|
Didn't take pictures last night. I decided that it still
was not right. Guess was that the drop in luma when full brightness
full saturation R appeared was caused by clipping in the detector
because the luma carrier was too far down on the 45 to 46.5
MHz slope, and the chroma was much much too high. I had looked
at the response crudely using a CH 10 rf sweep and it looked
like a sin curve from 46 to 41.25 MHz for the IF. I was unsure
of this since I didn't have a proper bias box.
Today I did an alignment. But there was a problem: the
cheapo Eico generator I got off Ebay leaked RF out the line cord.
That was fixed by removing the knot from the line cord,
shortening the leads to it, and installing new and better safety caps
as close as possible to the line in, the ground end soldered directly
to the chassis. The caps should be broadly series
resonant somewhere between 40 and 160 MHz. This worked.
Then I did the "Overall IF Alignment" on page 19 of the
RCA service manual. All the traps were already quite accurate so
I touched only 1T109, 1T110 and 1T111. This took a while with careful
notes of where I left everything since I wanted to be able to go back
if needed. 1T109 and 1T111 were within 160 degrees of
correct, but 1T110 had to go 1 1/3 turn clockwise. This resulted in
an almost perfect curve. The little bump
below 41.25 MHZ was a bit smaller than shown in Fig. 27.
I had earlier today generated true proper test patterns for an
IQ color set. I will make these available somewhere. They are
jpeg files generated in Photoshop by calculating Y, I, and Q levels
mathematically for various color patches.
The 2 main files have +I, -I, +Q, and -Q all at the same color level
and the luma calculated to make high saturation without going more than 5%
blacker than black or whiter than white. One file has full bandwidth
color (to 4MHz) in the jpeg, the other has I limited to 1.5 MHz and
Q to 0.5 MHz. (This was done by converting to Yab in Photoshop and
applying a "motion blur" to a and b only.) The -I to +Q transition was
filtered at 0.8 MHz.
I then copied the files to my Sony Blueray player and played them.
The result on the filtered IQ test are amazing. There are no off-color
fringes except a tiny one in -Q. The +-I transition is clearly three times
the resolution of the +-Q one. The unfiltered file shows
the expected 0.5 to 1.5 MHz crosstalk that so
mars cheapie color decoders (and encoders).
Its not as good as my 55 inch LCD TV on the same NTSC, but its close.
And the problem with saturated reds and yellows is gone.
Next is dinner.
|