How I Set Up My CT-100
 

Here's something started in 2009 that I never published on my CT-100 website, which, as you may know, is available now on the ETF site.


If you are going to set up a vintage color television set for accurate 1953 NTSC color, you must be able to recognize or measure Illuminant C.

Probably the toughest challenge you will face is to know when Illuminant C is beaming from your 15GP22 (or perhaps even the early 21AXP22; vintage documentation describes an identical spectral response for both tubes). The test for illuminant C can be a measurement by a colorimeter or a visual comparison.


Measurement
It is time-consuming, but accurate and rewarding, to use a colorimeter (such as the ColorVision Spyder2PRO used in the 2007 ETF demo) to set the white point of a Vintage NTSC Color Television set to Illuminant C, which is considered to be Average Sunlight: 6774°K (x=0.310, y=0.316).

By comparison: Illuminant D6500 is considered to be Daylight: 6504°K (x=0.313, y=0.329).


Visual Comparison

I now use visual comparison to set illuminant C: I recall vividly the Illuminant C hue I witnessed in 2007 because it was so much more red than the bluish P4 white I had been accustomed to. The contrast between P4 and illuminant C is so striking that the difference makes a lasting impression.


Adjustments

Here are the step-by-step procedures I use to set up a CT-100 or a 21-CT-55 (CTC2 or CTC2B chassis).

The controls involved are:
Color
Brightness
Contrast
R Screen
B Screen
G Screen
B Gain
G Gain
B Background
G Background


Presets:

Color down (CCW)
Contrast down (CCW)
Brightness near maximum (CW)
R Screen down (CCW)
B Screen down (CCW)
G Screen down (CCW)
B Gain (n/a)
G Gain (n/a)
B Background down (CCW)
G Background down (CCW)


Procedure:

The procedure is performed with a luminance-only image and assumes the purity has been set up.


1, Advance R Screen control fully clockwise, then back off slightly. The red screen is typically dim, but do not retard the red screen control to ‘save’ the red gun. Engineering information suggests the screen controls were designed to be set to the high side of the visible range of the control, not the low side.


2. Use the G Screen and B Screen controls to achieve Illuminant C without adjusting the R Screen control.


3. Set Brightness and Contrast for a normal picture.


4. Use the B Gain and G Gain controls to adjust the luminance-only image for Illuminant C. NOTE: When a properly set up vintage color set is displaying black and white programming, the picture will have a noticeably warm (reddish) appearance.


5. Use the Brightness control to adjust the luminance-only image for a dimmer-than-normal picture.


6. Use the B Background and G Background controls to adjust the dimmer-than-normal luminance-only image for Illuminant C. Check your setup by rotating the Brightness control through its useful range for the distinctive Illuminant C hue (not necessarily extreme brightness positions, however, as the focus may lose regulation).


7. Add some color and enjoy.


Further Thoughts:

1. 1953NTSC green phosphor is described as P1. [Perhaps modern CRT's still use the same spectrum phosphor?]

2. 1953NTSC blue phosphor is toward cyan, but was changed relatively soon to the more violet blue. [i.e., there is a little bit of green in 1953NTSC blue, and it is easily demonstrated by viewing color bars through a green filter: the blue and magenta bars are visible on a 15GP22 but barely so on a modern CRT.]

3. 1953NTSC red phosphor is, in my perception, far a more vivid red than the orange-tinted red of modern CRT television. [This is evident in side-by-side demonstrations of a 15GP22 and a modern CRT: facial tones are clearly more 'rugged' and orange-ish on contemporary CRT's. The 15GP22 skin-tone reproduction is more accurate to my eye. This can also be a result of matrix changes in later years, I hear. Comments welcome... 29 Jan update from old tv nut: " The cadmium-sulfide red used in the all-sulfide tube of the early 60s was more orangey, and also turned even more orange at high beam currents. The differences you see in flesh tones on NTSC today are not due to the red phosphor, but due to the much larger difference in green phosphors and the approximate matrix corrections made for that."]


And Final Thoughts:

A wide-spectrum nicely saturated color source (e.g., 30's and 40's 3-strip Technicolor films) reproduced on a 15GP22 driven by a calibrated CTC2 chassis delivers astonishing color images.

For best images, the 15G must be viewed in a dark room -- not in subdued or dim light -- or that light-grey 15G phosphor screen will bounce light back into your eyes and desaturate (add white) to those amazing colors. I do use a 9 watt D6500 fluorescent lamp behind the CT-100. Emphasis on behind.

The 15G in my set has very little time on it (other than what I have added) and has been environmentally fortunate all its life. I feel very lucky!

Pete Deksnis
11-March-2009
08-August-2013
25-January-2014

Pete,

Thanks for the excellent exposition on setup of the CTC-2. As with all CRT setup procedures, the difficult thing is to get tracking for a constant white point at all brightness ranges. This was even more difficult in the early sets because the red phosphor required so much more current than the green or especially the blue.

As you know, most people do not get anywhere near illuminant C when setting up the CT-100, because without a proper comparison or measurement, they just don't realize how hard the red must be driven and how far the blue and green must be toned down. It is definitely worth the trip, because avoiding an excessively blue white means that the grays are closer to flesh tone and therefore color level does not have to be boosted, and variations in the source material are less severe.

Anyway, I'm sure you realized I would chime in on this topic, so here goes:

Illuminant C (and D65) are both meant to be approximations to a bright overcast condition, not direct sunlight, which is closer to 5500 kelvin. Illuminant C is actually a physical approximation made with an incandescent bulb shining through a particular bluish chemical solution of a certain density, and it misses daylight slightly by being slightly magenta. D65 is based on a series of daylight color temperatures produced by natural daylight spectra. As such, it is never found exactly in nature on any particular day, and its spectrum is only approximated by physical sources; but it gives a more accurate CALCULATION of the appearance of colored objects in natural light. If there is no illuminant C source nearby for comparison, a TV set to D65 will look essentially identical to one set for illuminant C, as the viewer's eye will adapt completely to this slightly different color.

Philosophically, D65 is sort of the ideal artist's working environment. It is recommended for viewing of photos on computer monitors. I personally have a problem with this, as photography has historically used 5000k for viewing. Color transparencies films were designed to be projected with incandescent lamps, so probably had some blue shading to reduce the orange coloration. What this meant for the actual gray color aimed for on the screen I am not sure, but I doubt it was Ill C.

TV makers quickly found that they could not reliably produce the extreme current ratios between red and the other guns to make Ill C, without suffering differential blooming of the spot sizes of R, G, and B. Also, as a tube reached end of life, the red gun would run out of current sooner and shorten the useful life of the tube. It was decided to reduce the amount of red drive (although it was still considerably more than green and blue), resulting in the "9300 Kelvin + 27 MPCD" specs that you see on later sets. This was just an obscure way of saying "not enough red." The 9300K was a point on the black body locus, and the + (plus) 27 MPCD meant 27 "minimum perceptible color differences" perpendicular to the black body locus towards green. The result of these two coordinates was simply more cyan, or, the same thing, less red. This was unfortunate for color stability, as the color amplitude (and therefore the amplitude of variations) was turned up by most users to get flesh tones that were not too much toward cyan.

Further thoughts:
1) Yes, NTSC green was P1 (Willemite), the same phosphor used in oscilloscopes. It is much less yellow than the later sulfide green, but not as bright. Except: the sulfide green becomes non-linear at high current densities, so in early projection color sets, makers went back to P1 for a while. This messed up the color in some of these early projos, as the demodulator chips had been designed to make an approximate compensation for the color of the sulfide green.
2) You are right about the NTSC blue being more cyan. RCA couldn't make a decent sulfide blue, because of contamination by copper, which turned it into a sulfide green. Once processes were developed to prevent contamination, the sulfide blue could be used. I believe the engineers intended to have the more violet blue at first because it gives more vivid purples and magentas. They used such a blue in the triniscope sets, where each phosphor was confined to its own tube, and color filters could be used also to trim the color if necessary. It also appears that the FCC spec for the blue x, y coordinates may have contained a typo of interchanged digits that supported the more-cyan color, but we may never know for sure.
3) The modern rare earth red is not much different from the NTSC red; much of the difference in later NTSC sets came from the color matrixing, as you have heard. Neither original nor recent red can reproduce traffic signal red, although both can cover brake-light red. The cadmium-sulfide red used in the all-sulfide tube of the early 60s was more orangey, and also turned even more orange at high beam currents. The differences you see in flesh tones on NTSC today are not due to the red phosphor, but due to the much larger difference in green phosphors and the approximate matrix corrections made for that. The HDTV system, which is designed for the modern phosphors, and has the colors properly matrixed back at the camera where the signals are linear, can produce reds the equivalent of the early NTSC, but cannot do the Kelly greens and deep cyans of NTSC because the green is more yellow.

Net result, as you note, is that flesh tones from a proper NTSC source viewed on a CT-100 are more accurate; but flesh tones on a calibrated HDTV displaying a proper HD source should be equally accurate. The cases that get messed up are all the later NTSC sets with non-NTSC phosphors (or HDTV sets that are left in the sales-floor "searchlight" mode). From what I've seen, current makers still tend to set the white point bluer than D65. Custom installers make their living recalibrating sets to HDTV studio colorimetry.

Thanks to both of you for these informative comments on early sets. I'm enjoying seeing how different the red phosphor is on the CT-100. Watching a cop show this evening provided plenty of examples of flashing red lights of high intensity, and it definitely has a much deeper red effect on the CT-100 compared to later sets.

I doubt that I have Illuminant C properly dialed in on my set, but I can certainly assert that I'm driving the red very hard compared to the other two guns. Now that I've learned to keep the contrast down to make sure I don't drive the red gun into grid current mode, the colors are very nice and stable over the black to white range.

I've added Pete's procedure to our website.

Pete:

You may remember our discussion on this subject when I was setting up my CT-100. I found it helpful to use a device I had had for a long time. Its a TV-Color-Komparator made by Hellige GMBH in Giesbrecht Freiburg, West Germany sometime around the early '80s. It's specifically calibrated for D 6500 degrees and while this is not dead on NTSC C at ~6700 degrees it's certainly close if you tweak the whites just slightly bluer than what the match is on the Komparator. It made the job a lot easier once I knew what to do with it.

The attached pictures show the device The side shot has the entrance pupil on the left and the viewing eyepiece on the right. The knob in the middle is for changing the brightness of the comparison illumination. The top shot shows the meter used to set the calibration for brightness and color temp. I've never gone hunting on ebay for one of these, but they might pop up once in a while.

I just checked on eBay and there's one of the Hellige Komparators available with a starting bid of $19.99. It's exactly the same as the one I have with the exception of the handle. It's cat. no. 10702501.

Thanks for the reminder Ralph and for the eBay info; I have placed a one-time bid.

Pete

Visual comparison is the ultimate. You will find that it is so precise that you can have repeatable disagreements between different observers of when a match is achieved.

Decades ago, we had at Zenith a visual comparison unit using a fluorescent source (so the current did not require setting as it does with an incandescent source). One of the older engineers and I disagreed on how much blue was needed on the CRT to match the comparator, he requiring considerably more blue. I never found out the cause, but one possibility was natural browning of his eyes' lenses with age. Also, he had dark hair and brown eyes, an indication of likely greater deep violet / near-UV attenuation in his eyes (I am blond/blue). My experience with cataracts and my own lenses' aging is that they never got to the point of requiring more blue, even though after getting one operated on, I could see the difference in V/UV response to daylight.

I would be interested to learn a bit more about "matrix changes" to NTSC over the years. I don't know a lot of the background information on this subject, but it would seem logical to me that if the CRT phosphor colors changed over the years from deep red to a more orange red, then the cameras and modulation scheme would also have been adjusted to a different "standard." Using the old deep red phosphors like a CT-100 with 15GP22 does would then result in some errors in color using today's cameras.

Would it be correct to interpret this to mean that the CT-100 had good color fidelity back in 1954 when used with a camera designed for it, but with today's cameras, there are some color errors?

As I compare what I see on the CT-100 to what I see on my Panasonic HD CRT set (from 2004), I do see what appear to be "errors" in the red, orange, and yellow regime on the CT-100. Is this in part because of changes in cameras over the years? There could well be some nonoptimal adjustments in my set, of course -- I'm not assuming my set is dialed in perfectly. The "errors" are not really problematic -- but it would be interesting to understand them better.

I would imagine that the matrix depends on the filters used to separate RGB in the camera, the mathmatical relationships used in the RGB adders in the set (and some corresponding system in the camera/modulator), and the actual phosphor colors in the CRT.

Can one of you shed more light on this?

Wayne and/or others can best respond to the 'meat' of your questions.

But I too had been confused about the differences in color reproduction between my RCA HD CRT set (from 2002) and the CT-100.

It was actually justified in part until I revisited the quadrature transformer set up: critical tests had green areas on the HD CRT seen on the 15G as blue. Not because of a 15G phosphor difference, but because of a nonoptimal adjustment.

I now suspect that the differences that may remain are in large part a result of systems design rather than setup.

Pete

(Raising his hand)

I have explained some of these in multiple threads, but since it's hard to find it all, let me summarize directly in response to your questions.

The original phosphors in the 15GP22 and 21AXP22 were very carefully matched by the color filters used in the TK-41 cameras. In fact, this is one of the reasons the TK-41s needed so much light - the spectral response was adjusted by some "trimming" filters that attenuated the light quite a bit, but made the hue at the center of the filter response correspond to the hue of the phosphors as closely as possible.

The design of the tk-41 spectral response also followed a long tradition of design for film and graphic arts processes, in that the overall filter bandwidths were small and thus emphasized differences in spectra between different objects, thereby increasing the color saturation. The result of all this was that direct applicaton of the R, G, and B signals to the CRT gun currents could be made without any significant electrical matrixing to increase saturation or change hue. (Slight adjustments could be made by the receiver color and 'tint' controls, even though those are at the wrong point in the system due to the non-linear 'gamma' of the CRT guns.)

If the responses of the camera do not correspond to the phosphors well, the proper place to make corrections is to the linear RGB signals in the camera, before the non-linear gamma correction is applied.

The first departure from correct colors was the introduction of sulfide blue and green phosphors. The sulfide blue is more violet than NTSC blue. This means that the complementary color (yellow) has to be greener to make the specified white/gray color. This change was mostly ignored in TV receiver design as far as I know. The change to sulfide green produced noticeable differences in color due to its yellower color. This would move yellow back away from green, and is probably why the change in blue could be ignored. However, the yellower green meant that the spread of color from red-orange-yellow-green was reduced and looked unnatural.

The proper way to fix this would be to insert a linear matrix in the camera; but there was no standardization of just how yellowish the sulfide greeen would be. FCC rules simply said the signal should be "suitable" for the original phosphors. Also, TK-41s were notably noisy, and adding a matrix would increase the noise. There was also a question of maintaining color balance stably with such extra stages in the video processing. So, until Plumbicon cameras came along in the mid 60s, the only live sources were TK-41s with optics tuned for the original NTSC phosphor colors.

Each TV manufacturer played with the matrixing (color demodulator gains and angles, which have the same effect) to approximately compensate for the new green phosphor. Since the new green is yellower, that means that turning on green is also like turning on a little red. The way to fix this is to increase the gain of the (R-Y) demodulator, so it turns off the red more thoroughly when (R-Y) is negative; so, as green increases, red is decreased to compensate for the extra red in the new green phosphor. This would work perfectly if it was used on linear signals; but in the receiver it is being used on the gamma corrected signals which then get applied to the very non-linear CRT guns. The result is that on saturated red objects, the red gun gets turned ON extra hard, resulting in overly bright reds. So, while medium colors were approximately correct, strong colors had significant errors.

Meanwhile, CRTs with the new phosphors were being standardized for studio monitors; at the same time, cameras with more efficient prism optics and different pickup tubes were developed. The studio monitors were built with a switch to turn the approximate matrix on or off, so they could be calibrated electrically with the matrix off, and then would give an approximation of correct color similar to home sets with the matrix switch on.

The more efficent camera designs had wider optical pass bands, which meant they did not naturally produce fully saturated colors required to drive either the NTSC phosphors or the new ones. So, linear matrices were introduced in the cameras to correct the color rendition. At this point, the whole NTSC system was an approximation, and while the correct camera matrix could be calculated, it is certain that there was final visual judgement on the new CRT monitors to see if the matrix should be tweaked a bit. It was like the train leaving the station when the factory whistle blew, while the factory time keeper set his clock when the train left.

PAL put a stop to this nonsense by standardizing to the new receiver phosphors and doing linear matrixing in the camera to match them. NTSC never standardized anything. It is likely that the strength of correction in receivers has been decreasing over time as the tweaking of cameras has occurred, but that is a highly individual design choice of each TV manufacturer.

Adjustment of color was always needed when converting between PAL and NTSC. It's not clear if this always happened outside entertainment material involving enough money and time to be careful.

HDTV has, worldwide, standardized on the modern phosphors. HDTV color rendition is now very stable and repeatable, with material being exchanged between different places based on the same color standards (ITU-R REC 709). These primary colors have also been standardized as sRGB for still cameras and computer monitors. This is the default colorimetry for jpg images and web applications, many of which don't bother to read any information about what standard was used to produce a file.

The most noticeable drawback of the modern phosphors is a lack of highly saturated true greens ("kelly green") and especially of saturated cyan colors. For example, some cigarette packaging is inside the NTSC gamut but outside the modern gamut - but cigarettes are no longer advertised on TV anyway.

Any HDTV material (Blu-Ray) and probably most SDTV (DVD) material has been adjusted for rendition on a modern monitor. This means that when played back on a 15GP22 or 21AXP22, you may see the saturated greens and cyans, but that's not what the studio colorist saw and adjusted for. If he had a wide-gamut monitor, the source was processed to limit its color to Rec 709 so that he saw what people would see on their modern HDTV at home.

There are some computer monitors with wider gamut available, but unfortunately not built to any one standard yet. Most have a green approaching NTSC green, and usually have a red deeper than NTSC red, and an even deeper blue than the non-NTSC modern blue.

There are also some semi-standard wider color spaces, usually associated with photography applications, such as Adobe RGB and prophoto RGB. Taking pictures in camera raw mode allows matrixing the color to any of these spaces, which can have advantages for printing or displaying on a wider-gamut monitor; but many non-aware applications, like common web browsers, will mess up the colors if the files are not converted to sRGB before posting.

There are some proposed standards for expanding the color range that can be carried by digital TV signals, particularly on recorded media (Blu-Ray, DVD), but not all makers agree this is proper. As far as I know, no material has been released with a wider range.

There are some flat panel TVs that have some expanded saturation capability by using a fourth pixel color (yellow), but there is no legitimate source material to watch on them at this time.

Digital cinema has standardized (at least for now) on a wider gamut than the modern phosphors, so the source material shown at theaters has been adjusted to use and accomodate the wider range. A different matrix is used to produce the rec709 file for Blu-ray and DVD distribution

RCA "rare earth" phosphor has not been mentioned. I do not remember the details of the ads.

A slight refinement - I said that the change in the blue phosphor was mostly ignored. This may have been true early on, but by the 70s, people were doing full matrix calculations for receivers to minimimize the overall errors, and the phases and gains of (R-Y), (G-Y) and (B-Y) were all adjusted, unfortunately differently in each maker's sets.

I also didn't mention the setting of white point to a very cyan "9300K + 27 MPCD," isntead of the standard "Illuminant C," which made the choices for the receiver matrixing (gains and angles) very subjective, and in fact, led set owners to make wide variations in preferred settings, depending on source variations and how their visual color adaptation was reacting at any particular time. This white point along with the response of the TK-41s to polarized light reflections from hair resulted in a lot of complaints about green hair in the early days.

Rare earth reds are quite close to NTSC red, the major difference being their higher efficiency, which allows more nearly equal beam currents in the CRT. They did not significantly impact the matrixing issue.

So a CT100 with a 15GP22 needs video from a TK41 camera to in order to show the original colors of objects? But a CT100 driving a CRT with modern phosphors (or LCD colors) matching HDTV/REC 709/sRGB will show the original colors of the current HDTV video and the colors would then match HDTV set colors?

The new Rec 2020 http://en.wikipedia.org/wiki/Rec._2020 for UHDTV is a very wide gamut color space. So it seems there will have to be a new DVD/Blu-Ray type of disk made to support its larger color space. Will it then be technically possible to view a version of the Wizard of Oz on a UHDTV set that can reproduce all of its colors?