FM Switchover- response deadline looming

[Resurrection Man]

to #13:

Yes Dave, many of your points I''d agree with. The local FM licences will be published as and when the legislation is written and then you can apply and see how far you get!!

As to WTP I think it's not an approriate measure. Anyway it isn't the point any more, sets are available at the WTP point that surveys have established so why isn't there a mad rush to buy?!? Consumers are not not buying DAB because they are expensive it's because they don't like DAB and more channels doesn't attract enough, there's plenty of choice on FM already!!

Neither do they (consumers) appear to be bothered by interference on FM as suggested by the WTP question and on which the whole WTP argument rests (for the most part).

Flawed. Flawed. Flawed.

[Dave2002]

Gordon

Because it is more or less on the Shannon limit probability is that T2 is the end of that particular road [better be careful predicting].

I rather doubt that. I believe it's rather hard to get really close to the Shannon limit. However, I would agree that several factors might indicate that we are getting close to practical limits. TV multiplexes go up to at least 64 QAM. Maybe 256 QAM would be possible (is possible). That'd only give a 33% increase in capacity, which could at least partially be offset by the need for any extra error correction features. I doubt that 1024 QAM is realistic, but that would only give a 67% improvement over what we have now with the same caveats re error correction.

Digital compression could be improved further. Arguably this would lead to further audio quality erosion, though it might be argued that if high quality broadband services may become available that quality constraints (are there any?) would be less important. I suspect that by spreading the digital signals over time that considerable savings could be made for some types of material. However this would require significantly greater buffering delays at the receiver, and possibly also at the transmitter, though for pre-recorded material this could be done in advance. I'm not sure how acceptable services would be with greater buffering delays - probably not at all.

Another way to improve capacity would be to increase the transmitter power, which would perhaps reduce some errors. This would tend to nullify one of the supposed advantages of DAB - that it's possible to transmit more for the same power as an FM station.

I'm not really disagreeing. I think that for the moment we are quite close to practical limits, though I still feel we are way off the Shannon limits.

Another problem area is that if there is the possibility of future improvements we may lose out if the specification is set in stone too early. That's already a problem with vanilla flavoured DAB. Will future sets have dynamic updating, like many new DVB systems? Alternatively will some still be hard wired?

Will users take kindly to radio sets which update themselves? Will manufacturers guarantee to keep their old models up to date? I suspect many of us have working sets from the 1970s, and if we take AM services into account some may still be using even older sets. The economic assumption that many people are prepared to replace/discard sets every 5 years may just be wrong. Many manufacturers of FM sets have already gone out of business, yet their sets continue to work. If DAB sets are to do dynamic updating, then there are severe potential threats due to lack of business continuity. Who will care? (!!)

The whole thing is a minefield, and if those people looking for "public opinion", by which they seem to mean tweaking until they get the answers they want would just let up for a bit, then in the long term something more sensible might happen.

[Resurrection Man]

I just 'had' to look up the Shannon limit...

In information theory, the Shannon–Hartley theorem tells the maximum rate at which information can be transmitted over a communications channel of a specified bandwidth in the presence of noise. It is an application of the noisy channel coding theorem to the archetypal case of a continuous-time analog communications channel subject to Gaussian noise.

So there you have it. Happy to help.

[Dave2002]

A few more points. DAB in cars is pretty much an unknown quantity. I bought a new car nearly 5 years ago, and when I asked if DAB was available the salesman looked at me with disbelief, contempt or something like that, and told me there was "no need, no demand, car manufacturers not interested." So we really don't know how well DAB works in cars because hardly anyone has one. We don't know what the DAB coverage is, or is likely to be. At the moment of buying a car my WTP would have been about £100 cf. an FM only staandard radio.

The first step, surely, if HMG really wants to promote DAB is to ensure that all car manufacturers install DAB sets (DAB+FM) as standard. Then, if further developments are to be consumer led, wait and see how things pan out.

it's also interesting to read from Gordon's fairly comprehensive response that car based DAB really is quite difficult - though I don't know why.

Secondly, in the USA I believe that satellite rsdio has had reasonable market penetration. Satellite radios don't, yet, really work as portable units, but they are OK in cars. I do wonder why there has been no interest in satellite radio in the UK. It might be a better option for wide area coverage than DAB.

Thirdly, and maybe a bit wackier, if enough bandwidth is available, then on demand services may become possible even using broadcast methods. This would probably require 10 to 20 times the bandwidth of a radio channel, which means it should easily fit into an HD video channel. **

Also related would be the notion of using a data carousel mechanism to supply downloads at the same time/in parallel with live broadcasts. This is relatively easy (seemingly) with wired internet connections, but harder for broadcast services. However, even there (Internet) AFAIK unicast is often used where multicast could be effectively used to deliver comparable services with less load on networks and servers, and only, in some cases, a slightly increased complexity for each receiving device.

** in fairness this may be harder than that, but some programmes could be done using broadcast on-demand transmission. Made you think though, didn't it!?

At one time I thought some digital radio enthusiasts were really thinking ahead, and would have been interested in such ideas, but now I think the drive for "monetization" (i.e. filthy lucre) by certain interested parties is muddying the water and creating a smokescreen - to mix some well known phrases!

[Gordon]

#17 and 18:

Thanks RM for that, it helps people get a grasp of some complex techy stuff and points the way to the fact that there is lot behind throwaway remarks!!

Beware though, as usual the devil is in the detail. This limit is theoretical and is not fixed in stone like say the speed of light is as a fundamental. The "channel" will determine the limits of communication and communication channels are not all created equal.



Shannon simply stated that there is a trade to be had between C/N and bit rate. You choose one and live with the other and do well if you take proper acount of your channel. These show where we are in relation to Shannon: in T2 for a given bit rate we are around 1-2dB away from the limit across the scale, 6-7dB for T1. So if we fix at, say, 16dB for C/N we get about 24 MBit/s for T1 and 36 for T2 and the limit says we could get 40 in a perfect, unachievable world. Diminishing returns rule.

Notice that the conditions of the channel are noted at the top of the graph, fixed not mobile reception. "Stationary" doesn't mean physically fixed here it's a techncial term about the channel noise state [try googling "stationary noise" or "ergodic noise", it'll help break the ie at parties]. QEF means Quasi-Error Free at 10^-11 error probability after the correction in a noise limited channel. This assumes that any multipath has no effect having been absorbed by the processing. There is NO natural interference which is of course always present and Shannon can do nothing much about except take a bit off the C/N to acount for it.

[Gordon]

....The first step, surely, if HMG really wants to promote DAB is to ensure that all car manufacturers install DAB sets (DAB+FM) as standard. Then, if further developments are to be consumer led, wait and see how things pan out.

All in hand. European legislation is in place. About 30% of new cars for 2013 will have DAB either as standard or as option - SMMT says so.

it's also interesting to read from Gordon's fairly comprehensive response that car based DAB really is quite difficult - though I don't know why.

In a factory fit nowadays the radio is installed behind a dash and operated from buttons on a panel shared with other functions. Its wiring is all hidden. The antenna is wired both for FM and DAB in different bands which means a compromised aerial with a diplexsr in the feed. The antenna has to be cosmetically pleasing rather than technically good. The car body is made of conductive metal forming a complex Faraday cage which means an external antenna is best but not always fitted. A plastic roof means no ground plane to help the roof mounted antenna. On screen printed antennas are used; they are not very good and the polar plot is usually very poor and exacerbated by the Faraday cage effect which usually ruins it. A BMW result I saw tested had a really dreadful polar plot and low gain but it looked nice. A poor polar plot means that you can have good reception going South say until you turn a corner to go East and then its gone. I could go on but I think you'll get the flavour.

[Dave2002]

Very interesting. i didn't think we'd got so close to the limit as that yet. Also surely Shannon "is set in stone"! The theory cannot be violated.

However most signals are not purely random, so it is also possible to exploit redundancy there, and get an improved result. I'm not sure how far we are away from the limits there either. Most lossless compressed audio only results in a 2-4 times improvement. Lossy compression, where we are now considering the channel all the way to the ear/brain can do better, because the perceived response for most people is very similar, even with lossily compressed audio.we can argue about the point at which losses becaome detectable/indetectable, but for broadcast purposes I'd suggest that most codecs running on 320kbps streams are good enough.

[Dave2002]

Re msg 21. Not wishing to play into the hands of the gungho switchover police, but would DAB only car radios be better/easier because there would be no need for the diplexer? I'm not sure how significant the diplexer is. Perhaps in a fixed environment it would be easy to cope with - just a few dBs loss, with little injected noise, but presumably in a car where everything's running at the margins (or beyond) this becomes very significant. I'm now thinking that the "needs" of mobile users are really not being taken properly into account in the switchover proposals, and there's little evidence or user experience to rely on, particularly in the wide flung reaches 50 miles or more north of Watford.

[Gordon]

Very interesting. i didn't think we'd got so close to the limit as that yet. Also surely Shannon "is set in stone"! The theory cannot be violated.

With Shannon you can negotiate within the trading space so there is some flexibility which is why I said it is not set in stone. As a "law" of nature you are right, it is a firm bound.

Most codecs use buffers to smooth over peaks and troughs of "redundancy"; variable bit rate avoids that need to some extent but we tend to stick with fixed rates. The buffer introduces delay. This is more significant perhaps in video than in audio for obvious reasons. Audio is not so clearly systematically repetitive.

Removing redundancy involves removing all of that which is predictable in a signal. What you need therefore in a coder is a very good predictor which is able to track the input very precisely [using some statistically derived model perhaps] and send minimal, preferably no, information, which represents the failure to predict exactly, to a synchronised decoder. To do this for music is tricky. Any random noise will thwart this aim and the more noise there is the worse it gets. Don't compress dirty signals.

[Gordon]

#23: there are two sets of car problems. I mentioned the new car situation [BTW Europe says they all have to be DAB/FM!!] but some also transfers to old ones. If you think of the 30 million cars already out there the majority of which do not have a DAB fitted [but may have a PURE highway or something which is fine for good signal areas not very good at service limits] but do have FM, perhaps integrated to the dash too, then converting those may involve pulling the dash apart for one thing to get at the receiver. It isn't always just an ISO/DIN standard slot replacement. Possibly a DIY job?? If they all need converting professionally in a period of 2 years then 15,000,000 a year need doing, or about 50,000 a day! If a trained person can do 5 swaps in a day that needs 10,000 "competent" persons flat out. Doesn't sound undoable with all the garages around. People being people will not do that though, they'll wait to the end of the period and then everyone will need doing in the same 3 weeks!! Tough on them if they've been warned!

Another way is to make converters [low power FM transmitters] that take DAB and put out FM to give to the existing receiver on a given FM channel. More fiddling behind the dash. Challenge there is to take the Traffic and Travel data off DAB and send it to RDS - they are not directly compatible. I'm told that some receivers can be located anywhere in the body these days. The fitted antenna will be a band 2 and will be bad at band 3. New antenna needed? If consumer wants to keep FM to get the new local FM stations eg then the old antenna needs to be kept or a dual band fitted. Diplexer? In the case of the old monopole extending ones [the older cars] all one needs to do is shorten the thing to the new band 3 and so a bit of retuning the antenna!! Adjusting it as required when changing band? OK but a bit tricky on the move. Not perfect but might work for some.

The car and commercial vehicle market is very wide and its requirements are rather peculiar and need a range of fixes. The motor industry has been at it for some time and have come up with a few solutions that should satisfy most needs. But, given the reputation of the motor industry at local garage level would you trust anyone to fix yours? Halfords will surely offer a fitting service?

Replacement DAB car radios should meet a Minimum Specification that has been prepared buit it won't be mandatory so there is room for cowboys. As usual buyer beware.

[Anna]

I have two wonderful, vintage, valve radios. Oh well, down the tip with them. But what of us rural dwellers, who cannot get DAB? Oh yes, we can listen via our computers and our satellite tellies.

Do not the powers that be recognise the UK is not London?

Yours, Disgusted of Llanwrtyd Wells

[Dave2002]

I have two wonderful, vintage, valve radios. Oh well, down the tip with them. But what of us rural dwellers, who cannot get DAB? Oh yes, we can listen via our computers and our satellite tellies.

Do not the powers that be recognise the UK is not London?

Yours, Disgusted of Llanwrtyd Wells

Well done Anna! Agreed.

Re Shannon limit - apparently Low density parity check codes can get very close to the limit.
I did not know this before.

LDPC - http://en.wikipedia.org/wiki/Low-den...ity-check_code

[Nick_G]

Near enough!! I thought I recognised the name from other boards - I suspect we live in the same town on the Test and the A34. I live in K Park. We must have a beer sometime!! We have enough pubs to choose from - but we'd best be quick before they all close down.

Thanks, Gordon. I may take you up on that It's a small world eh!

You may know that I'm into DXing (long-distance radio reception) and tuners, if you've seen some of my other posts. I have 3 tuners in my current system, none of them with DAB! I sold the Arcam Alpha 10 DAB tuner on eBay about 3 years ago as I didn't use it any more. For digital listening I use a Squeezbox Touch, which, for home listening, is miles ahead of DAB.

Does anyone know when the results of this latest CBA are going to be published? I wonder how much data they'll leave out this time :rolleyes:

Regards,
Nick

[Dave2002]

All in hand. European legislation is in place. About 30% of new cars for 2013 will have DAB either as standard or as option - SMMT says so.

Nearly missed this. Weasel words and numbers again. Is that another way of saying:

About 70% of NEW cars Won't Have DAB as a standard or option in 2013.

Words fail me! ( well, obviously not ... )

[Resurrection Man]

Nearly missed this. Weasel words and numbers again. Is that another way of saying:

About 70% of NEW cars Won't Have DAB as a standard or option in 2013.

Words fail me! ( well, obviously not ... )

These people would have us believe that it is much more than 30%....http://www.getdigitalradio.com/digit.../manufacturers

In reply to some of the posts above suggesting all sorts of wonderful high-tech stuff like satellite radios etc, I think we are in danger of missing the point.

If it ain't bust, don't fix it

While FM might not be perfect all over the country, there are many millions of people listening to it and are perfectly happy with it