Source:  Andrew Fremont, Marconi studio department

Digital Beginnings

As we moved to the mid 70’s interest was mounting in various aspects of digital TV. There were a number of learned presentations at places like the Institute of Electrical Engineers; they were beginning to realise that there was more to life than big generators etc. Additionally the BBC put on some demonstrations. By the mid seventies we at Marconi had taken on production of the IBA designed digital standards convertor which, for the first time, would make possible the interchange of programmes between the UK and USA at a good standard. The convertor, known as DICE (Digital Intercontinental Conversion Equipment), was launched in the spring of 1978 at the bi-annual Montreux broadcasting symposium. By that time a colleague and I had convinced ourselves that a fully digital telecine was achievable; all we needed was the money. I should note that while the system was to be digital internally the output had to be analogue as fully digital TV studios were still a few years away. We cornered the MD at Montreux and it was agreed that a proposal would be prepared. The money was released that autumn and work started.

The B3410 Telecine

The new system would depart from the previous multi-projector format and provide just one; however it would be able to run 35, 16 or 8mm film (see picture – 35mm film loaded). It was to provide multi-speed projection and be easily switched between the European and USA video standards.  At the project start it was assumed that the system would mostly be run directly to air, i.e. the output would not be recorded. As will be described shortly, this was all to change as home video recorders came to the fore; this caused us considerable problems.

The studio development labs were lightly loaded at the end of 1978 and I was lucky enough to have a relatively free hand in picking a design team. A strong team was essential as it was going to be a major technical step forward that no one else in the world had yet tackled (in fact I believe it was not until 1990 that anyone else did). Experience from the DICE project would be of great use but, for the most part, the rest of the new design work was going to be the first move into real digital design for most. This was true for both the video and control side of the system.

The only suitable image sensor at the time was a Fairchild device designed for facsimile use (see photo). This device is still available today although it may have undergone a few design tweaks. The sensor had a single row of 1024 photocells (pixels) each 13 microns square. Charge was gathered for about 50 micro-seconds and then transferred out into two transport sections handling either odd or even numbered cells. The tiny charge packets were stepped down the chain to reach the output. This type of transport system is often called “bucket brigade” as the charge is successively ‘tipped’ from one ‘bucket’ to the next. For peak white each charge packet would typically be up to about 10,000 electrons. Unfortunately a few get left in each bucket and this has to be compensated; see the Bias Light patent information here.  Three sensors were required to give the red, green and blue signals and these were arranged around the colour splitter block (see photo). The sensors were arranged to image across the width of the film and due to their nature only ‘saw’ a very narrow strip of the film frame at any one time. The film was moved through the projection gate at very constant speed (sometimes called continuous motion) to provide the vertical scanning. Thus the frame was scanned in one pass using progressive scanning. This was what a modern flat screen TV specification might refer to as 576p (for the UK); it had to be converted to 576i for transmission (normal 625).

While the sensor was a ‘digital’ device the output was still analogue (much to the amazement of senior management!) It was important that the signal from the sensor was converted to digital as soon as possible and this had to be done with great care; we used 809 samples (but most of the pixels) per line. Once in the digital format the signal becomes very robust and excellent accuracy can be obtained – nothing drifts, there is nothing to tweak. At the time we started on the development there was no single device that could provide the required performance for the conversion and a certain amount of design cunning was needed to achieve our aims. This was true of much of the design and large numbers of fairly basic integrated circuits (ICs) had to be used to ‘hand craft’ the functions we wanted. I seem to remember that we used about 4000 ICs in all. Today you could do the same task in a fraction of the size. We could note that the microprocessors we had then ran at a thousandth of the speed of what I am currently writing with. I think most will be familiar with the small memory cards used in digital cameras. These can store hundreds of high resolution images on a square inch. We needed about 300 square inches to store just one ½ mega pixel colour image – times have changed. I should note that the image store was essential to providing the progressive to interlace scan conversion and the ability to run at a large range of speeds, including providing upright pictures when running backwards.

I mentioned that the system was multi-gauge. All one had to do was to change the projection lens / gate assembly (see photo – 16mm film loaded) and reverse the spool retainers. Everything else was automatic. Initially there was no interest in the 8mm option. The projector could run at something like 2000 different speeds and came with a comprehensive set of controls (see picture). Film run time could be adjusted within limits to stretch or compress an item; this was the main function of the panel to the right of the picture. It was possible to use an external pitch corrector to make the sound correct, i.e. just faster or slower but not squeaky or low pitched.

Reproducingnegative colour film had given me some problems on the B3404 system. This was one thing that the new system put right. I shall not go into detail but the negative performance of the B3410 was very good. Gone were all the nasty edge effects and poor colour. Some users were to find this feature was their preferred route to the best reproduction.

The video processing system of the new design was much like that of the previous one but being totally digital it had an accuracy that was arguably perfect; that’s what we told the customers anyway. Traditionally one spent hours fiddling at exhibitions, but no longer – just switch on and load up the film. More importantly at the end of the exhibition day you turned off and headed for the nearest bar, leaving the analogue chaps still tweaking!  Naturally the video boys wanted a nice new control panel (see picture).

The planned launch was to be at the 1980 Montreux exhibition, making it an 18 month timescale. This was exceedingly tight and towards the end we generally had to work a 12 to 14 hour day, except for Sundays when we finished at 4pm. I my case there was no overtime pay! Indian or Chinese take-away became our evening main diet; possibly didn’t do us a lot of good.  The machine was due to pack for the exhibition on the Friday but we still had problems and asked Dave the driver to leave the lorry for us to load. We finally ran a full film through for the first time on the Sunday afternoon, switched off and packed. A bit too close for comfort! Being spring it was, of course, French seaman strike time and the equipment was delayed a couple of days at Dover. As a result it reached Montreux somewhat later than expect, making things a bit of a rush.

The telecine had its own little raised platform that is now my shed floor. A wrap round curtain had been provided so that we could work without the competition looking on. We manoeuvred it onto its platform and pulled the curtain. Someone said “Let’s see how the boards travelled” and pulled out a printed circuit card. To our horror there were very few of the about 100 ICs that should have been on the board there. Being a prototype all the ICs had been fitted in sockets and hundreds of them had vibrated loose and fallen out. I think we were too astounded to say much. Thankfully there was a punched metal grille on the bottom of the machine and that was where the components were sitting. Luckily none appeared to have suffered damage and we just had the long job of identifying what went where and putting them back in. Hoping that we had recovered from this near total disaster, the power was connected up and the on button was pressed. The machine sprang into life and a film was laced up and successfully run (only the second film to date); the feeling of relief was unbelievable.

We had decided that the curtains would stay closed until the show opened. While the fact that we were to exhibit the world’s first fully digital system was generally known by anyone who had read the exhibition handouts they did not know what it looked like. We therefore wanted to make the first appearance a little special. With about half an hour to opening time to go my opposite number from Bosch Fernseh appeared (fern-seh translates as far-see, the German for TV). Could he have a little preview he asked. In general fellow engineers help each other, for example you could ask a rival for a spare component if you were stuck and most likely get it if they had one. We didn’t want to give away too much and set a time five minutes before the official opening. With Dieter’s little demo out of the way the curtains were opened and the delegates appeared. The response was unbelievable, something I had never seen before and was never to see again. We had a queue of around twenty or more for virtually the whole of the exhibition. We had to set a maximum time of about five minutes for each group to see the machine. Members of the team went along the queue and gave those waiting an introduction so that they were already briefed by the time their five minutes came. Those who have been involved in this sort of thing will know what ‘Exhibition Foot’ feels like; were we glad to throw off our shoes once the hall had cleared.

The engineering thinking was that we would show the machine to make a kind of statement and then spend a year fine tuning the design. For example we did not have a solution for showing Cinemascope film at that time. Additionally we knew that a better analogue to digital conversion method was needed. Unfortunately no one seemed to have told the sales team and the first off of a so far un-commissioned production run was sold for a September delivery, i.e. about five month’s time. To make it easy (not!) it was sold into Hollywood. So another hard slog followed.

Did you ever wonder why you sometimes have black bands at the top and bottom of your TV screen or why the cowboys sometimes looked very skinny? A few words about widescreen might help. At the time are talking about there were a number of wide screen formats in use. ‘Normal’ TV (i.e. what we had up to a few years back) had an aspect ratio (width to height) of 4:3 or 1.33:1. Cinemascope has a ratio of 2.85:1 but it is compressed horizontally on the film by a factor of 2; other widescreen formats (1.66:1 and 1.85:1) did not have this. Obviously the telecine needs to provide a means of decompression to avoid the skinny cowboys on c’scope but the problem was that the screen (at that time) was only 1.33:1. The two ends of the choice for were to have 25% blank screen at the top and bottom (called letterbox) or to have full height but lose half the action to the left or right of the screen; you could pan to ensure that important detail was not lost but for the big shoot out you could only see the bullets! (See picture of control panel). Sometimes you would see the still compressed version during the titles. Even with today’s 16:9 (1.75:1) widescreen TVs it is impossible to avoid either some horizontal loss (approx 38%) or blank screen at the top and bottom; or some halfway compromise.

Anyway, the machine was completed more or less to schedule and shipped.

Now at this time the telecine world was set to change. Home video cassette recorders (VCRs) had started to hit the shops in larger numbers and the Facilities Houses (who transfer film to tape) started to record anything they could lay their hands on for sale to the home market. A lot of the material was old and not in good condition meaning that a fair amount of work was needed to correct it to a reasonable standard for recording. There was then a demand to have much more flexibility in projector control as they wanted to stop at each scene and set the correction. If you stop a continuous motion telecine the live video will just be a single line of the film and you cannot see the results of any correction. It was therefore necessary to make the system run back and forth over the same frame; this was not easy with the system at that time. We ended up with attempts to do on-site development with various bodges. This lasted some time and I made a couple of visits. On the second occasion I had a nice little apartment just across the freeway from the main Hollywood studio area. Anyway, I eventually decided that we were not getting anywhere with this remote development and pulled everyone back. It was not until I finally convinced management that we needed a microprocessor controlled design that we achieved what was needed. The final system gave most of what was wanted and we provided a Shot Change Detector that would automatically detect a new action segment and park the machine on the first frame. The correction parameters were stored on floppy disc and later used to record the film in a single pass (see screen shot – looks a bit like Ceefax? No point in reinventing the wheel!).

IC technology was now accelerating at a wonderful speed and by about 1982 we were able to reduce the image storage area by half (150 sq.in) after a further three years we could get it down to 6 sq. in.

Further information

For more technical details click here.

The B3410 line-array Telecine by R. Matchell 1982 click here

Cinemascope facility for the B3410 Telecine by G.R. Boustred and C.S. Cranston 1984 click here

Digital video processing for telecine by A. de M. Fremont 1983 click here

Progress in digital telecine by A. de M. Fremont 1986 click here

Comment

Peter Smolka ex Marconi TDU/OSG reckons there are, as at 2014, still 20 B3410 Telecines working around the world from the 84 manufactured.  He should know he has serviced most of them!