FM stereo

[Synchrodyne]

Originally Posted by dave walsh

This extremely technical debate is fascinating but like most types of audio reproduction it has a psychological aspect at the end of the day.

I think it’s a case of “get the underlying science, mathematics and engineering right, or as right as the applicable constraints allow”, and you’ll have a robust and durable transmission system that does not materially get in the way of or indeed non-negligibly modify the signals that pass through it. With a functionally near-transparent transmission chain, the psychology part is then primarily dependent upon the original programme material and the individual listener’s receiving equipment as well as his/her preferences.

That requires a rigorous engineering approach throughout from the initial definition of the requirements through seeking out and evaluating candidate solutions and then selection of the optimum, thence to implementation. Such an approach is also required for any subsequent changes/improvements at both the transmitting and receiving ends. The Zenith-GE system has handled well those subsequent changes. For example, the benefits of PCM programme distribution before transmission were clearly transparent to listeners, as were the improvements brought by PLL-type IC decoders at the receiving end.

The rigorous approach does not really allow for the intrusion of the pseudo-science that infests the audio industry. If you look through the specifications of the Zenith-GE system, you won’t find any comments for example about the use or non-use of negative feedback in the circuitry, or that directional, oxygen-free cables must be used between the audio source and the encoder, or, back in the valve era, that single-ended triode signal amplifiers must be used.


Cheers,

[Ted Kendall]

Agree. The one thing the multiplex system is not, given half a chance, is a quality bottleneck.

[Hartley118]

Originally Posted by Pfraser

That's an intriguing question.

'Relatively silent background' is key, I think. I have a Sony FM & DAB set; also an FM & CD stereo which will accept memory cards and sticks. Both contemporary, average items. Both exhibit low-level but distinct background noise.

In contrast, I recall hearing Sony's first CD machine for the consumer market, the CDP-101, in a radio studio. There was no background noise I could discern.

A later model CD machine, from Aiwa, was much slimmer and lighter! It also produced low-level, distinct background noise.

I imagine that modern digital devices could be designed so as to generate no audible noise, as was the CDP-101. I'm not sure if the average consumer would really notice the difference. Personally speaking, I am sure that I was made more critical of hi-fi performance by my exposure to broadcast-standard studios.

I suspect that an extra-special effort was put into the Sony CDP-101, to assist in launching the CD format. Later models generally didn't need to sound quite so good.

Yes, I used the word 'relatively' silent because digital audio with no background noise at all has always worried me deeply. It implies that the least significant bit is taking no part in the action, usually because dither is either inadequate or absent. Not only are low level signals unresolved, but higher level signals may be subject to quantizing distortion.

It was actually the success of the (properly dithered) BBC 13-bit PCM transmission system that, at a time in the 1970s when many were casting doubt on the quality of digital audio, I became convinced of its future.

Martin

digital audio with no background noise

Quite true at zero, I have a new bedside radio (DAB, Bluetooth etc.) that uses a digital volume control. I can (could) go quite loud this meant that low volume levels had a lot of noise on them when a signal was present (sounded bloomin' awful), it was using only a few bits at that point, no dither. Attenuators (series and parallel resistors) in the 'speaker feeds solved that one.

[Ted Kendall]

And the Corporation bangs on about Digital Radio as if it's something marvellous. When they got digits right, they didn't make a fuss about it. Now they're making a up of it, they won't shut up about it.

[Hartley118]

Originally Posted by merlinmaxwell

Quite true at zero, I have a new bedside radio (DAB, Bluetooth etc.) that uses a digital volume control. I can (could) go quite loud this meant that low volume levels had a lot of noise on them when a signal was present (sounded bloomin' awful), it was using only a few bits at that point, no dither. Attenuators (series and parallel resistors) in the 'speaker feeds solved that one.

An abomination from the way you describe it. Processing of a digital signal demands redithering to avoid truncation and hence distortion. Your analogue solution is sensible.

Martin

[G8HQP Dave]

Originally Posted by Hartley118

the theoretical 6dB degradation in signal-noise ratio on switching to stereo often seems a great deal worse in practice,

No, the theoretical degradation is somewhere nearer 20dB - which is what you find with a good FM tuner fed insufficient signal. A poor tuner will do worse than this.

Wider bandwidth is part of the problem; 3x bandwidth would give 4.8dB degradation, other things being equal. Other things are not equal, because raw FM noise is not white but rises with frequency. Noise around 38kHz would be 22dB worse than around 3kHz baseband. The DSB signal gives an improvement of 3dB, so the net result is about 19db degradation. Note that this is a rough estimate.

You can judge the quality of an FM tuner by comparing the signal strength needed for 50dB S/N mono and 50dB S/N stereo. The latter should be only around 20dB more; if it needs much more than this then it probably has poor AGC performance. Many tuners fail this test; some are so bad that the figure is not published.

I did find some figures which do ring true in the spec of my trusty vintage Technics ST-3500, which quote (at +- 40kHz deviation), mono S/N is 75dB and stereo S/N 60dB, a 15 dB stereo degradation which is pretty audible.

It should be possible to achieve stereo S/N 6-10dB better than this.

After half a century of FM stereo, with today's crowded FM band and with today's listeners accustomed to the relatively silent background of digital media, is it time to consign the GE/Zenith FM stereo system to history?

No, definitely not. Used correctly at both ends it can still give excellent audio; certainly better than current DAB - although if DAB was used correctly (256kb/s for Radio 3) then it might just beat FM into second place.

[G8HQP Dave]

Originally Posted by Hartley118

Yes, I used the word 'relatively' silent because digital audio with no background noise at all has always worried me deeply. It implies that the least significant bit is taking no part in the action, usually because dither is either inadequate or absent.

Dither should be inaudible, unless the volume is turned up really high. Even then you are probably more likely to hear noise from the analogue parts of the system than the dither.

Digital volume controls need to be done properly. Sometimes they are not.

[Hartley118]

Agreed - today's optimised shaped-spectrum dither is pretty quiet. It's when the digits get truncated and act like a noise gate that I worry for the reputation of digital audio.

Martin

[Ted Kendall]

If I remember right, there is also a noise degradation in Zenith-GE because the stereo information is DSBSC AM. not FM, so the "FM improvement" in s/n ratio I dimly remember from Dr Betts' communications lectures is not available. Agreed, though, that optimum s/n ratio on stereo needs 20dB more signal at the receiver input than mono, all other things being equal.

However clever dither is, it should just be audible as increased hiss, if at all.

because the stereo information is DSBSC AM. not FM

But it is on an FM carrier, the mono audio is also AM with a zero frequency carrier.

[G8HQP Dave]

FM enhancement requires peak deviation to be significantly greater than maximum audio frequency. For mono this is 15kHz vs. 75kHz. For stereo it is only 53kHz vs. 67.5kHz.

If the subcarrier used FM rather than DSB then it would need to be wider in bandwidth to get any FM enhancement, but then it would be too wide to fit in.

Zenith-GE is actually quite a clever system and well-balanced. Like all clever systems it is difficult to improve it. You could gain something only by losing something else.

[G6Tanuki]

To get decently-quieting stereo with the Zenith-GE system you need not only a good signal but it also has to be relatively free of multipath-distortion components.

I was always amused by people in the 1970s/80s who would spend significant amounts of money on a good quality tuner or receiver, feed it from a silly 300-ohm 'tape dipole' and then complain about background hiss and burbly-noises on stereo.

OK, perhaps if your only interest was in listening to the BBC, and you were within ten miles or so of one of the higher-powered transmitters.... but I remember a somewhat hi-fi-fixated friend coming to visit and being astounded at my being able to receive Capital Radio, and Radio Boulogne Littoral, in hiss-free stereo using a humble Teleton GT202 tuner when I was living in Ewelme (south Oxfordshire). I took him outside and pointed out the twin-mitre-bracket-mounted 15-foot mast, Fuba UKA8 and Hy-Gain rotator on the chimney. [there was a 5-ele 2-metre Jaybeam up there too].

A week later he'd got himself a UKA8!

Whatever the technology, it all starts with a good signal and that means a good antenna.

[Ted Kendall]

Originally Posted by G8HQP Dave

Zenith-GE is actually quite a clever system and well-balanced. Like all clever systems it is difficult to improve it. You could gain something only by losing something else.

Absolutely - the BBC and others spent a long time arriving at their conclusion, and fifty years on it's still fit for purpose. The requirement of a strong signal is too often honoured in the breach, but it's never been a secret.

[Hartley118]

I'm just suggesting that, by modern standards, Zenith-GE FM stereo is pretty fussy. Very few users are in a position to actually provide their tuner with a sufficiently strong signal to achieve a signal to noise ratio comparable with that available from other modern sources. The theoretical ideal noise performance is all very well, but a single twittering 'birdie' in the background is a major distraction in a quiet music passage - and that's all too frequent in my experience.

When it comes to music on the move, there's nothing more distracting to me than the continually varying background 'shush' so often provided by FM stereo. It's all very well to claim that all would be OK if I had 20 dB more signal from the antenna, but that's not practical in a car. So my car radio is DAB, which gives consistently enjoyable noise free reception virtually nationwide.

Martin

[PJL]

I assume much of the noise over the theoretical level can be attributed to the receiver bandwidth compromise, selectivity vs distortion. I also assume that the decoder itself will introduce further noise.

When I added a stereo decoder to a valve FM tuner I spent some time realigning the 10.7MHz IF to try to keep the response reasonably flat over +/-100Khz but this is well below the recommended width. With AM it is pretty easy to understand the impact of restricted bandwidth but for FM stereo it is far too complex.

[Ted Kendall]

I take the point about mobile reception, although EOS can mitigate the effect, although many chip-based mobile receivers have sub-optimum s/n ratio anyway. Mind you, the noise level in a car is such that the limitations of DAB are less audible.

[Electronpusher0]

Originally Posted by Radio Wrangler

There was a fashion for having a pair of rear speakers connected in series differentially across the live sides of the feeds to the two normally-connected speakers at the front. An elaboration was to have a suitably beefy variable resistor from the otherwise floating junction of the two rear speakers.

Sorry for referring back so far but I have only just found this fascinating thread.
I remember making a bit of pocket money with a friend making and selling a simple box, aluminium u chassis in a veneered sleeve, with DIN speaker sockets on the back and a switch on the front that selected, Stereo (front speakers only), Surround (rear speakers connected Hafler style across the plus outputs of the front speakers) and what we called "four speaker stereo" where the rear speakers were connected in antiphase across the opposite front speaker (left rear being antiphase right front). We advertised in a national paper and sold quite a few.
Peter

[Radio Wrangler]

Lots going on here!

If a broadcast system of any sort gives a perfectly silent background, then it must have less dynamic range than the source material. Studio microphone preamps and mixer summing nodes are not noise free. There is always noise associated with any source of material unless you're into cryogenically-sourced electronic music. If the noise of a reproduction system, at the listener's volume setting can be got below the listening place's ambient noise level, then its perception should disappear.

Doing a volume control in digits means that the DAC must handle the combination of the dynamic range of the incoming signal PLUS the required range of volume control. It is unlikely that there will be enough bits or a good enough DAC for the loss to be made unnoticeable. Doing a volume control whether analogue or digital ahead of a high-gain amplifier stage is a standard recipe for exaggerating the amplifier's noise figure. Being able to modify the amplifier's gain itself can be advantageous - and we're back to the concepts discussed in the Cambridge P40 thread.

Care and sensible allocation of gain can produce a good enough result, though. The difficulty comes with people who believe that there is no limitation to the perception of their hearing. The inevitable consequence is that for them nothing can ever be good enough. Even issues that just might in theory create an infinitesimal difference have to be fixed, at least in the mind associated with that remarkable hearing. There's nothing we can do about this. Logic, reasoning and a sense of scale do not work in this belief-system. Pseudo-science carries as much weight (sometimes more) than the real stuff. So sit back and let them entertain you.

The circus that is the audio industry has some wonderful, breathtaking acts. No animals are harmed. There are showmen and snake-oil salesmen, and it's generously provisioned with all sorts of clowns.

Where I live, about 25-30 miles from the BBC VHF/FM transmitters, I get a good signal to just a dipole. Both Revox and Sony tuners give satisfactory audio, good enough to be preferable to the compression effects I get on my DAB receiver. The FM reception function included in that DAB set is definitely sub-standard, and listening via that set alone will give anyone the inclination to select DAB. I suspect FM's inclusion was just a tick-box exercise and there wasn't a box for whether it worked properly or not.

Travelling around Scotland, where the scenery is a bit more 3-dimensional than the rest of the UK. The odd bit of background shoosh on FM is not uncommon. My car's system is quite good at blending stereo to reduce its intrusion, and it has two antennae in opposite rear side windows feeding diversity receivers. It's about as good as it gets. All too often it's in the garage, and the garage's loaners all now have DAB sets. Travelling my familiar routes simply indicates that the provisioning of DAB services is not yet up to the standard of provision of FM services. Either system in good areas is quite adequate for use in a car. The newer system has just not (yet?) been brought up to the coverage of the older system.

At home, digital systems can be of any quality you want if the data rate can be afforded. However, the cost factor has certainly caused some channels to be compressed into intrusively low data rates. The less compressed ones available are still not out of the woods.

FM stereo, in a reasonable area is still a very hard act to follow. It was a very well-designed system and fully fit for all its purposes.

For car use, its chosen successor needs more transmitters to equal its coverage.

For domestic portable radios, there is still a difference in need for signal strength.

For domestic hifi systems, it'll take high rate streaming via the internet to equal what FM could deliver, though the Optimod wars have rendered all but R3 unlistenable.

It's a bit late, now, but people are beginning to learn just how good what we used to have was.

David

[G8HQP Dave]

Originally Posted by Hartley118

I'm just suggesting that, by modern standards, Zenith-GE FM stereo is pretty fussy. Very few users are in a position to actually provide their tuner with a sufficiently strong signal to achieve a signal to noise ratio comparable with that available from other modern sources. The theoretical ideal noise performance is all very well, but a single twittering 'birdie' in the background is a major distraction in a quiet music passage - and that's all too frequent in my experience.

FM is no worse than DAB in needing a strong antenna signal. Both can fail with insufficient signal. The difference is that FM just goes noisy while DAB goes bubbly. You can switch FM to mono or add an HF filter - or a combination of both; you can't do anything with DAB. FM birdies are as likely to be due to too much signal as too little signal. For home use both systems are engineered on the basis of a decent antenna on the roof; the fact that few people do this is up to them. I suspect that they do not because mono FM can often cope with a telescopic on the set. DAB and stereo FM need more.

Originally Posted by Radio Wrangler

I suspect FM's inclusion was just a tick-box exercise

Indeed. I recently learned that in order to get the Digital Tick logo a set not only has to include DAB+ but also FM. The spec does not say why: FM to continue? FM as a reliable fallback in DABless areas?