Appendix D: How to Navigate T2

Imagine yourself standing in front of the central cubicle of the 405-line vision transmitter. In the centre of this was an Envelope Monitor (a CRT screen which displayed an envelope of the radiated waveform - basically, pure RF to the Y plates, and a saw tooth waveform at either line or field rate to the X plates). Flanking it were two large rotary handles, looking rather as if they had been provided to wind the thing up. They were in fact interlock handles, ensuring that you could not get inside the various units whilst they were powered. Turning one of these handles removed the power supplies and then released locks on each of the cubicle doors. Two handles: one for the modulator side, the other for the RF cubicles.

Stretching away to either side of this central panel were rows of doors, one to each cubicle: modulator to the left, RF to the right. There were similar doors, also interlocked, to the rear of each of the cubicles, reached from a rear corridor.

The RF Stages

Let us start with the RF side - this was the more straightforward. The rightmost cubicle of T2 in fact housed switchgear and fuse panels, but immediately to the left of this was the first low-level RF unit: working leftwards from this were increasing RF power stages, until the penultimate RF cubicle was reached. This was known as the Pen RF stage for obvious reasons, and unlike the earlier stages which never gave great trouble, was the cause of much grief over the years. Not that this was all the fault of this particular stage- most of the real reasons were further on. I can't recall now the original valve type used when the station was first commissioned, but in the early days it was always a little short of wind, and touchy. In the 'fifties it was rebuilt to use a pair of TY7-6000 valves by Mullard. These were originally designed for use in industrial RF heating applications I believe, and were very robust - a good thing, in view of the indignities that were heaped on them. Basically all they had to do was provide about 5kW of constant power at carrier frequency to the final Modulated Amplifier (Mod Amp) stage. Unfortunately the load provided by the Mod Amp to the Pen RF stage was very peculiar, often causing the latter to indulge in fits of the sulks, which usually ended in violent instability and a flashover somewhere. When that happened, the transmitter would be off-air in short order, with a mimic panel full of overload lights and alarm bells ringing furiously.

Moving one cubicle further to the left brought us to the Mod Amp cubicle - the horror of horrors around which most of the catastrophes attending this transmitter revolved. This was basically a very straightforward grid-modulated RF amplifier. Originally it used a pair of CAT21 triodes with video modulation fed in parallel to the grid circuits. From an RF standpoint, it acted as a grounded-grid amplifier with RF fed to the cathodes in push-pull. Neutralisation was applied in the forlorn hope of making the whole thing stable. The valves were later replaced with a pair of BW165's, rather meatier devices which could in theory give the transmitter a rated output power of 50kW peak white (the old CATs could only manage about 35kW). In practice, you would have been very ill-advised to attempt running them at this power level - violent flashovers were much more frequent at these higher powers, and over the years the output power was quietly crept down to a level of about 43 to 44kW, at which level life became much more placid.

The valves, with their anode cooling water jackets, each sat in the top of a large-diameter plated brass pipe, the pair of which formed a resonant section. These, together with a capacitive tuning plate, which could be slid up or down between the lines, formed the output tuned circuit. Behind these lines was the output-coupling loop: this was a massive square-section loop built up out of silver-plated brass. In series with this was the output coupling tuning capacitor, again built up from plate, having very thick vanes with heavily rounded corners to prevent corona discharges starting. Of split-stator construction, the whole item was perhaps an eighteen-inch cube and the rotor was carried on an insulating glass rod about an inch in diameter. (See 'the Infamous Spacewarp Incident' to find out whether or not this glass rod was a good idea.)

To finish off the RF side of things: a high-level modulated transmitter, like this one was, had to generate a double-sideband signal, whereas what was wanted was a vestigial sideband (VSB) one. Accordingly, the unwanted section of the upper sideband was removed by a massive VSB filter, constructed from sections of coaxial feeder about six inches in diameter, and mounted on the wall at the rear of the transmitter. Effectively the filter diverted the unwanted sideband power into a water-cooled dummy load mounted adjacent. This unit was a fine example of the designers not knowing quite what to expect - this unit was, after all, the first VSB filter to be built in this country, and Sutton Coldfield was (as they never tired of telling us) the most powerful TV transmitter in the world when it opened. They thought that a 5kW load should have been big enough. In the event, this nice high-power load never had more than a few hundred watts dissipated in it. Other items of bad-guessing to be found around the transmitter were meters that never read (because the expected 50mA of current turned out to be 0.5mA in practice).

The Modulator

Time to look at the modulator side of the transmitter, another fertile field of concern. Start at the beginning - walk up to the leftmost cubicle, and open the door. You could do this, for these two end doors were not interlocked, (and a good thing too, considering the amount of time people needed to spend in there), for there were no dangerously high voltages enclosed.

The Pre-Amps

Inside you would find two identical units, coyly labelled 'Preamplifier'. Those of you of a suspicious turn of mind may feel that there was rather more to these than meet the eye, and you would be quite right. When EMI designed these units, they felt that they were going to be troublesome, so they provided a duplicate installation with a changeover relay panel, operated from the control desk. They were quite correct in their surmise.

These preamps had to do rather more than just amplify. They each had their own sync-pulse separator feeding a clamp-pulse generator, used for DC level restoration in both the pre-amp and the Sub-Modulator, which we will be meeting shortly. The pre-amp also had to provide linearity correction, pre-distorting the vision signal to compensate for the later distortion that was going to occur in the Mod Amp stage. (Those of you familiar with amplifiers but who have never worked with video circuitry may say at this point 'Why don't you use negative feedback?' A valid question, to which the answer is that it won't work. There is far much too much time-delay between the instant that the signal goes into the transmitter, and the time it comes out. Any attempt to use feedback under these circumstances will only result in a bad waveform becoming rapidly worse).

When Alexandra Palace was built, no linearity correction was provided (the signal fed to the AP transmitter had a 50:50 picture:sync ratio which gave the required 70:30 output ratio, thanks to transmitter non-linearity), so the Sutton pre-amps were very much a first attempt at sorting out a problem that had been bypassed in earlier years. Generally speaking it worked fairly well, but could have been rather more stable. All valves in this unit were small plug-in types with Loctal bases, a variety now mercifully almost extinct. They were mounted horizontally, sticking out forwards from the pre-amp panel and in clear view. Unfortunately this type of base had a bad record for poor pin-contact resistance, and due to the horizontal mounting the valves would have eventually fallen out of their sockets had they not been held in place by screw-down skirt clamps. These clamps however did little for the poor pin resistance, and merely made it impossible to waggle the valves in their sockets when the pictures became riddled with black-line flashing, the usual symptom of a bad contact somewhere. The usual cure was to switch over to the other pre-amp: if you were lucky, that one wouldn't be flashing quite so badly.

Finally, the pre-amp had to amplify the input signal (about 0.7V pk-pk, since it had been through a level-setting fader on the control desk) to an output voltage of about 50V pk-pk. This signal was then fed to the next stage, the entertainingly named Sub-Sub-Modulator.

The Sub-Sub-Modulator

It is in this stage that we first meet that strange valve, around which most of the modulator was designed, the ACM3, by Marconi-Osram. I have a feeling that this valve was in fact the civilian equivalent of a valve designed in World War II as a radar modulator valve. It was an interesting device, with good performance for the time, and only one serious vice. (See 'The Vice of the ACM3' below). It had a faintly futuristic look, slightly resembling the spacecraft that were to be seen navigating the screens of the B-movies of the period. Basically a glass tube about nine inches long or so, it had around the centre of its girth a set of anode cooling fins enclosed in a brass collar, about three inches in diameter. At one end, a stout brass cylinder protruded, the grid connection. At the other, a collar that was the cathode connection and also one side of the heater: the other heater connection being on a flying lead coming from its centre. The thing was indirectly heated, forced-air cooled, had an anode dissipation of about 1.5kW, and a high mutual-conductance for a valve of that type. In all, a handy device.

But back to the Sub-Sub-Mod. This used an ACM3 as voltage amplifier, followed by another ACM3 as cathode-follower. This arrangement was quite standard in this transmitter, and gave good performance. The main problem was the common heater-cathode connection, which in a cathode follower meant that the considerable capacitance to ground of the heater supply circuitry was visible to the signal path.

To stop this from degrading the video bandwidth, various cunning wheezes were adopted (bifilar supply chokes, padded out to a fixed loading using Constant Impedance Networks). These steps worked, at the cost of converting what was basically a simple circuit into a jolly complicated one. The sub-sub-mod took the signal level up to a few hundred volts, which was A.C. coupled into the next stage, the Sub-Mod.

 The Sub-Modulator

The Sub-Mod was a repeat of the sub-sub-mod, but used two ACM3s in parallel as voltage amplifier followed by two more in parallel as cathode follower. Due to the A.C. coupling from the previous stage, the D.C. component had to be reinserted at this point, and this was done by the Clamp Unit, a diode bridge that used four power rectifiers (UU4s, I seem to remember) as diodes. The Sub-Mod took the signal up to about 1300V pk-pk. and fed the Modulator.

The Modulator

The Modulator was a straight cathode follower, and used four ACM3's in parallel. This then fed the Mod Amp grids with about 1200V pk-pk of video, flat from DC to 3MHz, at an output impedance of some eleven ohms. As you can see, a considerable amount of modulator power was involved.

Power Supplies

The power supplies round the back were quite extensive. Each modulator stage had its own shunt regulator using (you've guessed it) ACM3's. The Mod Amp was fed with unregulated DC at about 6.5kV: this came from a six-phase rectifier assembly using hot-cathode mercury vapour rectifiers. The Pen RF stage was fed with stabilised 5kV obtained from the 6.5kV line via a series regulator (ACM3's again).

Well, that was the vision transmitter, more or less as built. Over the years many modifications were made. Mercury vapour rectifiers have an unfortunate habit of backfiring if they are not operated at the correct temperature (sometimes they just backfired, anyway). A backfire (conducting simultaneously in both directions) usually blows every fuse in sight. The fuses that were used were 'Quenchol', a special high-rupturing-capacity type that were basically a spring-loaded fuse-wire immersed in insulating oil. They were evidently not quite high enough capacity, since when they ruptured they tended to blow their end-caps off, showering the cubicle with hot oil. When they became available, the mercury rectifiers were replaced with mixed-gas units which were much more placid. Later still, silicon rectifier stacks were used: strangely these did not prove to be much more maintenance-free than the mixed gas units that they replaced.

The Mod Amp filaments were originally lit with DC from a mighty rotary converter - this was done to keep hum levels down. Many years later it was found that the hum penalty incurred by using AC heating was only very marginal, and so when it wore out, the rotary converter was replaced by much simpler transformers.

The most important modification was that which (mostly) tamed the Pen RF stage. The problem here centred on the enormously varying load that the Mod Amp placed upon it. When radiating peak white, the Pen RF delivered about 5kW, and it didn't mind doing that. The problem came during sync pulses, when the Mod Amp was cut off: at these times the Pen RF stage was completely unloaded, so the RF had nowhere to go. Instability was the usual result. The cure was the installation of a gadget referred to as the 'Pen RF Load'. This was simply a resistive load, poked in between the cathode tuning lines of the Mod Amp. The idea was that instead of the loading swinging from heavy to zero, it went instead from heavier to small. This small loading proved sufficient to stabilise the system. The load itself was simply a block of stainless steel; however, in line with the tradition that power levels inside this transmitter would never be what were expected, upon first powering after the modification this load was seen to rapidly assume a cheery orange glow. Further rapid modifications involved drilling the stainless steel block to produce galleries, and circulating tap water through them. This did the trick - not much water flow was needed, about that of a slowly running domestic tap. The warmed water was just run down the nearest drain. I suspect that, when installed, this was to have been a temporary arrangement. Like so many temporary arrangements, it lasted in fact for the lifetime of the transmitter. I hate to think how much water was used over that period, but at least it was all paid for (the station feed was metered). In retrospect, I think that leaving it so may have been the right decision - at least it was a lot more reliable long-term than installing pumps, heat exchanger and fans so as to reuse the water.

As mentioned earlier, the ACM3 was a pretty good valve but did have one problem. As it grew older, it tended to suffer from grid-emission. This phenomenon was caused by the fact that the rare-earth coating on the cathode (thorium, and all that stuff) very slowly evaporated off the cathode with the passage of time. This vapour tended to condense on anything cooler than the cathode. The control grid was a good candidate for this, so after a while the grid became provided with a little emissive material of its very own.

Since the grid also ran hot (not from grid current: it was just placed very close to the cathode) after a while it would also start to emit electrons, resulting in a small grid current being developed. In most of the stages of the transmitter, this didn't matter two hoots, but there was one stage where it was a killer - the Sub Mod. The control grid to this stage was AC coupled and in fact didn't have a grid leak at all: any leakage had to be made good by the action of the Clamp. The effect of grid current was to produce a slight tilt across each individual picture line, too small to be actually visible on picture. The problem happened when the clamp fired and pulled the grid potential back to its proper value. The result was a severe bend, or coggle, as locally called, in the back porch of the waveform. This could be visible on earlier receivers that didn't have adequate flyback blanking, as a faint wobbly band of light up the centre of the picture, colloquially known as 'rope'. A good TA on the control desk would spot the onset of this problem on his waveform monitor, and tip the wink to the shift engineer. The problem was rarely so bad that something had to be done about it then and there - usually the valve would be changed after the end of programmes.

This event tended to be a marker in the lifetime of ACM3's. The Sub-Mod always attracted new valves, young and vigorous and with no bad habits. As they aged, they fell into bad ways and started showing signs of the dreaded grid emission. They would then be retired from that spot, being replaced with another new valve. But grid emission apart, there was still plenty of life left in the old reprobate and he would be put into a socket where grid emission mattered less, releasing a still older specimen. The engineer would take this still older one and prowl round the back of the power supplies, looking for an antique ACM3 there that was near expiry, and move the tube into that spot. The power regulators were regarded as the retirement home for elderly valves: they were so lightly run that any tube that wouldn't work in them wasn't worth keeping. So, for these unfortunate valves it would be the scrap yard for most, but not all.

Over the years, various people entertained the belief that dud versions of this valve could be converted nicely into a stylish base for a table-lamp. For many years, such a lamp, locally made, was one of the standard supplementary presents made to retiring staff - "Just to remind you of Sutton Coldfield". I suspect that on arrival at home, most of these lamps went straight into a cardboard box, and thence into the attic. Mercifully, before I myself retired, Band I had closed, the supply of ACM3's had dried up, and the custom had died out, thus saving me a lot of trouble. I don't have an attic.

There was one latter-day defect that the ACM3 suffered from - suddenly, in the 'seventies, it was found that quite new specimens were failing very early on in their lifetimes - often whilst still under guarantee. Such devices were sent back to GEC with a protest note, and they investigated the matter. The result was faintly amusing. By this time, the BBC was about the only customer for these devices, and they were being hand-made in quite small quantities. The defect was traced to the grid structure: 'the old geezer who wound the grids has just retired, and nobody else quite knows how to do it properly...'. In the event they had to bring the poor fellow out of retirement just to show someone else how it was done. I hope his palm was adequately crossed with silver.