4-tube cameras and illegal signals
 

I just got a copy of the October 1966 RCA Broadcast News, with articles touting the TK-42 4-tube studio camera and TK-27 four tube telecine camera.

You may have heard of illegal signals being generated when computer image files and video are converted among different standards, but it turns out this was a problem with four tube cameras back then, because the separate luminance channel generally would have a too-high signal output on saturated colors compared to a correctly matrixed three-tube camera.

This issue has two pages of cartoon diagrams and explanation of special non-additive matrix circuits that were used with the TK-27 to drive the waveform monitor so that these out of bound signal conditions could be detected and corrected (but they don't say how - just iris/gain down?) before putting out signals that might overload the transmitter.

Four-tube cameras avoided loss of detail in the luma due to tube misregistration, but it was a really dumb idea for color reproduction. Of course, there are lots of testimonials in this issue of how much the purchasers of TK-42s like the color reproduction. Blecch.

The transmitter had a proc amp that clipped everything to 100%, besides regenerating sync and burst. Some models were Vital, Grass Valley and RCA.
The model proc amp I liked was the RCA TA19 which had video AGC.
For the cameras, they had auto iris, auto black level and auto sensitivity (auto target).
Looked pretty good when everything was working right.

Yup, proc amps. The savior of inattentive camera ops and editors the world over... Weren't there FCC monetary fines for stepping out of bounds on your signal? (May still be, but probably less common due to digits and all.)

The major FCC no-no I recall was accidental expansion of the blanking interval. Like it made any difference to CRT sets, which all over-scanned to some extent anyway.

The question of 3 vs 4 tube cameras was still an issue when the UK started its colour service in 1967. At least we had Plumbicon tubes rather than IOs. Hence no incentive to tray and match a single IO for Y with 3x vidicon for RGB. Or make a huge camera with 3x IO.

There were 3 cameras available in the UK in 1967. The Philips PC60 was 3 tube, the Marconi VIII and EMI 2001 were both 4 tube.

I think the Philips PC60 was sold in the US under the Norelco name. The Marconi VIII was widely sold internationally. The EMI 2001 saw little use outside the UK.

4 tube cameras gave better registration and grey scale. The problem of resolution on 3 tube camers was later solved by taking high frequencies largely from the green channel.

The other big argument was about constant luminance. Becuase all the RGB/YUV matrixing is performed on non-linear signals some Y travels in UV and vice versa. If Y and UV all had the same bandwidth this would hardly matter but they don't so you get various artefacts. 4 tube cameras got much closer to constant luminance than 3 tube.

In the early days in the UK, an EMI engineer called Ivan James was the main proponent of constnat luminance. I can see that I wrote almost the exact same post here in 2010: http://www.videokarma.org/archive/in.../t-248163.html

See also Poynton whose explanations are pretty good: https://poynton.ca/notes/video/Constant_luminance.html

The biggest problem with illegal signal level occured with quad tape. You could record overdeviation that was impossible to get out on playback.
I only remember one FCC visit. I don't think we were fined but the FCC made a recommendation we buy a new transmitter. Which the management did.
The one we had was a converted B&W transmitter that looked smeary on the air. It was a 25 year old tube model (TT25BL).
A TA19 attached.

The problem with non-constant luminance is not fixed simpy by using the Y output from the fourth tube, since the missing luminance is only at the high frequencies that the UV channels cut off. So, for example, a red rose has generally the correct brightness and chroma over its broad area on a color receiver, but greatly reduced detail contrast. (It also appears too dark overall on a monochrome set, which does not produce the missing low-frequency luma that is carried in the UV channels.) Using the complete fourth tube signal for luma means that the low frequency luma becomes correct in a monochrome set but wrong in acolor set because it is carried in both the UV and in the Y. For a color set, you really would like to have the Y' produced by a three tube camera, augmented by added high frequencies from the fourth tube.

By the way, the proportioning of the luma signal from 30% R, 59% G, and 11% B was somewhat erroneously called constant luminance in some early texts, but this referred to it being nearly true for small chroma noise fluctuations on neutral or nearly neutral colors. An important principal of analog color TV, but not the issue of saturated colors having reduced detail contrast, which is the usual use of the term.

Most fines in the 70's day...my day...were for blanking error no matter where they came from. Early 3/4" tapes were a killer. Quad was a roll of the dice. Blanking errors and the resulting empty spaces on either side of the scan were considered "non-program material" Front porch, back porch, it did not matter. Your transmitted picture was not wide enough. I don't remember any procs in our line to our RCA TTU-30 xmtr. GIGO.

Plumbicon tubes inherently have unity gamma. So the cameras had 2 chances to tweak the colourimetry. WIth matrices before and after the gamma correctors. It took a great deal of work by the designers to get good results. Matching an IO for Y with 3x vidicon for RGB sounds like a nightmare.

As soon as you convert from RGB to a YUV colour space there is the possibility of illegal colours. Just as true for digits as analogue. Some years ago I designed a piece of kit for a client that took SDI in, legalised it and gave an SDI output.

Doing this crudely isn't hard. Convert to RGB, clip and convert back to YUV. This can look horrible in practice so I designed some soft clip methods and desaturation algorithms to avoid the YUV>RGB>YUV conversion. It proved surprisingly tricky to get results that were both technically correct and looked OK. One part of the algorithm needed several divisions at full (13.5MHz) video rate. FPGAs have had fast multipliers for many years but division (I think it was 26 bit numerator, 12 bit denominator) took a lot of logic. I think I used a single divider running at 54MHz to do all the divisions as separate ones would have meant a stupidly large FPGA.

Re: Philips PC-60 TV cameras. Yes, sold in the USA as Norelco PC-60. ABC used them as well as other facilities.

OLD thread, I realize, but just wanted to chime in that the Bosch KCU and KCK 40 cameras, both 4 tube, did an outstanding job with color. Probably by their era, B&W was no longer an issue. I had 6 of them in a remote truck doing sports. They had a Y Gamma control, which amounted to a contrast adjustment. For a camera. Helped a lot in odd lighting situations. Detail came out of the high bandwidth Y channel as well. The designers degraded the blue channel to save bandwidth I suppose.

Later they adopted the entire system to triax, from the original multicore, by putting a box of electronic tricks underneath the camera, and adding rack chassis to the CCU. Worked pretty nicely until you'd lose one of the reference signals.... then the camera would go absolutly nuts. The head had a lot of motors in it for registration, and they'd all start going crazy, along with the signalling system that would blink, buzz, and ring. We called it "The One Man Band".

memories.....

Did the Bosch cameras have a larger tube for luma, or were all four tubes the same size?

If this info for KCK cameras is correct for the KCK40, it indicates it was a 3-tube camera, but instead of RGB, it was RBW (red, blue, white), so luminance still did come from a single tube.
https://www.tvcameramuseum.org/bosch/kck40a/kck40p1.htm

Yes to the KCK. Memory is fading I guess. Yes also to the Y channel being a larger tube. Same memory again lol

The original brochures are also available...

http://www.tvcameramuseum.org/pdfs/bosch/kcu40broc.pdf
https://www.tvcameramuseum.org/pdfs/bosch/kckbroc.pdf

So all three tubes were 1.2". However, at least on the KCU 40 the blue channel had a smaller image size, which was a bit tricky and the method to match the channels was later modified against the original design.

The four tube camera delivered by Bosch Fernsehanlagen was called "KC 4 P 40", and it was in essence the General Electric PE-350, offered as an interim solution after they abandoned the project to design a IO colour camera (they gave up after prototypes suffered from unmanageable registration) and went straight into the development of a three Plumbicon design. I don't know if any of these four tube cameras have been delivered to TV stations outside Germany at all.