Martin Banks, Personal Computer World 01/87 - checked

January 1987

I'm not entirely sure that I like surprises. They can be fun, it's true, but they can so often be nasty little things that sneak up behind you when you're not looking and jab their fingers viciously into your ribs when you least expect it. Letters can be like that.

Down the stairs you come to see a pile of them lying on the mat, threatening you. Cautiously you circle them, trying sniff out the treacherous ones. The rest, in my case at least, can be determined from their general appearance. They will, in the main, be press releases that come in various levels of terminal boredom. Their one advantage is that they will usually contain, in the US corporate sense of the phrase, no surprises.

Sometimes, however, one of them can prove treacherous and do something nasty like set me thinking. I'm not entirely sure that this is something I like, mainly because I'm not sure I'm any good at it.

One of these treacherous surprises came the other week and set the synapses crawling around in an unnerving fashion. It was from a company called Neutral, In East London. It was a draft proposal for a low-cost local area network. That nearly got a "ho hum" and the short journey to the waste bin.

What stopped the journey in this case was a bit that I had somehow failed to read first time through. This was the statement `multi-speed', a statement that had the unfortunate effect of making me think.

Neutral, out of an apparent sense of pure altruism, wants to see its proposals as widely accepted as possible. I suppose it doesn't take too much thought, even for me, to see where their advantage might come if it was so accepted. That however, is hardly the point. What set me thinking was the underlying reason behind its idea, and the fact that it is needed.

It takes not too much observation to see that the current status of the Local Area Network business is something of a dog's breakfast. There are nearly as many different types of LAN as there are manufacturers producing them, and even where systems are supposedly the same, such as with Ethernet, they are often, in practice, quite different from each other.

The problem with any differences, as pointed out by Neutral, is the fact that the LANs are all very fussy about what you hang on them. For example, get a nice and tidy network of PC/XTs all working happily together on something or other and then stick a shiny, new PC/AT in as a workstation. the effect, not always but quite often, will be the creation of some interesting problems caused by the different operating speeds of the machines.

Yet, as Neutral points out indirectly in its draft proposal, the use of these different machines is made possible by simply plugging them, willy nilly, into the same, standard, 13 amp sockets. Wouldn't it be nice, says Neutral, if we could all do the same with networks and the information accessible therefrom.

It was this simple statement that got the grey matter moving that morning. It would indeed be nice, thought I, if it did become possible to take any system of any operating speed and plug it into a standard type of socket and get communicating. It would certainly be a potential boon to the smaller network users, who maybe don't have the financial clout to specify all the expensive bits and pieces needed to make a single-standard network operate. For them, the ability to link together the most cost/effective devices regardless of their speed, could be a very useful trick.

I have no way of knowing whether the Neutral proposal is the best solution for this task. It is however, worth an airing, for I do feel that this is an area of that thorny subject, standardisation, where the advantages of commonality far outweigh any restrictions on technological development.

The Neutral proposal itself is based on the RS485 standard for multi-drop communications. The idea is for a bus-based system using RS485 based around 6-way connectors that are similar, but not compatible with the new British Telecom connectors. These would link two or three twisted cable pairs, with one of the pairs running a balanced clock. This is an important bit, because it is this that will allow slow-speed devices to communicate with high-speed ones.

According to Neutral, the system would indeed be low cost as the key component would be a single chip serial communications controller. This means an interface card for a PC could cost as little as £50 (built in, the system could cost as little as a tenner). The wall sockets cost around £4, and connector cables could be made for around £3 each. The hand tool needed for this costs a fiver.

This means that devices could be added to a network for £60 maximum. Yet this would in practice be less, for though the maximum number of nodes per segment allowed is 32, it is quite permissible to connect in more wall sockets than that, allowing users to put in anyone of their 32 devices at the most convenient point between the job in hand and a wall socket.

And this would work using an interesting little trick called Dynamic Node ID. The interface, once plugged into the network would simply start sending out questions...is anything Number One? This questioning is continued in numerical sequence until there is no affirmative answer. The unit then adopts that node number as its own. Physically, it doesn't matter where it sits.

The important bit for this system is the multi-speed aspect. Neutral's suggestion is a minimum speed of 250K bits per second with a maximum frame size of 64 bytes. The maximum speed allowed under RS485 is 10M bits per second, but most high speed devices can be made to run at slower speeds.

High speed systems will be able to talk to slower ones, therefore, by enquiring of the recipient what speed it runs at and working accordingly. the reverse, slow system communicating with fast ones will be automatically controlled by the clock lines of the network.

I can't help feeling it is to the credit of Neutral that it has offered its proposal to anyone who wants to contribute to the idea. It is not saying that their's is the best or only suggestion, only that something of this sort is needed and needed soon. I can't help but agree with them.

end