During the 1980s I worked for a company who, in order to protect the guilty, shall remain nameless. Suffice to say they were then, and are still today, one of the main suppliers of alarm and security equipment in the UK. In moving on from Dymar, I was employed to develop RF-based products. That meant wireless alarm systems for domestic and small-to-medium business use. The government had recently released a VHF channel at 173.225MHz for unlicensed use for this kind of application, so the opportunity was there to make some unique products.
The basic point was that alarm systems, as retrofitted to a home installation, needed lots of cabling which was time consuming to install and had to be done very neatly to be acceptable aesthetically. Clearly a wireless system had many obvious attractions. Trouble was, RF expertise wasn't all that easy to come by and was considered something of a "black art" as a result. Naive attempts at making workable wireless systems were doomed to failure for a variety of reasons:
GIven that only one single narrow-band channel was available, modern approaches such as spread-spectrum were not really available for consideration, and in any case the technology was immature and much less well understood. In addition, it was still the age of only medium scale integration with respect to RF, with most "front ends" for a receiver being discrete components and likewise for most of a transmitter. This made receivers in particular hard to miniaturize, so while transmitters, being simpler, could be quite small, they couldn't be coupled with a receiver in order to 'listen out' for other devices on the system - in other words it was a given that transmitters had to transmit "blind". In turn this leads to other problems - what if two transmitters transmit together? Result - interference and very likely complete loss of those messages at the receiver.
The first wireless system that the company marketed was a design sourced from an unrelated company in the USA (ITI), modified in several respects for the UK. Most of the modifications had to do with its built-in dialler, which had to be approved for connection to the UK telephone system (BABT), so much of the engineering effort was expended on this, with very little on the radio side. The radio communications aspect of this system was woeful. The receiver was of the "super regenerative" type in order to keep it cheap to manufacture, but its performance was rubbish. Such designs had ceased being popular in pre-war valve sets, and the far superior "super heterodyne" type of receiver was pretty much standard - though more complex and slightly more expensive to make. This receiver design was at the heart of this system's crappiness, but it didn't stop there. Transmitters were awful too. Each one was about the size of a bar of soap and housed in a hard-edged beige plastic box. Aesthetically it was a case of "ok, you've lost the ugly wires but you've got these ugly beige boxes on all your doors". So much for progress. The wireless protocol consisted of transmitting just 11 bits of information using a hideous on-off keying method. Transmission was so slow it took about 1.5 seconds for each message to be sent, followed a few seconds later by a second transmission of the same message "to make sure it was received". Hmmm. The on-off keying created so many spurious emissions it was a jolly good job that transmitter power was only 10mW! Combined with a nearly deaf receiver, range was limited to only a few tens of metres.
The slow transmission, long key-up time and message repeating meant that this system was a failure waiting to happen. Clash between transmitters wasn't just a possibility - it was virtually a certainty. Its developers didn't really get this - their solution to the problem of clash was to add yet a third retransmission! That only made things worse.
The reason for going into this awful piece of shit is this: it laid down the expectations of what a radio system was inherently like in the minds of our management. Everything that followed always seemed to come back to this. So in trying to develop a better product, we had to fight the legacy of this one with management at every step.
Shortly after that time I had the great fortune to work with a very talented engineer called Graham Sharples, who understood radio as I did, and recognised that a whole new approach was necessary. This came about because of a relocation of the R&D branch of the company and a merger with another company. The result was a much larger R&D department and a pooling of engineers and ideas from the two companies. Clearly, radio as a concept was attractive, and between us, Graham and I managed to convince management that a completely new design was needed developed with sound radio principles in mind from the start. We had more or less a clean sheet and no legacy to support so it was a dream opportunity. The ITI system had failed to make any significant sales - in fact many were returned in anger by customers who tended to recognise a dysfunctional product when they have to use one. Modern Windows users being the exception that proves the rule.
Fired with enthusiasm, we quickly established a very reliable radio system for the basic communication, based on a VHF FM transmitter and correctly designed superheterodyne receiver which was very sensitive (0.2µV) and reasonably selective (cost was a much bigger constraint compared to Dymar for such designs, so expensive crystal filters couldn't be justified, but we got by with simpler 2-pole filters and a second IF stage using ceramic filters). Transmitter power was still limited to a very modest 10mW, but with a much more sensitive receiver and correctly designed antennas, ranges of 500m were easily achieved. In terms of getting data back to the receiver, a protocol was developed that used a 110-bit format at 1200 baud FFSK, giving a transmitter key-up time of under 100mS. Compared to the 11 bits of the original, it was a luxurious amount of data, and the much shorter transmission led to a number of immediate benefits. Battery life was vastly extended, and the chances of clash were much less likely. The protocol itself used a variety of techniques to eliminate clash, with some error correction and redundancy in the message itself, and a randomising element to avoid transmitters becoming synchronised over time. However, it was complex enough that a custom IC was needed for the sensors (the prototype board ran to something like 30 CMOS ICs). At that time, cheaper ways to implement custom chips were coming on the market, and the then-new 1 micron process offered significant power savings. So somehow we convinced management to give us the go ahead to develop the sensor chip in partnership with a custom IC vendor (ESS).
Development progress had been so rapid that immediate management were largely in the dark about what we'd achieved. So there was an engineering meeting at which we presented our design. This is where the entrenched thinking based on the ITI system started to rear its head. Our transmitter and sensor chip could run off 3 volts, and the low duty cycle and quiescent current of the chip meant that we'd worked out that a battery life with a 3V lithium cell would be a minimum of seven years. This compared with 12-18 months for the 9V PP3 (MN1604) -powered ITI transmitters. Seven years was long enough that it allowed the entire sensor to be disposable, with the battery just another soldered component. And since the battery was not user-replaceable, it meant that the whole sensor could be made small and put INSIDE the door frame, so it was goodbye ugly beige boxes. It was obvious to us this was the right way to make the sensors - small, cylindrical so they could be installed into a drilled hole, and that also allowed a much longer and more efficient antenna. Fit and forget - after seven years replace it altogether. The boss couldn't get his head around this at all. He started from the "given" that the power source had to be a PP3 9V battery that was user replaceable. Never mind that 6V had to be thrown away as waste heat. Never mind that doing that cut the lifetime to 2 years (and hence forced user replacability and hence external mounting on the design). His "contribution" was to spend a weekend in his garden shed whittling out wooden models of his "vision" of what these sensors would physically look like. He was very proud of his efforts, but guess what? Beige boxes about the size of a bar of soap. Only with more rounded edges. Well, that's innovation!
We realised that if we were to make sure this project ended up being what we wanted and not hamstrung by entrenched thinking, we'd have to be cunning. So from then on we had the real project more or less going on as a hidden effort, one of several so-called "black projects", and dummy "public" versions that incorporated all the boss's Heath Robinson ideas. Most of the real work got done out of hours. It was no way to develop a product, but we knew we were right!
Another aspect to the design was the receiver or main control unit. At that time most of the company's products used a pair of 7-segment displays as a user interface - everything was displayed as 2-digit codes (with a few "letters") which had to be looked up in the manual. Since at that time ('86) I had recently been introduced to the Mac and its friendly GUI, it seemed to me there was a better way to do it. We could have an LCD panel with iconic symbols representing the different states of the system, with each sensor having its own state indicator and so on. The layout of the LCD was fixed (not a bitmap display, which was very expensive at the time), but because the symbols were high resolution and could be whatever we wanted, it seemed a good way to make a much simpler to use system. We got as far as developing a simulator for the user interface (in MS-BASIC on the Mac) and talking to manufacturers about pricing and volume and so on before the boss got wind of it. Again, he just didn't get it. Even though it was no more expensive than the current approach, and far easier to use, he couldn't get his head out of the 2-digit display. I'm sure he still uses DOS to this day. Another conventional wired system that was being developed at that time by a different set of engineers was adopting a 1-line text display, so it was decided that our system had to have the same "for consistency". Like that matters - how many people have more than one alarm system in their home they need to know how to use? Well, we lost that battle so a 1-line text display it was. If you buy an alarm system today from that company (or any other, pretty much) you'll generally get a 1-line text display.
As we neared the productionisation stage, the next dinosaur to get involved was the purchasing manager. He had to make sure that we could build it for the right price. One of his concerns was that at that stage we only had a single source for the custom chip we needed, and he wasn't happy about the unit price. In fact it wasn't that bad as I recall (maybe one pound per unit at 1988 prices), and as we were very much technology led with respect to the power consumption and so forth there simply wasn't another supplier available. Unfortunately the purchasing manager just treated our supplier (who had been very closely involved in the design of the chip, and had provided a great deal of effective R and D for free on the basis that they would recover the investment in unit sales) as just another ten-a-penny supplier, and tried to screw them down on cost. They gave a little, but obviously not too much, they needed to make a profit on the deal too. So in the end they got very pissed off and refused to budge. A clash of egos ensued - the purchasing manager "I can always get it cheaper!" refused to buy from them unless they gave him a better price. We told him it had to be them - they were the only one that had the tech we needed, but that cut no ice. That stalemate lasted a while, but we assumed it would eventually be resolved, because after all, there was no alternative in the short term. (It was obvious to us that there would eventually be alternative suppliers, but for the initial production we'd only have one. But he seemed to think that backing a single horse was a gamble too far, even though ESS were highly successful and still exist albeit as part of a larger conglomerate. By the way when we developed that chip the head of ESS was Robin Saxby (now Sir), who got closely involved in our work, and who went on to an even more stellar career, founding ARM. So I'm not talking about some cowboy outfit - in fact it was embarrassing, it was obvious they were way too good for us, especially when the P/M started to play silly buggers.)
Well, it turned out there was an alternative - complete cancellation of the entire project. In a fit of pique, the purchasing manager effectively pulled strings (he was the MD's golf buddy, for example) to get the project cancelled on the basis that it was too expensive to make because we were locked into a single supplier. We knew the costs, so we knew that it was competitive with every other system we made, but the simple fact was that the purchasing manager just didn't like being told what to do. And given the old boy network that operated amongst the management, he got his way. Talk about cutting off your nose to spite your face.
As engineers we were completely demoralised by this. The boss immediately set in train a new project to develop a radio-based system (they still needed one) but made in his image. Thus it had a 2-digit 7 segment LED display. He commissioned an outside developer to develop a receiver for it - this was madness, our receiver was perfectly functional, already fully developed and debugged, and had much better performance than the external developer's effort (who had some experience with broadcast receivers but none with narrowband data receivers - his design looked like a broadcast receiver with some added filtering). And naturally the transmitters were PP3-powered beige boxes using some very crude signalling that didn't need a custom chip. They clearly hadn't learned a thing from the ITI farce. We were expected to enthusiastically get behind this bastardised version of our much more innovative design. To say that morale became very low and work slowed to a crawl is an understatement. Graham and I spent nearly all our time just working on whatever took our fancy - we developed a quite neat radio tracking device that could have been used for security van tracking applications, had the company seen fit to get into that - and some very easy to use software that interfaced to one of our sheltered housing receivers. But it was clear that nothing we did was what the company management could understand. It seemed to us as if we were forever doomed to sit on the trailing edge of the possibilities.
Both Graham and I left at about the same time - maybe a year after the cancellation of the project. Graham went on to found the company Radiometrix , where he basically redeveloped our transmitter/receiver designs into standalone OEM modules that could be used in any radio data linking situation. Radiometrix thrived and is still going strong. I went to another company, despite offers to go and work for Graham (maybe I should have done that!). Curiously, our boss also left not long after we did and set up his own company making... radio alarm systems. This was odd, as we knew he knew next to nothing about them! A few years after he started, he apparently gave up on developing his own RF elements using the same consultant that had designed our 'broadcast receiver +', and started buying them from Radiometrix! I have no idea whether his endeavour succeeded in the long run.
It's fair to say that if you're a bright young engineer looking for cool stuff to work on, the security industry isn't likely to be the place for you. If you really do want to work on alarm systems, start your own company if you want to be able to innovate. I've become a firm believer in the idea that alarm systems don't work anyway - better off getting decent door and window locks. While I have met some excellent engineers in my time, most of them weren't while I worked there. Graham stands out as being far and away much better than the average, and his subsequent efforts proved that. For a while after this debacle I was somewhat bitter about it, though once I moved on, it ceased to matter. This is the distilled version from 20 years on and many of the details have largely been forgotten. It may sound like we were just pushing technology for its own sake without reference to marketing or any other input. That's not the case, we had many meetings and discussions with others in the company who were crying out for this stuff. We had a very close relationship with the head of marketing who was 100% behind our design thinking.
I'm fairly sure our system would, at least within the limited importance of the security industry, have been a great success and a new model for how security systems could be. Much of what got developed around that time became the model for systems that persists to this day. The main opportunity for technology-led innovation was around then, and what there was of it came from pre-existing work done by the company we merged with. Our boss was a widely despised and ridiculed figure (among several) for his obvious embodiment of the Peter Principle, though occasionally a nice guy outside of work. One of his engineering contributions to the company's product line was an automatic dialler with voice output - provided by a custom "flexidisc" vinyl record! That was designed in 1982 when speech synthesis was already readily available. Another was his simple wired alarm system that became a perfect medium-wave transmitter when the sensors were wired up via longish lengths of cable. To say that he didn't understand what he was doing is very much an understatement. Real engineers couldn't respect him, and pretty much all the problems and farce that followed stemmed from that basic fact. It is telling that during my time at the company, we had a very high turnover of staff at the level above us who reported directly to this guy - at least four as I recall.
© 2006-2008 Graham Cox
