Sixty years of electronics

Citation

Ellis, B. (2007), "Sixty years of electronics", Microelectronics International, Vol. 24 No. 1. https://doi.org/10.1108/mi.2007.21824aaa.001

Sixty years of electronics

Sixty years of electronics

When I was invited to write this Editorial, as an accompaniment to my last Internet Commentary for this journal, I thought it would be a good idea to summarise my experience in the last half-century or so and to comment on the numerous changes I have seen.

My adventure into electronics started in the early 1940s when I built a crystal set so that I could listen to the BBC Home and Forces programmes, while lying in bed at night. Fiddling with the “cat's whisker” to find the best spot on the galena crystal was, I suppose, an entry into the world of semiconductors, even if I did not understand why. I quickly evolved into using amplification with an Osram LP2 triode valve (vacuum tube) and then super- regeneration. By the age of 16 I had become the UK's youngest “ham” and had started my studies at the Heriot- Watt University in Edinburgh, to graduate as a Radio Engineer (I think Southampton University started a course in Electronics Engineering, not just radio and telecommunications, a year or two later).

After completing my National Service, I started my professional career with a telecommunications company in Cambridge in January 1954. At this time, the germanium point-contact transistor was still in the laboratory and everything revolved round the valve, mostly with B7G and B9A bases, but also the octal for medium power devices. The smallest passive components were about 15mm long with 50mm axial or radial wires. Most of the wiring consisted of soldering the ends of the wires directly between the tags of the valve-holders, using a massive uncontrolled soldering iron with a 7 or 8mm-wide bare copper bit that had to be filed every couple of hours. The soldering wire was activated rosin-cored Sn60 alloy, 1mm diameter (0.8mm for miniaturised) and this was the only item that had a specification, DTD599A (which is still seen today on some UK solder reels). If I remember correctly, this applied to the flux and limited the percentage of hydrohalide activator that could be added to water- white rosin. Of course, we did not bother about cleaning after soldering, because the uncontrolled temperature of the soldering iron ensured that all the activators had sublimated, despite the high anode voltages, often 250V for non-transmitter applications, going up to 1,500V for the 2kWHF transmitters I was involved with.

The commercial printed circuit was not yet in common use. I had seen a couple of pre-war applications, which could be described as using printed circuits. In one case, it used a moulded phenol-formaldehyde substrate with copper-loaded ink scraped into grooves. The other was my 00-gauge electric train which had a flat commutator of a rolled copper foil disc stuck onto a phenolic resin paper substrate, the metal being milled to separate the segments. My first contact with true printed circuits, as we know them today, was serendipitous, as it would shape my future career. My boss asked me to visit another factory in the group where they made portable radios. They were experimenting with a machine for soldering. I went along and met the late Rolf Strauss, who became a good friend, and his prototype Fry's Flowsolder machine. Those old enough to understand will do so when I say it seemed as if it were designed by Heath Robinson, with little resemblance to the commercial machines of a few years' later. Of course, it presupposed the use of printed circuits, which were made by Technograph, using an SRBP laminate with rolled copper foil, made by Formica Ltd A couple of sets had been previously hand-soldered to act as a control, while another handful had the components “stuffed”. On the great day, the gas-ring under the solder pot was lit, the fluxer (a couple of paint spray guns with the triggers taped down) was supplied with compressed air, a screen having been installed since the previous trial had resulted in an explosion as flux spray hit the flame, and the solder pump was started. The pallet with the first assembly sizzled and spluttered like hell when it reached the wave, as the notion of preheating was a later addition. We waited half a minute for the solder to solidify and then turned up the pallet to admire the perfectly soldered board – except that all the components fell to the floor because there was no copper left on the board! Four weeks or so later, I was invited to the next trial, which was the breakthrough. The difference? The copper foil was electrodeposited on flat sheets by the Royal Mint at Tower Hill in London; this was a part of their metal-purification process for “coppers” which were actually made from bronze.

A year after this, I made my first transistorised apparatus, actually a 1kHz 0dBm sinusoidal signal generator the size of a cigarette packet, including batteries, used for the field calibration of telecoms networks. This was with two germanium PNP point-contact transistors that failed with a good glare at them. The leads were short, only about 3cm long and the maximum temperature of the junction was about 40°C, so we had to use thermal shunts when soldering them. These were crocodile clips with thick copper strips soldered in the jaws. It was no fun, trying to solder three leads, as quickly as possible, with three of these monstrous devices blocking your view!

So these three techniques have made the biggest impact to electronics in the past 50 years: the printed circuit, the wave soldering machine and the germanium (later silicon) semiconductor device, and all three have well and truly celebrated their golden anniversary by now and all three are still in common use. As an anecdote, two out of these three techniques owe their genesis to Jews who had fled the Hitler regime to the UK, my friend Rolf Strauss for wave soldering and Paul Eisler for the printed circuit. The latter, whom I met very briefly only once, was an opinionated character who had a knack of getting everyone's back up, while the former was one of the gentlest persons I ever met; the contrast between them was astounding.

Of course, another great change in packaging technology has been the advent of surface mounting and the concomitant solder paste reflow, but this was not a revolutionary change; it was more a development forced by a need for miniaturisation. The fact that it has now become the de facto mainstream technique is only incidental, because it became more cost-effective with the economy of scale offered by modern assembly machines. Cast your mind back just 20 or so years to the slow, cumbersome, unreliable assembly machines for through-hole techniques.

No historical review would be complete without a word about the environment. Prior to about 1970, the electronics industry was generally perceived to be light and non- polluting. This changed slowly over the years, initially within the PCB fabrication sector which, let's face it, had many polluting black sheep, mainly small “bucket shops” who would not think twice about sending their used etchant into the public sewers or a water course. Hopefully, this no longer happens, but the assembly guys also polluted the air with flux vapours, cleaning solvents, as well as polluting water in some plants. The perception of the industry did change in the public's eye and the Montreal Protocol (1987) contributed to this in all sectors. This was followed by a raft of new regulations in most countries. It followed that this hit SMEs more than large corporations, as the cost of implementation was always disproportionate to the volume of a company's turnover. This resulted in thousands of small companies being forced to put their key under the doormat, out-sourcing their production (often to countries with low manpower costs), combining or being bought out. Unsurprisingly, this paragraph would not be complete without a mention of RoHS, the summit of bureaucratic ineptitude in the field of the environment, both because it causes more environmental harm than it avoids and because it is so badly written that it will take years of costly brouhaha before all the finer points are resolved. I sincerely hope that no industry ever has to be submitted to anything so stupid again, although REACH may be an even more costly EU enterprise.

A positive step forward that some of the companies in our industry have embraced is the sensible use of the internet as a means of obtaining technical information. In a few cases, this has even resulted in a better implementation of concurrent engineering, although this is a technique that should have been much more heavily used than is the case. I foresee that this will become a major improvement over the next decade, in order to reduce costs.

Another costly error perpetrated by our industry, especially over the last 20 or 30 years, is the abandonment of commonsense. We have replaced it by a slavish desire to stir paper. The desire to be seen to be clever in this field is demonstrated by the adoption of working to standards, whether they are relevant to the job in hand or not. For example, I know SMEs who all but bankrupted themselves to obtain an ISO 9002 certificate, tripling their consumption of paper to do so, increasing their payroll of non- productive staff, yet their products were no more reliable or well-made than they were beforehand. How many times have I seen messages posted to internet forums asking for chapter and verse on standards where none exist or are needed? Sometimes others ask for interpretation of standards that are not even relevant. This replacement of commonsense by being enslaved to standards, where they are not strictly necessary, is an expensive luxury. Remember that you can follow every standard under the sun and still have a product that is rubbish, while your competitor uses a little nous and produces a very useful gizmo that undercuts your own. This goes hand- in-glove with the paperless factory. We have been talking about this since the 1970s, but we are no nearer to seeing them generalised than we are to living on Mars.

So, quo vadis? I do not know. The present leaning to have everything manufactured in the Southern and Eastern Asian countries seems ineluctable. What I think I do know is that this will not last for ever. The trend started in places like Singapore and Hong Kong. In turn, these have become uncompetitive as their standard of living rose and the baton was passed to Thailand, Malaysia and Taiwan. Now it is going to China and India. Eventually, it will end up in Africa. When? I am not sure, but my crystal ball tells me about 2020 will see the start of the drift towards the more politically stable sub-Saharan countries. In the meanwhile, on condition that it, too, becomes more politically firm, I foresee Iran may turn out to be a high- tech industrial nation. On the technical front, I can see that miniaturisation will still be a keyword, as a means of cost cutting and for environmental reasons. Is there a limit to what can be done? Obviously there must be, but as integrated circuits increase in the number of transistors, so must the interconnection problems become more acute. No one thought, 20 years ago, that a CPU would have a pin-out of 775 or more, requiring as much motherboard real estate as one with 48 pins at that date, neither did we dream that it could ever work at 4GHz, rather than 25MHz. I foresee that the printed circuit board, ever more sophisticated, is likely to remain the mainstream interconnection for at least the next decade, possibly longer. But we may not recognise it, or its assembly, compared with today's, any more than anyone from 1986 would have recognised what is current for us. Progress is inevitable.

Brian EllisCyprus