Thanks to everyone who responded.

I agree with Franc's interpretation of the SanKen data sheet. Moreover, it
is possible that the IC's I have were made to slightly different criteria
(after all, the data sheet is from the late '80's), or subject to
manufacturing variables. (In this regard the IC's I used are new, branded
SanKen and from a reliable supplier.)

In addition to Chuck and Leonard's comments, a few additional observations
suggest to me that my original results might not be as serious an issue
as I first thought. I come to this, in particular, as a result of seeing
SanKen (and NTE) data sheet.

The data sheet indicates an output voltage of 135 +/- 0.8V with Iin of 7mA.
Iin at 7mA is not much of a load, certainly far less than the real-time
situation in which I am measuring (Iin around 500 ma.) The specification
for load regulation is 1V between 250 to 500 mA Io. What happens
between the no-load (7mA) level and 250mA is not indicated.

Even at the specified Iin of 7mA, the data sheet does not say what the DC
input voltage is. The NTE data sheet refers to Vout of 135V +/- 1V at a Vin
of
120VAC. With full wave rectification, a good sized filter capacitor,
virtually no load, and a nominal RMS AC line of 120V, I would expect the
input voltage to be close to 170VDC.

Here are my measurements, with the TV running and Iin around 500mA.

AC Line input VDC output VDC

118 153 130
119 155 132.5
123 160 134
126 165 134.4
130 170 134.9
134 175 135.4
138 180 135.8

(The normal line voltage here is 120VAC.)

There could be measuring differences: The TV's circuit and that of the data
sheet are not exactly the same. The data sheet has R3 of 12K Ohms and R4 of
220k Ohms whereas in the TV the equivalent resistances are 15k and 470k.
Also, the DC input to the regulator has about 6V p-p sawtooth ripple
(there's no RC filter as in the test circuit, only a capacitor). I don't
know how well the IC handles this. Moreover, I don't know how my
digital meters handle the ripple, on top of their normal accuracy
tolerances. If it is a simple averaging of the ripple, the DC input
voltages that I have recorded at different line levels might be incorrect,
say, within a 3V range.

All to say that I tend to agree with Chuck and Leonard that my observation
of input regulation is probably not a defect. I would add that the raster
is quite stable with varying brightness.

Michael