Thermionic standards converter......?

I doubt any of us has the room big enough to house it or the funds to pay the electricity bill the thermionics would consume. Just look at the foot print the original transistor one took up https://www.radios-tv.co.uk/standards-converters-from-pye-to-the-aurora/ As a hypothetical musing, it should prove interesting to see what ideas come to the fore. I bet David is already planning it One thing for sure, its would resemble Colossus.

Well I suppose the 16mm film standards converter had valves in it for some task but perhaps not what you were thinking of. 

16mm film was the way programmes were exported to other countries.

I do recall a news item on one of our local TV channels about a chap in West Yorkshire, who had built his own converter in an upstairs room. That one was solid state device and occupied two alcoves (one each side of a fireplace) and built on two racks, each the size of a very large wardrobe. What I was wondering about was the viability of a thermionic converter in public service broadcasting, way back when. Clearly, there were a number of standards around the world and it just seems probable that content sharing in those days would require conversion at some point - I wondered if it was ever done electronically before solid state devices, or even if it could be done, material and running costs aside. I could envisage a hefty bit of kit akin to a "Colossus" kind of construction, or, if you've seen it, the 'cloaking system' mocked up for the original film "The Philadelphia Experiment" and the massive room full of 'toobs' that it eluded to. I reckon it would need a good water cooling system to sink all the heat that could be generated by such a converter. It would be very interesting to see a theoretical circuit for one..... Edit..... Ooer, a crossover in postings, but obviously thinking on the same lines.....

You're thinking in terms of standards conversion where there are different field rates: where the field rate is the same it was done electronically very early - see the Pye converter a.k.a. the BBC CO6/501.

Where the field rates are the same you do not need frame stores - a few lines for interpolation is all.

I would surmise among the biggest hurdles to overcome, in a thermionic system, would be problematic delays and timing errors introduced by longer signal runs through such widespread devices, similar delays caused by lengthy wiring runs, slow and/or random response times in the less uniformly manufactured components (capacitors in the main?) of the day.

A lot of careful and thoughtful design would be necessary to design out such errors.

What a machine it would be though!......

Katie_Bush said
I would surmise among the biggest hurdles to overcome, in a thermionic system, would be problematic delays and timing errors introduced by longer signal runs through such widespread devices, similar delays caused by lengthy wiring runs, slow and/or random response times in the less uniformly manufactured components (capacitors in the main?) of the day.

A lot of careful and thoughtful design would be necessary to design out such errors.

What a machine it would be though!......  

Timing is easily dealt with; it was in analogue TV studios, down to a few nanoseconds. As ever, it would be the sheer amount of power the thing used; nobody has, as yet, miniaturised the watt.

A thermionic standards converter is not impossible; it is just impractical. Given a CO6/501 took up two 19" racks and was fully transistorised, you'd be looking at a lot more in thermionic technology. At a guess, you'd be looking at a CO6/501 for architecture.

Generating a 405 sync pulse train using valves is easy enough but locking into the 625 standard is not so simple, easy with TTL.   In order that the two standards are locked together in frequency and phase  the lowest common frequency is 2,531,250c/s. This signal will divided down by 162 to produce the 625 reference of 15,625c/s and 250 for the 405 line frequency of 10,125c/s, in practise the division will be 125 in order to trigger the 2H frame sync pulses, further division by two to  produce the 405 line sync pulses of 10125c/s.   To make those divider chains is not going to be easy using valves.  However, the 405 eight broad frame pulses of 40 microseconds each can be initiated from the 625 frame sync but then the line frequency stability will have to be maintained by some other means. So we do lock the two line standards together then comes the question how do we store the active video part of the lines? Valves are out of the question and delay lines are difficult to implement so it has to be optical conversion, that is, a 625 picture monitor and a 405 line camera.  Before WW2 the Scophony company made mechanical TV receivers for 405 lines. The key element in these receivers was the "Jeffree" cell.  This revolutionary device was invented in 1934 by John Henry Jeffree and was called the "Supersonic Light Cell".   An article about the Jefree cell appeared in the May 1935 "Short Wave and Television magazine.  Perhaps this device could have been used in a line store converter? Till Eulenspiegel.

Till Eulenspiegel said
Valves are out of the question and delay lines are difficult to implement so it has to be optical conversion, that is, a 625 picture monitor and a 405 line camera. 

"...delay lines are difficult to implement..."

But, does that mean just difficult? Extremely difficult? Or simply impossible?... At the moment we're exploring the possibility. Not that anyone is likely to build such a converter, so it's just a look into the theoretical, but I can see now why it would not have been done electronically.

Hi Marion,   “…delay lines are difficult to implement…”    I should have written impossible to implement.   The line stores have to accept the 52microsecond active part of the 625 video waveform and stretch it out to 80microseconds.

Delay lines were employed in the 1968 Mexico Olympics 525/60 to 625/50 colour TV systems converter.  The active video part of the waveform is almost the same in both standards whereas it is 80microseconds for 405 and 52 microseconds for 625.

Till Eulenspiegel.                                                                                                                                                                                                                                                                                                         

Till Eulenspiegel said Hi Marion,   “…delay lines are difficult to implement…”    I should have written impossible to implement.   The line stores have to accept the 52microsecond active part of the 625 video waveform and stretch it out to 80microseconds. Delay lines were employed in the 1968 Mexico Olympics 525/60 to 625/50 colour TV systems converter.  The active video part of the waveform is almost the same in both standards whereas it is 80microseconds for 405 and 52 microseconds for 625. Till Eulenspiegel.                                                                                                                                                                                                                                                                                                           

There is, if you will excuse me, some confused thinking going on here. It is perfectly possible to use analogue delay lines with valves.  The thermionics are not required for the delay, but to correct for the losses in the delay network.  If you look in "Birkenshaw", there are diagrams of LC delays and equalising amplifiers based on valves.  This is a sensible technology if you only need delays of a few microseconds, after which point the LC - equalise approach gets out of hand and alternatives are preferable.  Camera vertical-aperture correctors require two delays of one line each, which were done using quartz glass blocks and a carrier system - nothing here that can't be done with valves as the active devices.  The delays in the frame-rate-changing converters were also quartz blocks, but because of the delay required (of the order of 17ms) the block was in the form of a polygon about (from memory) 15" across - but, again, no reason in principle why the necessary modulator, demodulator and amplifier could not be based on valves. The challenge of standards conversion is to retime the video, arriving according to the input standard, to suit the requirements of the output standard.  To do this, as a generality, requires a method of delaying successive picture elements by different amounts.  This is equivalent to, as you correctly say, the need to be able to put the signal into the system at one rate and get it out at a different one - so what is needed is a storage network rather than a simple time delay.  In the analogue domain, this is much easier to do if the maximum time for which storage is required is fairly short.  The CO6/501 625-405 line converter stores one active line on several hundred capacitors and requires a bay to do it with semiconductor electronics.  If you could get the leakage in the various switches low enough for this to be a sensible approach, this implies that to store one whole 625 line picture would required 575 bays of electronics at the rate of one per active line, and there would seem to be something of a real estate problem arisen.  For this reason, the analogue standards converters that were required to change the field rate did the conversion in two stages: 525/60 -> 525/50 -> 625/50.  The first conversion involved the big glass delay lines, the second used, in effect, the arrangement of the CO6/501, though with modified component values on the store cards to suit the different interpolation characteristics required. There is a little wrinkle in that the CO6/501 picture element stores incorporated a device for which there is no thermionic equivalent: the OC140 symmetrical junction transistor.  This is because the write side needs to be able to source charge from the storage C as well as sink charge into it.  However, this obstacle isn't insuperable although it would require the store to be redesigned to use valves. So far, we have only considered the luminance signal.  It must be borne in mind that colour television effectively involves three signals, Y, (R-Y) and (B-Y) which occasionally share the same communications channel.  If you want to standards convert colour, you must handle the signals at baseband at some point, as the modulated chroma signal will not survive interpolation without severe distortion.  In effect, take what has been said above and multiply by three. Is it possible to build a television standards converter using only valves as the active elements?  Theoretically, yes, though there are three powerful practical reasons why you might want to think more than twice before doing so: (i) size: thermionic equipment is, in general, much larger than the semiconductor equivalent.  In the days of the white-fronted transistor video distribution amplifier, these came eight across the width of a bay at 3U high.  The miniature valve equivalent came two across the width of the bay: the octal equivalent occupied the whole of the bay width.  A CO6/504 (one of the analogue field-store converters) occupied, from memory, 7 or 8 bays.  You would likely need in excess of 30 for the valve functional equivalent, and suddenly your real-estate problem has returned, albeit perhaps on a slightly smaller scale; (ii) power: every EF80 equivalent you design in your circuit requires, as a minimum, 1.89W of heater power (6.3V x 0.3A), not to mention anode and screen-grid supplies.  You will find the power requirements add up quite quickly, especially across 30+ bays; (iii) ventilation: 1V across 75 ohms (the output signal) is about 14mW: the rest of the power you have supplied to the kit at (ii) above appears as heat, which must be removed somehow [1].  The working rule of thumb in my day was that 1kW of waste heat requires a further 3kW of air-con power to remove it.  You will need a BIG bag of shillings for the meter.   [1] although it occurs to me that the "W1A" solution would be to equip the apparatus room with wooden benches and sell tickets to it as a sauna

So from the foregoing, the tube room aboard the Philadelphia wouldn't be large enough? Sinking all that heat - a tad on the risky side, but liquid cooling might work? Would need a pretty meaty pump though, to shift enough coolant. I suppose then, the old generator house (once contained 3 x 0.5MW generators) on this WW2 airbase might just about be big enough to house the converter. I suppose a single generator of half a megawatt should provide enough power to run the thing?

I'd just add to Occiput's excellent explanation of the /504 that there were six quartz polygons in the converter - each one had a delay of 3.33ms. Working out the multiples of 3.33ms should give a clue why that figure was chosen. One of the polygons used to be on display in CAR at Television Centre; it seems hard, but not impossible, to reconcile the size of a CO6/504 with the two 1U Vistek units I own that are multi-standard in and out and of a performance the /504 could only hope to achieve. If it were valves in our hypothetical standards converter, I'd opt for something like an ECC70 for size: Dave Grant of Dinosaur fame has been modifying a Murphy TPG11 with these to improve performance in critical areas. I'd also add that, given the propensity for anything with three electrodes to get sold to the Golden Ears brigade, that anyone with a /501 or /504 should buy these before they get turned into bloody fuzz pedals

A diversion from the particular topic under discussion, but if you happened to have access to LOTS of money and real estate (like the '50s US military....), large and powerful processing applications were possible with valves- in this case used to drive missiles into 3D coincidence with enemy bombers from radar data input. Supposedly, these computers were a major user of the "red base" 6SN7s that the audiophiles get worked up about;

https://en.wikipedia.org/wiki/AN/FSQ-7_Combat_Direction_Central

Whilst a very different beast to a standards converter, the architecture of fast access to memory with dynamic and developing input and attention to system timings still apply and would seem to indicate that "will and way" could make a thermionic standards converter possible- if admirably ludicrous!

Cathovisor said

One of the polygons used to be on display in CAR at Television Centre; it seems hard, but not impossible, to reconcile the size of a CO6/504 with the two 1U Vistek units I own that are multi-standard in and out and of a performance the /504 could only hope to achieve.

Indeed so: the process of miniaturisation has continued apace, the route being discrete transistors -> transistor arrays -> generic integrated circuits -> ASICs.

I'd also hazard a guess that the Visteks synchronise as well as converting: something the analogue standards converters were unable to do and which, indeed, gave rise to most of the operational (as opposed to engineering) problems with them.

When the first ACE [1] was brought into service in 1979, it was regarded as a genuine "miracle of modern technology" - it would synchronise as well as convert, the conversion had a 4-field 4-line aperture, which gave a significant improvement in converted quality, the conversion was done in one stage instead of two ditto, and all this filled just one bay - this latter being, perhaps, the most notable achievement.

[1] CO6/510 Advanced Conversion Equipment

How about this? A Murphy TPG11 pattern generator. This unit can be the source of 405 sync pulses.  The syncs can be locked up to the mains, how that can be effective I'm not sure because there is always going to be a difference in frequency and phase between the 50c/s frame pulse generated in the unit and the 50c/s from the mains supply. There is a copy of the TPG11 service manual to hand, time to do some swotting up. Anyway, it's a start for creating a valve standards converter. Although I'll have to use solid state line stores.  Nevertheless, the early stages of the 625 line part of a converter can be constructed by using valves. But before any work starts on the restoration a power unit will have to be constructed for the TPG11.  Lots and lots of 6SN7GT double-triodes employed in this unit.  6SN7s are greedy things, requiring 0.6amp for the heaters. Second attachment shows the divider chain adjustments. Everything is done properly in the TPG11. Till Eulenspiegel.

Till Eulenspiegel said
How about this? A Murphy TPG11 pattern generator. This unit can be the source of 405 sync pulses.  The syncs can be locked up to the mains, how that can be effective I'm not sure because there is always going to be a difference in frequency and phase between the 50c/s frame pulse generated in the unit and the 50c/s from the mains supply. 

Given that synchronous working was abolished in the 1960s, if you really must use this as a source of 405-line sync pulses for this project you would be better advised to work out how to lock the start of field 1 in 405 on the TPG-11 to the start of field 1 in the incoming 625-line video.

The TPG-11 uses the 50Hz incoming mains to vary the frequency of the 20,250Hz master oscillator.

I certainly wouldn't use the 50Hz mains as the reference and instead it will be the frame sync derived from the 625 sync train.   The TPG11 was an expensive TV pattern generator and not to be unkind the and hardly sort of instrument one find in the workshop of the 1950s telly shop.  

Till Eulenspiegel.

I thought the Murphy TPG11 would have been used in the factory were TV’s were made. I didn’t see one in quite a large service department with about 15 bench engineers so certainly not in smaller shops.

The first pattern generator I had was the RBM dual standard which we obtained for the start of colour in 1967.

My friend's widowed mother ran a TV shop. The only test equipment was a rather posh-looking valve tester and a hand held multimeter. The shop was demolished just before colour came in.