Historic AM audio quality

[Synchrodyne]

Originally Posted by Nickthedentist

Well I never! A chap who lived about 1/2 a mile from me in this part of Oxford, and who had a well-known radio/hifi shop which survived (albeit as a B&O centre) until last year.

He died in 2009: https://reader.exacteditions.com/issues/34226/page/9

Geoffrey Horn also designed the prototype Quad FM tuner, of which a small number (a single digit count, I think), were built in 1952. John Collinson finished the production design in 1955, including the addition of the signature twin neon centre-channel tuning indicator, the idea for which he is said to have obtained from an article in an American magazine. I suspect, but don’t know for sure, that this would have referred to the Sonocraft FM tuner. But that is getting off-topic. As far as I know, both Geoffrey Horn and John Collinson had strong RF backgrounds, something that was not often found in the specialist hi-fi amplifier companies.


Cheers,

[Synchrodyne]

To put into perspective the introduction into UK AM broadcasting of ruinous brick wall filters with cutoff around 5 kHz, circa 1972-73 the vast majority of AM receivers in use in the UK would have been of the narrow-band type, with AF responses that cutoff mostly below and rarely above 5 kHz. So the broadcasters could have argued that they were taking away that part of the AF spectrum that de facto was not being used by listeners anyway. (Although I suspect that the deleterious effects of brick wall filters might just be noticeable on receivers that went to 5 kHz with a gentle rolloff thereafter.)

Perhaps the argument could have been extended to say that listeners had made a deliberate choice in favour of narrow AF bandwidth by buying narrow-bandwidth rather than say variable selectivity receivers. But it was less of a deliberate choice and more of a forced choice. Presumably in part for economic reasons, the mass-producing setmakers had generally abandoned variable selectivity receivers with wideband capability by the early 1950s, although it remained available on a steadily diminishing number of hi-fi tuners.

A look through Hi Fi Year Book (HFYB) 1972 shows only the Quad AM3 with confirmed wideband capability. The Telefunken T250 had AM bandwidth switching, but drilling down into this showed that the IF bandwidths were 8.5 and 3.5 kHz, that is narrow and very narrow, but not truly wide. Although not stated in the HFYB 1972 entry, the Grundig RT100 also had variable selectivity of the narrow/ very narrow type. Here the choice was between the AF response being 26 dB down at 5 or 3.55 kHz. I suspect that where automatic (signal-strength dependent) variable selectivity was employed, it was over the range between narrow and very narrow. The Armstrong implementation in its 200-series products (I don’t know if it was used on later products) varied the IF bandwidth between 2 and 8 kHz.

Thereafter, following the general European maker abandonment of quality AM reception, one had to look to one or two American, Australian, New Zealand, and very occasionally Japanese makers to find true wideband AM receiving equipment. Something of a surprise in there was the Philips USA AH673 FM-AM tuner of c.1977, which had a highly-specified AM section with 10 kHz AF wideband capability. I don’t think that it had a Philips Europe counterpart.

The original post included the question:

Originally Posted by Scott37

Does this mean that medium wave / AM radio sounded much better in the old days than it does now?

I think the answer to that is, in the Japanese style, “maybe”. Assuming good receiving conditions, as might be expected in the inner ground-wave service area of a transmitter, then back in the valve era, the main limitation was the narrowish bandwidth of much of the receiving equipment. Now it is mainly the narrow bandwidth of the transmissions, although much receiving equipment probably provides some further restrictions.

As a thought experiment, one could set up a comparative test in which a high quality audio source was fed, via selectable filters, to an AM modulator with wide bandwidth (15 kHz AF) and very low distortion, and then on to a selection of receivers (and tuners), old and new. One might extend the scope by also feeding the audio, from just ahead of the AM modulator, via a level control, to the gram or line inputs of the receivers under test, as a means of checking their AF performance.

The selectable filters, low pass, would include brickwall types that emulate those used by the broadcasters, and perhaps a variable type, like that include din the Quad 33 and 44 control units. The latter would allow determination of the point at which transmitted quality became the main determinant for given receivers.

Of course, another difference factor between then and now is the advent and widespread use of over-compression, but then that has afflicted FM as well. I once saw an Optimod advertisement whose message was that the sound of a classical music FM station could be improved through the use of its processors. I remembered that fiction the other night at an NZSO concert at Baycourt (a smallish hall with quite nice acoustics); one programme item was Schubert's "Unfinished", which does have a very wide dynamic range (visually a nice piece as well); it occurred to me that if the jokers at Optimod could find a way of effecting compression in the acoustic domain, they’d be hard-selling it the concert halls as a way of improving the sound…..

Originally Posted by Synchrodyne

Langford-Smith described differential distortion a lot better than I can.

Also, there’s a full discussion, with quantitative and graphical information, in an article by K.R. Sturley in Wireless World 1937 December 23, p.632ff.

Regarding off-centre tuning to, this is from the Quad AMII operating manual:

"Sometimes, too, adjacent channel interference experienced in the Filter or Wide positions can be reduced by detuning slightly away from the interfering transmission, without greatly affecting the wanted signal. Under these conditions such detuning may well result in an overall improvement of reception.

"(It should be noted, however, that the true tuning position remains unaffected in spite of this audible improvement which is also accompanied reduction in high frequency response and apparent increased volume due to reduction of the AGC voltage)."

A diode-type AM demodulator will produce distortion if presented with any sideband asymmetry. But if the asymmetry, and hence the distortion¸ is restricted to frequencies above say 7 kHz, it is less likely to be problematical. This would be the case for a wideband receiver. In the Quad AMII case, offsetting of up to 4 or 5 kHz would be possible in the wide setting before the asymmetry affected frequencies below 7 kHz. Distortion, probably to a greater extent, is produced if the carrier amplitude is reduced relative to the sidebands by any material amount. This is more likely to happen in a narrow-band receiver, where the IF response curve will have a curved top to some extent. In a wideband receiver, the IF response has a flat top over quite a wide bandwidth, so mild offsetting would not cause carrier reduction.

Quasi-synchronous demodulators with unfiltered reference channels seem to do better when it comes to offsetting. Any sideband asymmetry will cause quadrature distortion in the reference channel limiting process, but this is not a big nuisance. The exalted carrier nature of this demodulation type means that within reason, carrier reduction caused by offsetting is not a problem. This can be shown in for example the JRC NRD-525 HF receiver, which has this kind of AM demodulator. It can be offset to the edge of the IF passband (and a bit beyond) without major distortion (beyond that always there in what is a not very good AF section).


Cheers,

[Radio Wrangler]

I'm following this with interest. To have designed and put into manufacture an AM transmitter in the 21st century puts me into a bit of a minority. It's readily apparent that you're going against the flow when you try to find suitably rated devices.

The curved characteristic of a valve used as an AGC-controlled amplifier is immediate and applies to all components of its grid voltage. Distortion of an AM signal is inevitable, especially at times when the modulation depth is great and the AGC moves the operating point into the curvier regions. The same goes for transistor stages with AGC controlled bias.

We do, nowadays, have a number of low distortion variable gain techniques. PIN diodes and Gilbert Cells. Back in the day CdS LDRs and neon bulbs would have been great, but the complexity compared to just shoving a variable-Mu pentode in, would have been too expensive.

Diode detectors constitute a non-linear load on whatever supplies signal to them. It's their nature. People would have baulked at the cost of adding adequate buffering.

Delayed AGC tends to make fewer stages run in curvier regions at certain levels and exacerbates the non-linearity of the stages.

Low distortion AM reception, back with wide transmissions, did sound rather good.

AR88s and HROs set wide sounded wonderful on the BBC. The general coverage receivers coming out of Japan for the amateur radio market in the late sixties onwards were grim on AM. They had lousy distortion on SSB as well. Evidently the designers, once they'g got through a crystal filter thought anything goes...

David

[simpsons]

As a schoolboy, working Saturdays for the Cable Television and Radio service in Sowerby Bridge Yorkshire, I built the Practical Wireless Luxembourg 2 valve TRF receiver, designed by S Collins, to give their customer’s an additional service.
I can’t remember what radio receivers were used for BBC reception but Radio Luxembourg wasn’t available. I imagine the main receivers were crystal controlled superhets and too expensive to buy one just for the minority youth audience of whom I was one. The main receivers had to be kept in tune day after day in all temperatures; after all there would outrage in The Upper Calder Valley if the Third Programme had drifted off to another station. Perhaps not, but you get the drift, no pun intended!
The PW circuit was on a blueprint and I don’t have a copy, but looking at the construction details in the December 1962 copy, it used 1.6 MHz IF transformers with a small fixed capacitor to bring the tuning mid point to 1,439 KHz which was Radio Luxembourg’s frequency.
For my sins, I had little understanding of AM audio bandwidth or distortion and was more than surprised that my construction worked. The aerial was strung along the rafters on the top floor where the twin-valve audio amplifiers were located. Each amplifier gave off so much heat that it was a bonus to be allowed to sit in the room during the very cold winter. As my receiver was there, I had a good reason to be able to enjoy the warmth and I did whenever I could during those very cold winters.
It may sound stupid to say but despite trying so hard to give the best quality sound from the TRF design, listeners heard all the radio stations through an “extension loudspeaker” box with very limited audio frequency response. As there was no monitor reference speaker at the source amplifier to check the sound quality of Radio Luxembourg being sent down the multi cable to subscribers, who knows what the overall frequency response was or what they thought of Mr S Collins TRF design.
Happy Days

[paulsherwin]

There is an interesting simple TRF tuner design for 1500m in a 1972 Radio Constructor article. This would have given good wideband reception in strong signal areas, but sadly the brickwall filters were added soon after, negating any benefit.

http://vintageradio.me.uk/radconnav/...72/circuit.jpg

My A22 does sound better from my SSTRAN 'modulator' (wide band AM) than from off air, no brick wall filtering just a smooth low pass from the IF (intermediate frequency) filtering. I suppose it is that a little of the high frequencies get through and there is no horrible phase shift at the end of the brick wall.

[Nuvistor]

No Optimod type filtering from the SSTRAN either. The MiniMod I use even with its simple circuit sound much better through my radios than off air stations.

[SteveCG]

Re Synchrodyne's post no. 62

I remember using a Sony 2001D receiver with its selectable sideband capability. This was useful on the short-wave as a way of making the most of the sound quality of a station that had a strongish adjacent channel source of interference. Obviously it can do no better than the limited bandwidth of the desired programme - but at least you could make the most of what was available compared with double sideband reception.

Not quite an AM III !!

[G6Tanuki]

In the past I built a number of AM broadcast-receivers using synchronous demodulation - this being an attempt to get round the horrendous selective-fading (where either one sideband fades but the carrier-and-the-other-sideband don't, or the carrier fades wrt the sidebands) when trying to listen to 208, the pirates etc. at night.

A double-balanced mixer with a really-strong locally-derived carrier in-phase with the received carrier is really rather effective: either generate this by phase-locking the 'BFO' to the incoming signal (which has the disadvantage that if the carrier fades the thing comes unlocked and you get a 'growl' superimposed on the audio) or my other technique was to split off a smidgin of the signal at IF, put it through a 'tight' crystal-filter to attenuate the sidebands and leave mostly-carrier, mercilessly amplify/limit it to remove the modulation, then feed it to a Schmitt trigger to generate essentially a square-wave which - being derived from the original carrier - was always in-phase.

The results were rather good: I really wanted to do a 'selectable sideband' version [with a narrow IF filter that would remove the carrier and one sideband, then detect the result as SSB using the 'processed' carrier as described] but paid employment got in the way.

These days you'd resolve the I and Q components and stuff them into the 2 channels of a PC soundcard, then do the rest in software.

[Synchrodyne]

There have been several efforts at using PLL fully synchronous demodulators in production consumer-level AM receivers over the years, with varying levels of delivered performance. There was a time when it was almost a de rigueur feature for HF receivers, which probably resulted in some less-than-stellar implementations.

Probably the best of these efforts was found in the Phase Track Liniplex F1 and F2 HF receivers of the 1980s, which were designed for those who wanted to listen to HF broadcast programme content, and not for DX’ing. They had an interesting origin. Their designer once worked for a major manufacturer of professional communications receivers, and when stationed overseas, found that the said receivers were not actually very suitable for listening to the BBC World Service programme content. So he designed and built a receiver specifically for that purpose. It had a relatively simple and “clean” RF side, and a tracking PLL fully synchronous demodulator with selectable sideband by the AF phasing method. Being optimized for HF reception, it had a narrowish AF bandwidth, although that was about the maximum typically usable for 41- and 49-metre bands back in the later 1980s and into the 1990s.

I bought a Liniplex F2 after some correspondence with its designer. When feeding a reasonable hi-fi system it was the only HF receiver I have used that passed what I called the “BBC WS Play of the Week” test. That WS programme turned up on Saturday evenings when I was living in Texas. The objective, suiting the day and time, was to achieve relaxed, hands-off listening, sitting in the armchair with a glass of something red in hand, cat-on-lap, without requiring undue straining to catch and enjoy the content. Missing a line or two in that kind of programme was unhelpful, to say the least. The Liniplex PLL was quite tenacious, and I don’t recall that it ever unlocked. The BBC WS evening frequencies (5975 and 6175 kHz inter alia) in that part of the world were subject to severe multipath fading, with the signal strength meter often swinging (at about 1 or 2 Hz) over a 20 dB or more range. Clearly the AF output in those circumstances was not fault-free, but it was devoid of any distracting distortions, and over extended periods did not offend the ear accustomed to listening to CDs and good FM broadcasts. I suspect that a relatively linear AF path following the receiver also helped.

Prior to acquiring the Liniplex, I had used a Sony ICF-2010. This at least pointed the way in respect of PLL synchronous demodulation, but it was not a device for extended programme content listening. I also had a JRC NRD525, good for DX but also not in and of itself a programme content listening device. I used it with the Sherwood SE3 outboard PLL synchronous demodulator. The combination was quite good, but not as good as the Liniplex. It was fiddly to use, in part because sideband selection was done by offsetting. But it was useful to have a second reception channel, as it were, as this could be setup in advance of an approaching on-the-hour BBC WS frequency change. (The line outputs from the HF sources were switched in a Quad 44 used as an auxiliary switcher, thence feeding the Quad 66/606, so changing from one receiver to another was a simple push-button exercise, after which the Liniplex could be reset for the new frequency.)

For MF reception back in those days I used a Carver TX-11a AM-FM tuner. This had a dual-bandwidth AM section, with AF to 15 kHz in the wide mode. It was equipped for C-QUAM AM stereo. The Motorola decoder ICs for the latter used an interesting mix of techniques. The envelope demodulator for L+R was a low-distortion implementation of the quasi-synchronous type with a wideband reference channel (necessarily so because it had to retain the phase modulation). But the I and Q demodulators were of the PLL fully synchronous type. Notwithstanding its wider bandwidth, I thought that the Carver TX-11a was not quite as good as the Quad AMII, though.

An interesting connection here is that the Sony ICF2010/2001D HF receiver used for PLL demodulation an IC set that Sony had originally developed for multi-system AM stereo decoding. That IC set – which I think used only PLL demodulation, unlike the Motorola ICs - was used in the Sony SRF-A100 small portable receiver, which had a dual bandwidth AM section with around 10 kHz wide AF bandwidth. Using the headphone output to feed a hi-fi system, it gave quite a reasonable account of itself. At one time my home office/study radio system consisted of the SRF-A100 and one of the ICF2010s, both with headphone outputs feeding a passive switcher and then on to the auxiliary input of the Beolit 707. (These days it is the Beolit 707 alone, permanently tuned to RNZ Concert FM

In the professional field, minimization of selective fading distortion had been an objective long before it appeared in the consumer domain. From the earlier 1950s at least, some broadcasters had used commercial point-to-point ISB/SSB receivers for the reception of HF relays or actual HF broadcasts for relay. In the reconditioned carrier mode, these used very narrowly filtered and limited extracted carrier to drive product demodulators fed with one or other sideband. 6 kHz sideband filters were the norm in such applications. As well as purpose-built receivers, outboard adaptors also featured in this role. For example, Racal offered a broadcast-type SSB/ISB adaptor to partner its RA17. In the solid-state era, sometimes the reconditioned carrier was obtained from a PLL driven from the filtered and limited carrier. I think that the Eddystone EC958/12 did it this way.


Cheers,

[Synchrodyne]

I should also have mentioned that a very early effort to bring locked oscillator synchronous AM demodulation with selectable sidebands into the consumer product realm was made by GE in 1948 with its YRS-1 outboard adaptor, all done with valves, of course.


Cheers,

[Pfraser]

With regard to high-frequency response: I once had a 'ghetto blaster' with very good performance on MW stations that could make use of it. This was most obvious on Dutch station Radio Monique, on 963kHz. It was a bit like 'FM on MW'! I understand that they used an Orban Associates Optimod to bring this about.

No other radio I had at the time could perform like this. My collection included a Vega Selena B212 and a solid-state Bush radiogram. I do recall that the Selena could give very mellow and sweet-sounding audio on FM. Separate bass & treble controls on a portable!

I'm virtually certain the ghetto blaster in question was an Hitachi TK7100E. I paid £25 for it, second hand.

I did enjoy the sound from historic sets I had:- a valve Regentone radiogram with a large & sturdy Goodmans speaker was rich and quite bass-capable. My Pye Fen Man 1 was pleasant to listen to.

It's quite revealing to hear a MW station's signal on a professional monitor. I worked for a station using the Orban Optimod 9100B, set up for pretty heavy processing. On a monitor receiver, the audio was unpleasantly harsh on the treble end of the spectrum. Treble pre-emphasis was recommended, apparently, to compensate for roll-off in the majority of receivers. The AM monitor (Belar) was specifically designed not to have this roll-off.

A cheap way of listening without roll-off: crystal set + crystal earphone?

For the best compromise of punchiness, frequency range and clarity, I'd nominate Luxembourg on 208m. Pity about the fading!

[Graham G3ZVT]

In 1970 I used to get a lift to work from a colleague, he had a VW Beetle and I was very impressed with the MW/LW factory fitted radio.

I think the enhanced bass was due to the large dashboard mounted loudspeaker and the extended treble was thanks to the less restrictive TX filters and an IF response that took advantage of it.

Whatever it was, the sound of that set was the benchmark by which I judged other AM sets for years after.

[Synchrodyne]

Somewhere – I can’t find it right now - I have an SAE Paper on the use of pre-emphasis in MF AM broadcasting as it existed in the USA in the 1980s. It was all ad hoc, and some fairly severe curves – beyond first order – were in use. That chaotic situation had led to the development of the NRSC pre-emphasis curve, basically 75 microseconds with a shelf at 8.7 kHz (originally it might have been 9.5 kHz). Cutoff at 10 kHz was also part of the definitive NRSC specification.

Much of what has been discussed in this thread is neatly summarized by this 1992 article from Audio magazine:

Cheers,