In reviewing the RGB-seps, I noticed that color variation of one affected the other color(s). It's almost like one color putting a drain on the other, when displayed at their maximum brightnesses. Weak electronic circuits? Maybe. Also, maybe, tint or/and color controls not at their optimum settings.

I'm new to the details of analogue TV theory, but this little snippet I found ..

"They combined red (R), green (G), and blue (B) into a luminance, or brightness, channel (called Y for historical reasons), a B-Y channel (called U or Cb depending on the context), and a R-Y channel (called V or Cr). On the other end, the television would convert B-Y to B by adding Y to it, and R-Y to R the same way. Then G could be recovered because Y is a simple weighted combination of R, G, and B, so G is just Y minus a small amount of R and B. ... [They] put Color and Tint controls on the display to allow users to correct for mismatches in overall amplitude of the three channels. Broadly speaking, the Color control raises and lowers the level of the Cb and Cr channels relative to the Y channel. This can correct errors where Cb and Cr are both too high or too low by the same amount. The Tint control rotates the Cb and Cr values around the origin of a 2D space." - http://www.spearsandmunsil.com/portfolio/setting-color-and-tint-2/

.. leads me to think the TV errors mentioned therein might be the "impossible" variations in your color-bars capture.

If you're not sure how to proceed, I suggest making multiple captures and taking careful records. Work from the controls' default settings. Vary each control to create all possible control-step combinations for tint and color. Take your sample and make an RGB-sep analysis (like mine) for each one.

This should give you an actual setting to even out the color-bar variations. If not, at least you should get a feel for direction where to vary the controls in half or quarter steps for better results.