That's some decent BBC engineering from a team that works away quietly in the background and just makes things happen, or keep happening. I don't work in that field, but like many people in my profession (Systems and networking), I also tend to be invisible until something breaks, and then I am the fall guy if it's not fixed within a heartbeat.
/I was half expecting the article to outline how a box full of 80s technology had been replaced with a Raspberry Pi!
It is a bit odd that they shoehorned in an existing commercial board rather than just put the FPGA on the main board or even designed a custom daughter card if upgradability was a concern. As it is, they are now tied to the availability of boards with the same board-to-board connector and a similar form factor.
They used a commercial board that implements the Vita 57 bus, which is the standard way to interface with an FPGA [1]. So if they need replacements down the road, they just have to get one that implements the same standard, and standards for industrial equipment tend to stick around for a considerable amount of time.
The engineering team has a budget that big? Do you know what the budgets are for the other aspects of the BBC? Like TV and radio broadcasting, production of TV and radio shows, administration, HR and other areas?
Europe average R&D spend cross industry is just above 4%, but that's mostly pulled up by R&D on healthcare (which is closer to 18%). Without healthcare I think it's around 2%. I would hope to see them spend at least £100m on R&D, but that's obviously not just on broadcast technology.
This is one of the most interesting things I've read on HN in weeks. Awesome to see this kind of success story, and it's super fascinating to get insight into how radio works behind the scenes.
I'll be honest, I had no idea all of this goes into radio. I thought there were just, err, towers?, that broadcast... radio waves, and that was kind of it? I mean, I knew on some level it had to be a tad bit more complex than that, but you never see or hear about it.
Compared to U.S. radio in my area, this is indeed fairly complex. In my "market" we have mostly local stations and the complexity is minimal: rent office space for offices and DJ booths, send mix to a directional transmitter on the roof pointed at the tower site, tower decodes and blasts a 100K watt signal.
Of course, Clear Channel also has lots of robot stations in many markets in the country, which seems like it might approach similar complexity. However, of the few I've experienced, Clear Channel don't handle the complexity well and there's nowhere near the BBC's level of professionalism: they'll overlap commercials and songs for the entire duration, drop off the air (no signal) or broadcast dead air for 15min at a time ...
That's pretty much how it works in UK independent local radio (ILR).
There's been a fair bit of change over the past decade, resulting in a lot more network programming and automation amongst stations established prior to 2000 or so. Core programming hours can be 7am-6pm, with local presenters, news, sport, traffic and weather, but that's it. Post-2000 stations seem to be a lot more local, they only appear to serve larger cities - smaller towns are stuck with carbon copy stations. Specialist and talk radio stations are virtually non-existent.
It sounds like Clear Channel's problem is with automation and production. It's really not that hard to automate a station such that it sounds like it's being produced live. I would take a guess that they're using some early 90s - maybe even 80s - tech to drive their automation.
One thing you've got going for you in the US is local TV. It's something that's only just started to be done in the UK and it's pretty amateurish stuff. I can't see it succeeding - wrong era and stations exist on their own channels, rather than as network affiliates.
"Secret orders to the captains [of the nuclear subs] say that these deadly instructions are to be opened and acted upon only if the submarine cannot tune in to Radio 4’s Today programme for a given number of consecutive days. That is a reliable sign that Britain has been hit by a nuclear attack."
Que new James Bond plot, the bad guy disables radio 4 to launch nuclear armageddon - Bond in race against time to find the spare parts needed to restore the radio.
The operative words there are "it has been claimed that".
This may have been true in the past (although really no official confirmation has ever been given), but it most probably is not true anymore, as the valves of the 198 kHz transmitter at Droitwich are of a design that is not available anymore and there is a limited supply left; when that is depleted, the transmitter will shut down.
I've always been impressed by the BBC tech people I met, doing a high-quality job in an extremely bureaucratic neophobic environment.
As part of making the unit, we had to ensure it complied with all the relevant European directives so that it could be CE marked. We had the Codec tested for both safety and EMC.
Interesting, I thought this wasn't required for internal use; probably the BBC corporate structure means it no longer counts as "internal".
On "Radio Teleswitch":
[Droitwich] transmitter uses a pair of obsolete metre-long valves which are no longer manufactured anywhere in the world.[1]
In October 2011, the BBC admitted that the Droitwich transmitter, including Radio 4's longwave service and Radio Teleswitch, will cease to operate when one of the last two valves breaks, and no effort would be made to manufacture more nor to install a replacement longwave transmitter.
This is like the Domesday Book (1980s interactive multimedia system on Laserdisc): high-tech and sui generis at the time it was built, but eventually uniquely obsolete.
The BBC itself is very much a valve-and-laserdisc organisation in a Google world. We huddle round the warm glow, concerned that eventually some vital element will give out due to lack of money and the whole enterprise will give up.
Periodically someone comes up with the idea of shutting down analog FM radio. This is politically inconceivable as half the country has ancient radios that were tuned to Radio 4 in 1967 and have never played any other station (an exaggeration, but only a small one).
CE carries a number of subclasses that can be self-certified.
With respect to EMC and safety it is absolutely in their best interest to conform, not merely out of a bureaucratic obligation but out of a need to minimize any interference from or to other equipment housed nearby.
to allow electricity suppliers to switch large numbers of electricity meters between different tariffs by broadcasting an embedded signal in broadcast radio signals
I kind of heavily doubt these are signed broadcasts that cannot be replayed..
I believe that operating a rogue transmitter in a licensed band would get you located rather quickly - even though the tech itself may not necessarily be secure, there are other enforcement mechanisms.
The obvious benefit for a malicious person would be to set their meter to the cheaper tariff by replaying the signal.
It wouldn't need to be a powerful signal - if you were broadcasting right next to the meter even a few milliwatts would be enough. Build a faraday cage around the meter to be sure! Just remember to dismantle it before the meter reader comes round.
Surely you only need the rogue transmitter to be stronger at the receiver (your electric meter?) than the signal from the transmitter. That doesn't seem like a strong enough signal to easily locate you?
Seems a complicated way to do something - doesn't Economy7 operate between fixed times overnight?
I wonder if they can cut power using the transmissions - seems like something that could be used for emergency control of demand in times when the National Grid has problems.
Hm, that's probably the point: there is no threat (to the utility provider) in that model. Worst thing you could do is make someone's appliances run with a more expensive rate; hardly a likely event or one worth preventing.
(As for power cutting, that seems unlikely - the system is in place to provide larger power consumption when base production exceeds the load. Around here, the basic low/high rate info is provided by https://en.wikipedia.org/wiki/Load_management#Ripple_control , which needs a not-really-smart power meter, but runs on top of the power line, obviating the need for a side channel)
The smart metres that are being rolled out at the moment have a remote kill switch, although it couldn't really be used to implement rolling blackouts as you have to physically hit a button to reactivate it.
That's a DoS waiting to happen, IMNSHO. If it's easy to transmit on the frequency locally at low power and there's no auth, a drive around the neighborhood could create a blackout at anyone's demand.
At least in the UK, smart meters don't use radio-based time switching as they have their own clocks which are synchronized with their remote system. All commands to these smart meters (including the ability to remotely disconnect them) are cryptographically protected to ensure integrity and authenticity, and also have anti-replay protections.
Disclaimer: I've worked for a smart metering manufacturer.
I was under the impression that most if not all meters of any kind (electricity, gas, etc.) have GSM modems that not only allow for this kind of signalling but also report their reading to the energy company.
I'm surprised BBC is not only willing to make custom hardware, but custom RTL design as well.
I'm curious what the failure rate of the FPGA will be, I mean they are more susceptible to soft errors than CPUs or ASICs. Maybe BBC will fund making a custom ASIC :-) Well I see two systems, maybe they are redundant that way.
Why not make the project open source? Put it up on github.
I would doubt an ASIC would be worth it, giving the number of organisations needing this CODEC would be relatively low. An ASIC run is expensive and even more expensive if there is subsequent error in the design.
FPGA makes sense for this application. It can always be re-flashed.
They should at least enable the SEU mitigation circuit, which repeatedly performs CRC of the configuration data. Have it reboot if it detects an error. Maybe they already do this.
Very nice work. What is the rationale behind doing a bespoke design rather than sharing the development cost with all the European broadcasters, is NICAM unique to the UK? What do other broadcasters use for similar services?
I think NICAM as a distribution system to FM transmitters is a BBC invention, and presumably all the hardware, physical interfaces etc were designed in-house. So unless they managed to sell it to other countries they're probably the only ones using it. It later found another use for transmitting stereo audio to household with analog TV but that's basically dead along with analog TV and isn't quite the same anyway.
https://en.wikipedia.org/wiki/NICAM seems to indicate that while NICAM is used by a few broadcasters here and there, it was initially developed by the BBC, and it's no longer as widespread in Europe as it once was.
See makomk's comment nearby: NICAM's use for stereo sound with analogue TV broadcasts is not the same as the use for the BBC's internal audio distribution system.
Cool to see that there are still companies that can and will make devices like this in-house, but yes it just seems like something that a radio station would normally just do with an off the shelf solution.
There seems to be something uniquely British in its engineering culture which allows for the left-field hack-it-simply-together-and-hope-it-works approach.
I'm guessing that the initial build was exactly this. Maybe the equipment is highly specialised for the UK, and other countries have their own standards?I imagine these custom standards are hard to just rip out and replace. It is also hard to convince a bunch of engineers who know a set of tools, and proprietary formats and hardware to change to something they don't know, with something that is so critical to a nations infrastructure.
Somewhat related, a few years ago I saw a video of a guy starting up a "live TV studio" from 70s/80s I think at his cottage/cabin. It was quite interesting, unfortunately not in English. I'll try to find it anyway and update this post.
I appreciate a good success story about major transitions / upgrades / process changes / etc. Those situations where there is usually SOME glitch and you have to engineer the hell out of it to get a perfect transition. The only reference to testing was "We had the Codec tested for both safety and EMC." Was that in house or external? Surely there was some major transition testing low power broadcast tests, etc to make sure everything went smoothly. Or did they just get really lucky? Seems like there's more story to this.
I read the article with interest, waiting with anticipation for them to announce the resulting hardware and software had all been open sourced. And then...nothing.
I'm surprised, I thought in the UK these kinds of taxpayer-funded projects were required to be open source.
In any case, it's too bad. I was not familiar with the NICAM codec up to now, but I'm sure that software and hardware plans would have been very useful to some developing countries, many of whom apparently also use NICAM in their state-owned broadcast companies.
> I thought in the UK these kinds of taxpayer-funded projects were required to be open source.
There is no law, afaik, and even the BBC has commercial interests. iPlayer has never been open-source, for example.
The BBC has obligations in terms of content reach and content quality, but they're not particularly bound to specific ways of achieving those targets (beyond some stuff about not abusing their position in commercial terms). Actual engineers are mostly pro-OSS and try hard to push for it internally, but they have legal obstacles similar to any other company.
They aren't using PATA or SATA. They are using ribbon cables to connect boards to each other and to backplate connectors.
Some of those ribbon cables could be carrying serial signaling on some lines but to get it all on one smaller cable with fewer signaling lines like SATA you'd need to mux/demux the data on each end which would add expensive chips to save money on cheap cables.
They most likely aren't ATA cables, as in, they aren't used to connect hard drives -- so it's not obvious that whatever purpose they serve could be served by a serial alternative.
If you look at the photo of the back of the unit, you see a lot of very wide D-Sub sockets that these ribbon cables seem to connect to, so the unit is meant to connect to a bunch of other things. Actually, the need for these sockets seem to account for most of the size of the unit, the box appears pretty empty.
If you're designing your own boards with ribbon cables connecting them, I imagine that you could save a lot of money on parts by using 40-pin connectors with PATA cables--we're talking entire cents, for each of the 2 units.
Also, I'm willing to bet a place like the BBC has boxes filled with spare PATA cables and 16MB SIMMs and 8 GB internal hard drives, and maybe even a room with a bunch of dusty 17" CRT monitors in it. And sometimes, rather than fill out a purchase order, one of the engineers descends into the hoard, returns with a cobbled-together frankenserver, and installs NetBSD on it.
So they might actually be PATA cables using PATA connectors, but the signal between cards is not ATA.
Exactly. And it has to connect to the same system as the original 1980s units, so whatever protocol it's using will be old. DB25-serial? SCSI? IEE-488? It would have been nice to have a few technical details on the system.
Random BBC-related anecdote: the Cambridge University computer lab had, up to 1999, an info TV in the lobby driven by a 20 year old BBC micro in Teletext mode. This was apparently because it had an extremely powerful video output (more so than modern hardware) and the display was at the other end of 50m coaxial cable.
If it was a scavenged system, I'd be apt to believe you, but a new build like this, you almost certainly would use new ribbon cables. In bulk, the ribbon is about fifty cents a foot, so reusing old parts wouldn't be worth the trouble of having to troubleshoot problems down the road.
You can easily see from the pictures that the ribbon cables are different widths. It looks like every ribbon connects a PCB to a connector on the backplane, most of them D-subs. The photo of the backplanes as installed show that most of the connectors aren't even being used.
By coincidence, the large-sized cables have 40 conductors. The medium ones have 25 or 26, and the small ones have 15 or 16. Those are all standard sizes for ribbon cable, and you can likely buy new cables of the correct sizes off the shelf for less than $10 each, or make them yourself at 20% the price or less.
In comparison to the overall cost of engineering the box, the ribbon cables are just a drop in the well. Nevertheless, about 12 of the ribbon cables in the box could theoretically be replaced by old PATA cables and 35 million people would still never know. The potential problems would be that some ATA cables block pin 20 as a mechanical key, they would be longer than strictly necessary, and ribbon cables are obnoxiously effective antennas.
The 40-pin ribbon cables as installed could almost certainly be used as very short PATA cable replacements, with a little mechanical modification of the connectors.
Those look like custom digital connectors moving data to and from the codec boxes. They're not connecting hard drives here.
If you had to connect 40 low-speed signal pins from one board to another, you'd do it as simply and cheaply as possible. Serializing them doesn't gain you much of anything, other than complexity.
Really interesting article, but terrible clickbaity title. I wouldn't have read it if there weren't so many positive comments on HN already.
It could work if there were a subtitle. "35 million people didn’t notice a thing: BBC Radio's NICAM Codec replacement project". I dunno. But it is (or should be) very important that the title actually informs you as to what the article is about. "35M people didn't notice a thing" doesn't tell you anything at all.
Edit: Upon further consideration I may have been a little harsh. This is really a blog entry more than an "article" per se, and I suppose that's fine for a blog post title. The problem comes when we link to it from somewhere else, e.g. HN; we need the additional context.
Perhaps the title-writer has been infected by the BBC news editorial policy. They seem to be turning clickbait titles into a fine art, combined with renaming articles so you think it's a new piece on the same story, but turns out you've already read it. (If you read BBC News at all that is: my wife still does, I gave it up years ago and haven't looked back.)
I love many aspects of the BBC but sadly not their news output anymore.
(Sorry I realise this has very little to do with the R&D article.)
I used to think the BBC news was definitive 5-10 years ago. I'm not sure what has changed? Have the recent governments and their pressure on the license fee really made the difference or is it just my perception?
When I was young the BBC was always seen as pretty left wing (it certainly wasn't at it's inception though) it seems pretty conservative (and sensationalist) right now.
Both sides of the political spectrum regularly complains about the BBC being biased towards the other side. It's the best evidence that whatever bias there is either is minor or short-lived.
In my eyes, the biggest problem with the BBC isn't bias, but that they're terrified of seeming too biased, and as a result they're wishy-washy in ways that ironically makes them seem biased.
E.g. the left will accuse them of being right wing because they don't write in more critical terms about the governments willingness to deal with the Saudis, while the right will accuse them of a left wing bias for not writing in more critical terms about Chavez, and so on. Similarly they're often accused of bias for not reporting something, or very carefully reporting something, that either side believe to be fact, but where there's not (yet) actual firm evidence. Of course, if done "against" just one side, this too would be actual bias, but the BBC appears to be pretty consistent about it.
So we're left with reporting that is rarely very biased, but often exceedingly unsatisfying for everyone.
I'm not saying there's no bias - I don't believe that would be possible. But I do think that the allegations of bias against the BBC tend to be wildly exaggerated.
There was a comment here a few months ago which mentioned a term. I have forgotten the term since but it effectively described a situation where an expert in subject matter foo reads an article about foo in a well-regarded news source (like say the Economist) which is either pretty incorrect or pretty biased and then the subject matter expert goes "wow they know nothing about foo" and then turns the page and reads about another topic let's say bar (or maybe let's say Middle East politics) which our subject matter expert knows little or nothing about and then our subject matter expert says "wow, this news source is very insightful".
If we read about articles on key escrow or actually anything about cryptography and surveillance in general, we will find it hard to see BBC is accurate (neutral is not the same as accurate because to be neutral about anything -- and here comes Godwin's law -- involves saying that there was a legitimate excuse governments to act in a way that is not beneficial to millions of its citizens. Anti-terrorism rhetoric is way overboard and news outlets like BBC are fully on the bandwagon (although thankfully much less so compared to CNN, Fox News, and other low lives). Honestly, I think our fear of terrorists has caused more damage to the world economy than all the "acts of terror" combined. Ordinary people should not be afraid or even mindful of "acts of terror" in their daily life any more than they need to be afraid or mindful of a meteor shower hitting them in the head. Everyone goes over the top with "security precautions" because they don't want to be seen as doing nothing. The point here is BBC does not preserve this neutrality at all. I bet even CNN and Fox could do better if they were organizations that has their funding guaranteed, as opposed to being under the mercy of its advertisers. Sorry but what BBC does is nothing exemplary for someone in its shoes.
> Sorry but what BBC does is nothing exemplary for someone in its shoes.
I didn't suggest they were exemplary at all. On the contrary, I criticised them for not doing a proper job because they're worried about coming across as biased.
They've chosen to deal with criticism of bias by essentially trying their utmost to never say or do anything controversial, but the nature of the real world is that a lot of the most important journalism necessarily steps hard on someones toes.
When you just look at the website it is easy to forget just how much current affairs content the BBC produces. Across TV and radio their are numerous news programs with different skews. News Night or PM can sound completely different to Breakfast or Five Live. I would rather individual programs are allowed to show bias as long as the output in general is balanced. It is ridiculous that every single program acts like it is the only thing the viewer will ever watch.
Add things like the Jimmy Savile abuse scandal, and the ongoing threat from the conservative government to turn the BBC into the equivalent of America's PBS (irrelevant on the political landscape, lacking financial heft, and in close to nonexistent viewership) and you have a recipe for a neutered public news broadcaster too afraid to do anything meaningful.
Although it's foreign, I've found Christian Science Monitor to be fairly diligent and insightful. If nothing else, I've found it difficult to identify an institutional bias in their coverage and outlook.
Yep. For me, at least half of the value of an article comes from the (external) internet discussions about it, which summarize key points and weaknesses much better than the author is able to.
I think you're right; when skimming, I read the following section, which made it sound like they were updating it to send additional metadata, based on their mentions of the "codecs" and how they "will" allow radios to do useful things with the metadata. Also, I'm used to codec meaning "software program", not "machine".
>>The NICAM distribution system also carries the RDS data. This will enable the radio in your car to automatically retune when you can get a better signal from a different transmitter. It will also display the station name (for example “Radio 1”) and the current song name (for example “Now Playing: Coming With You by Ne-Yo”). RDS will also indicate when a travel announcement is being made on BBC local radio so that a radio can switch to the travel announcement at the right time.
The BBC seem to have moved to this sort of title with the recent updates to their web properties. They do these things seemingly confusing themselves with a commercial enterprise rather than a publicly funded provider.
That's a small example of what I mean, not the full totality of the threat space.
Building more readership makes them that much more difficult to cut funding to, in a variety of ways (more people to get outraged if the government tries, more evidence to show the government that they are worth the money, etc.). Clickbait headlines create more readership. We may prefer that to be false, but it's obviously not. From there the logic is simple, almost inexorable.
Being government funded does not mean you get to ignore funding issues, it just changes how you play the game. There is no source of unconditional funding in the world.
Also, one might argue, funding aside, that increasing readership (without sacrificing quality below some threshold) is the whole point of the enterprise. Click-bait headlines are annoying, but they aren't the actual body of the report themselves.
The BBC serves a charter which is managed by the BBC Trust. The Trust effectively interprets the charter. Their interpretation of the goals of the enterprise are (http://www.bbc.co.uk/bbctrust/):
"Make the most creative and distinctive output;
Innovate online to create a more personal BBC;
Serve all audiences; and
Improve value for money through a simpler, more efficient, and more open BBC. "
Increasing readership by reducing quality of headlines to match junk media clickbait doesn't appear to fit in there.
"The Public Purposes of the BBC are as follows—
(a) sustaining citizenship and civil society;
(b) promoting education and learning;
(c) stimulating creativity and cultural excellence;
(d) representing the UK, its nations, regions and communities;
(e) bringing the UK to the world and the world to the UK;
(f) in promoting its other purposes, helping to deliver to the public the benefit of emerging communications technologies and services and, in addition, taking a leading role in the switchover to digital television."
I'd argue that this sort of change goes against the stimulation of cultural excellence, the tail is wagging the dog.
It could reduce it, arguing there is a lot of waste at the bbc. As long as the public perceives the value in the bbc's offerings the odds of that happening are reduced.
Maybe people did notice but didn't care enough to complain? Or we're all used to various sound quality depending on what we're doing (cell, skype, voip, etc) that we've just internalized quality changes as not a big deal.
It reminds me how satellite radio was supposed to be "mp3 quality" whatever that meant. Everytime I rent a car with SiriusXM, my god, the sound quality is horrible. Its like they're using those early postage-stamp sized quicktime videos from the dial-ups days. Yet people still subscribe in vast numbers.
Apologies for that, I submitted this and did actually go looking for the HN title guidelines, which seemed to say use the original title. I'll try to be a little braver with title editing in future.
Even if you post a more informative title, the mods of HN will change it back to the original site title. It has happened countless times and is pretty much the site policy. The claim is that is prevents posters from making click-bait titles to gain karma, although in this case the original is the click bait.
So anyway, no, you did the only thing you could do. Anything else would have been degraded to the current information value by mods anyway.
It's interesting you classified it as click bait. Too many attempts to raise interest quickly, in a large number of people with low quality content, results in a small segment of those people "hitting" on the fact it's raising interest quickly and should be flagged. In your case, you "unflagged" it because we started talking about it, which implies some of us shouldn't (or won't) filter long term.
BTW, I agree with your edit. An R&D blog shouldn't be held to the same standard of journalistic excellence as a reputable news source, even if it's the BBC's blog.
/I was half expecting the article to outline how a box full of 80s technology had been replaced with a Raspberry Pi!