Editor’s Note: This “How it Was” story is told by David Ashton, an engineer’s engineer who is a frequent visitor to EELife, EDA Designline, and Programmable Logic Designline where he offers many astute comments and feedback along with his own articles.
In 1980 I started working for SITA – Société Internationale de Télécommunications Aéronautiques – in Harare in Zimbabwe.
SITA was, and still is to a large extent, the airlines data network. In those days it was pretty state of the art. Airlines had big mainframe computers in their home cities and they used SITA’s communications network to connect terminals in their reservations offices around the globe. The terminals were text-only devices and didn’t need very high speeds – 2400 bps (bits per second) was pretty standard and 9600 bps was a high speed circuit.
I’d worked in communications previously, but most of it was voice and teleprinter, so I had a pretty fast learning curve. SITA had sent down a tech from their Nigeria office to help set up the new Harare centre, but he seemed more interested in chatting up our (admittedly very pretty) secretary than showing me what was what. I nevertheless learned quite a bit, and by the time he left to return to Nigeria I had a pretty good idea of what went on.
SITA had sent down a line tester which could tell you everything you wanted to know about a standard 300-3000 Hz audio channel, which our main line to Johannesburg worked over. Apart from a basic frequency response test, the tester measured group delay – it measured how long a frequency took to get from one end of the line to the other, relative to 1004 Hz which was the standard frequency. (Why 1004 Hz and not 1000 Hz I never discovered).
It would also check for noise, impulse noise, phase jitter, amplitude jitter and a few other more abstruse parameters. There was a standard – I seem to recall it was CCITT M.1020 – to which all these parameters had to conform, and if they did the channel would reliably support our modems at 9600 Bits per second. Considering our channel went via a microwave system with something like 39 hops (repeaters) between Harare and Johannesburg, this was pretty impressive. The Codex LSI 96 modems were about the size of a shoebox and full of LSI ICs, and thumbwheel and DIP switches which had to be set just right or the modems wouldn’t talk to each other.
Modem clocking was another skill I had to acquire. A port of our comms equipment might go to a modem, or to a Modem Sharing Device (MSD) which in turn talked to four or five modems. And if the end of one of those modem lines went to an office where more than one device was connected, there would be another MSD connected to more modems and some terminal systems. Each of these devices could be set for Internal clock (it provided the timing), External clock (the terminal provided the timing) or Receive or “loopback” Clock (it derived the timing from the incoming signal). And woe betide you if you got the clocking wrong between two devices. That resulted in data loss when the clocks slipped out of phase with each other, and the airline customers’ terminals would “hang” with no responses. I got quite good at solving clocking problems and was always getting cryptic diagrams sent to me as “text” drawings asking for clocking settings.
Our Johannesburg centre had a 9600 bps link to London. They had a data concentrator with connections to many international airline offices in South Africa, and also a connection to South African Airways (SAA’s) mainframe in Johannesburg. Now when a reservation agent on a terminal makes an entry to enquire about flight availability, for example, it’s usually very short, something like A30JUNJNBLON which would mean “Give me availability on flights on 30th June from Johannesburg to London”. The response, on the other hand, would be multiple lines of up to 64 characters, each line giving date, airline, flight number, cities and times of departure and arrival, and seat availability for all the classes on that flight.
So data INTO a mainframe is very short, and data OUT OF a mainframe is typically quite long. So our engineers in the data centre in Johannesburg did a very neat trick with their 9600 bps circuit to London. Rather than split it into 2 x 4800 bps channels, one for SAA and one for all the other SITA customers, they split it into a 2400 bps channel and a 7200 bps channel. Then they crossed over the transmit and receive wires between the ports. So the SAA mainframe got 2400 bps in and 7200 bps out, and the international airline offices connected to the SITA concentrator got 2400 bps out and 7200 bps in. Not only the data wires, but the clocking also had to be crossed over. The engineer who came up with this arrangement explained it to me with considerable pride, and rightly so, it was a very novel solution to squeezing the best possible performance out of what was, even then, fairly limited bandwidth. SITA aimed at a response time of 3 seconds from an entry on a terminal anywhere in the world, to the response starting to appear on the screen. We usually got that, and better. And that 3 seconds usually included a couple of satellite hops which put on a quarter of a second or so each.
In 1985 I left SITA (though I went back to them later) and went to Marion Island for a year as a radio technician. Marion is an island in the bottom end of the Indian Ocean, about 2000 Km from South Africa. It is primarily a weather station and the primary comms was by short wave radios carrying FSK teleprinter traffic at 50 Bauds. We had an encoder called an Autospec, a fiendishly complicated device full of logic ICs, which converted the 5 bit teleprinter code into a 10-bit code with error correction built in. Even then, when conditions were bad, even the most usable frequencies necessitated several repetitions of the data (the messages were pre-prepared on punched paper tape). To give you a laugh, I have given Max a photo of myself in my Marion days. I think his hair length beats mine, but for wildness and wooliness I think I take the cake. I certainly do for greyness of hair now! (I think Max dyes his….. ;-)
David age 29 in 1985 on Marion Island (left)
and as a distinguished engineer today (right)
How far things have come in just 25 years. Before I left SITA in 1993, we were using 19200 bps modems on all international lines, and a 64K ISDN channel between South Africa and London. Many airline offices were using 9600 bps or faster modems on local circuits, and instead of terminals were using PCs networked with Coax or twisted pair at 10 Mbps, with “gateway” cards communicating with SITA. One of our concentrators used the original Ethernet – thick coax cable using large N type (I think) connectors – for communication between its modules and console. Marion Island has now got a 128 Kbps internet link for their comms – that is not fast by today’s standards, but it’s some 2500 times faster than my 50 baud teleprinters!
I’ve been with my current employer in Australia for 10 years. When I joined we had a microwave system between our main offices that had 16 x 2 Mbps channels available. That’s still carrying traffic, but is slowly being phased out – we lease bandwidth of 100 Mbps for these links now, and our rural offices which had one or two 56 K modem links or maybe a 128K ISDN service for data now have 2 Mbps ADSL services, and we’re considering upgrading those to ADSL2 which should give us 8 Mbps +. If we were to consider upgrading our microwave system we’d be looking at 155Mbps Ethernet links at a minimum. Links around our main offices, and even within towns, are on optical fibre at 100 Kbps or even 1Gbps. We still do use serial ports but USB has simplified things so much – no more worrying about clocking, or even bit rates, parity, number of bits, number of start bits…..
When I started, if we had to log in from home to check a fault, we’d use 56K modems. We progressed through ISDN to “air cards” – basically a mobile phone used as a data link. They used to connect on the serial port, then by USB – now they’re built into our laptop PC and we just slip the SIM card into a holder under the battery. I just used mine today, and the connection speed I got seemed well-nigh as good as my 100 Mbps Ethernet cable connection back in the office. At home, my wife’s laptop connects wirelessly to my ADSL router at 54 Mbps (and that’s slow today!) I sometimes wonder what we do with all this seemingly unlimited bandwidth, but then I look at downloading MP3s, You Tube videos, movies, indeed at some of the great tutorials available on EETimes, with large detailed diagrams, that download to my screen in the twinkling of an eye. Often even faster than SITA’s 3 seconds!Click Here
to see other articles in this "How it was..."
series...Editor's Note: It would be great if you took the time to write down short stories of your own. I can help in the copy editing department, so you don’t need to worry about being “word perfect”. All you have to do is to email your offering to me at max@CliveMaxfield.com with
“How it was” in the subject line.I can post your article as “anonymous” if you wish. On the other hand, what would be really cool would be if you wanted to add a few words about yourself – and maybe even provide a couple of
“Then and Now” pictures like David's above.
If you found this article to be of interest, visit EDA Designline
where – in addition to blogs on all sorts of "stuff" – you will find the latest and greatest design, technology, product, and news articles with regard to all aspects of Electronic Design Automation (EDA).
Also, you can obtain a highlights update delivered directly to your inbox by signing up for the EDA Designline weekly newsletter – just Click Here
to request this newsletter using the Manage Newsletters tab (if you aren't already a member you'll be asked to register, but it's free and painless so don't let that stop you [grin]).