.96 to 6.95 MHz. Next comes a variable divider (divisor 396 to 395) to obtain a 10 kHz output. This is then phase compared with a 10 kHz reference in PD1, the output of which then controls VCO1 in 100 kHz steps, allowing for the division by 10 on the input side.
The 10 kHz reference is obtained from a 20.48 MHz crystal master oscillator by dividing down, by 2 to 10.24 MHz, by 2 again to 5.12 MHz, and then by 512 to 10 kHz. The division by 512 step is determined by the chosen PLL IC.
The PLL1 mixdown signal is generated by VCO2, which is controlled by PLL2. For feeding into PLL2, it is first mixed down to the range 131.5 to 231.4 kHz using a 30.72 MHz mixdown signal derived from a variable crystal oscillator (VXO). There it is compared in PD2 with a 131.5 to 231.4 kHz reference signal obtained by dividing down by 100 from a 13.15 to 23.14 MHz signal obtained from VCO3. The PD2 reference signal varies in 10 Hz steps, thus VCO2 is also controlled in 10 Hz steps over a 100 kHz range. In turn, as the mixed down VCO1 output is constrained by PLL1 to move in 100 kHz steps, it follows that the 100 kHz variation range allowed for VCO2 is also reflected in VCO1. Thus the 100 kHz steps provided by PLL1 are interpolated by the 10 Hz steps provided by VCO2.
The VXO frequency may be varied over a ±800 Hz range, to provide Receiver Incremental Tuning (RIT). This variation thus also occurs in the VCO2 mixdown, and given that the latter is constrained to move in 10 Hz steps, it is also reflected in the VCO2 output to PLL1.
The nominal 30.72 MHz PLL2 mixdown signal is chosen to be related to 20.48 MHz master oscillator frequency, being 1.5 times (3/2) of that value.
As noted, VCO3 generates a 13.15 to 23.14 MHz reference signal for PLL2. The output of VCO3 is also mixed down to a 2.91 to 12.90 MHz range using a 10.24 MHz mixdown signal. Next comes a variable divider (divisor 291 to 1290) to provide a 10 kHz output, which is compared with a 10 kHz reference in PD3. Thus VCO3 is constrained to move in 10 kHz steps. The 10.24 MHz mixdown signal is obtained from the 20.48 MHz master oscillator by dividing down by 2. As with PLL1, the 10 kHz reference is obtained from a 20.48 crystal master oscillator by dividing down, by 2 to 10.24 MHz, by 2 again to 5.12 MHz, and then by 512 to 10 kHz. The division by 512 is determined by the chosen PLL IC.
The LO2 feed of 61.44 MHz is obtained by multiplying the 20.48 MHz reference by 3. This is the close to the 61 MHz difference between the 70 MHz and 9 MHz assumed IF1 and IF2 target values.
It may be seen that the fact that the chosen PLL ICs include a division by 512 of the incoming reference signal, the latter will be a multiple of 512, and so will have the general form 2n. In turn those signals also derived from the reference will have the same form, including the PLL2 and PLL3 mixdown signals, and LO2. The chosen 20.48 MHz number is both divided down to 10.24 and 5.12 MHz, and multiplied up to 61.44 MHz.
The PLL1 mixdown logically would take the signal to a range that was below the LO1 range but above the tuning range, using a nominal mixdown signal that was simply related to the master oscillator frequency. The 30.72 MHz number does this, albeit that it was modified slightly to allow for the 10 Hz tuning steps and the RIT facility.
The reason for the choice of the 131.5 to 231.4 kHz range for the 10 Hz increment input is not readily apparent. Possibly it was to suit PD2 in PLL2, but that is just a guess.
The net result is that VCO2 output is in the range 30.8515 to 30.9514 MHz.
With the PLL1 mixed down range fixed at 39.6 to 69.5 MHz, in turn that means that the LO1 range is 70.4515 to 100.4514 MHz, and that IF1 is 70.4515 MHz.
And IF2 is (70.4515 – 61.44) = 9.0115 MHz.
The assumption of a 9 MHz target for IF2 is reasonable. ICOM needed two subsequent IF numbers for its passband tuning facility, the other being the industry standard 455 kHz. 9 MHz was a recognized 1st IF number in amateur (trans)receivers, which with ICOM was involved. Another possibility in the standard repertoire was 10.7 MHz, and it might have been considered. But in this case, conversion from 70 MHz could not have been done with an LO2 that was as simply related to the 20.48 MHz master oscillator frequency. In that regard, 9 MHz was a good fit as well as being a “known” number.
Cheers,
Steve
Posted : 24/12/2023 7:55 am
Topic starter
Similarly, there is an expanded narrative for the ICOM R71.
The assumed givens are:
The same tuning range as the R70.
The same IF structure as the R70, namely 70.4515 MHz IF1, 9.0115 MHz IF2 and IF4, and 455 kHz IF3.
The use of a different PLL IC, one designed for amateur radio receiver applications. This IC has a divide by 1024 function in its reference channel. It does not require the use of a prescaler. In conjunction with a swallow counter, it can do the necessary division directly on the input frequency.
The use of the same LO1 mixdown range, 39.6 to 69.5 MHz.
The use of the same 30.72 MHz nominal mixdown frequency.
The lack of RIT allowed simplification to two PLLs, main and subsidiary.
The main loop, with its divide by 1024 function, used a 10.24 MHz input in order to obtain a 10 kHz reference frequency.
As before, the main loop stepped in 10 kHz increments, but because there was no prescaling of the LO1 input, then it also controlled LO1 in 10 kHz steps, rather than the 100 kHz steps in the R70 case, where the LO1 input was divided by 10.
Thus the 10 Hz step interpolation, obtain by narrowly varying the mixdown signal, was required over only 10 kHz, not 100 kHz.
Without the RIT function, a fixed, not variable 30.72 MHz crystal oscillator could be used to generate the mixdown signal. It could also be used as the master oscillator, divided down by 3 to supply a 10.24 MHz reference input to the main PLL, and multiplied by 2 to provide the 61.44 MHz LO2.
The subsidiary loop provided the 10 Hz interpolation signal of 230.00 to 239.99 kHz. This was mixed with the 30.72 MHz oscillator signal to obtain the mixdown signal. In this case the mixed signal was used directly by the mixdown mixer, there was no intervening VCO.
The PLL IC in the subsidiary loop used the same type of IC as in the main loop. Its reference frequency was 5 kHz, obtained by internal division by 1024 of a 5.12 MHz feed, the latter in turn obtained by dividing by 2 the 10.24 MHz feed generated for the main PLL reference. The subsidiary loop controlled a VCO over the range 115.00 to 119.99 MHz. The VCO output was divided by 500 to obtain the 230.00 to 239.99 kHz signal, which went directly to the 30.72 MHz mixer, with no intervening VCO. The division by 500 of course changed the 5 kHz PLL step to a 10 Hz step.
This was overall a simpler system than the three-PLL configuration used for the R70. In the latter case, the 3rd and 2nd loops respectively fed the 2nd and 1st loops from VCOs at their output ends. In the R71, such intervening VCOs were not used.
The different interpolation oscillator swings, 100 kHz for the R70 and 10 kHz for the R71, meant that the mixdown signal ranges were proportionally different. They were 30.8515 to 30.9514 MHz for the R70, and 30.950 to 30.95999 MHz for the R71. That would have produced a slight difference in the theoretical tuning ranges available between the main PLL lock points, 0 to 29.9999 MHz for the R70, and 0.0985 to 30.00849 MHz for the R71, but this was immaterial against the 0.1 to 30 MHz specification range.
The foregoing narratives support the notion that the IF1 and IF2 targets for the R70 and R71 were respectively 70 and 9 MHz, but they do not necessarily prove it.
Looked at otherwise, in respect of the main PLL, mixing down of the LO1 feed from the VCO might or might not have been necessary to stay within the PLL IC capabilities, although dividing down (prescaling) was necessary in the R70 case. Nonetheless, mixing down provided the opportunity for injection of a narrowly variable signal to provide fine interpolation steps between the coarser main PLL steps. So it was justified on that basis alone.
There could have been good reason to keep the input range, the mixdown range and the LO1 range in separate blocks to avoid mutual interference possibilities. Whereas in professional equipment selling for perhaps ten times the price of the R70/R71, excellent intercompartment screening was possible, that was not always the case at the consumer level. On this basis, at minimum, without any guard bands, or a space for the mixdown frequency, the blocks would be RF signal at 0 to 30 MHz, mixdown at 30 to 60 MHz, and LO1 at 60 to 90 MHz, with a mixdown signal of 30 MHz.
Again, for mutual interference avoidance, the mixdown frequency would preferably be simply related to the reference frequencies used in the PLL system. In turn these were determined by the PLL requirements, in this case multiples of either 512 or 1024. Amongst the possibilities, closest to the notional 30 MHz mixdown requirement was 30.72 MHz. Use of this whilst avoiding overlaps would require pushing upwards both the mixdown range and the LO1 range, say to 32 to 62 and 62.72 to 92.72 respectively, with an IF1 of 62.72 MHz, not taking account of the effect of the incremental signal.
As with the mixdown case, the LO2 frequency was preferably going to be simply related to the reference frequency, limiting the possibilities. Amongst these were 51.20 and 61.44 MHz, which with a 62.72 MHz IF1, would produce IF2s of respectively 11.52 and 1.28 MHz. If the target was 9 MHz, then neither was satisfactory. On the other hand, looking at it from the other end, starting with 9 MHz and adding those LO2 possibilities gave respectively 60.20 MHz and 70.44 MHz as the corresponding IF1 numbers. 60.20 MHz was below the 62.72 MHz derived lower bound for LO1, but 70.44 MHz was fine. Adding in the interpolation tuning increment took that to 70.4515 MHz, and IF2 to 9.0115 MHz.
Looked at that way, it is not necessary to include an IF1 in the vicinity of 70 MHz as a starting assumption. A single assumption of an IF2 in the vicinity of 9 MHz, coupled with the PLL requirements and derivation of feasible values for LO2 gets to the same point.
By the way, that second look came about after another effort (not successful) to unravel the workings of the Kenwood R-5000. One aspect noted by Kenwood was that above the PLL, as it were, the key determinants were the BFO (effectively IF2) and LO2 frequencies. (Below the PLL, it reduced to the reference frequency and four of the PLL variable division ratios.)
With the R70 and R71, IF2 and LO2 may be rationalized as having been key determinants.
Thus, with some receivers, seemingly those at the consumer level, at least, PLL workings are an integral part of IF determination.
With professional receivers, say from the Racal RA1772 onwards, it is usually possible to treat the PLLs as black boxes (my preferred approach), configured to supply the requisite LO frequencies howsoever it was done, although sometimes with an obvious relationship between IF1 and IF2. As seen upthread, with earlier models, such as the Plessey PR155 and Redifon R550/R551, the IFs were to some extent determined by the workings of their relatively simple synthesis systems. And in the Redifon case, those workings were also affected by the initial condition that IF2 was chosen.
Cheers,
Steve
Posted : 27/12/2023 10:14 pm
