One thing that I haven't seen mentioned is the temperature compensation effect of the two 'normal' diodes in series with the zener.
If you look at the this paper it explains the reasoning- https://www.edn.com/design/analog/4334638/Designing-a-zener-diode-regulator
When you heat or cool the zener you do it in isolation, what you should be doing is applying the heat/cooling to all three components together, only then can you see the actual effect.
Zeners below about 5.6V have a negative temp coefficient, those higher have a positive coeff due to the avalanche effect. I guess Sony picked the setup to minimise the overall effect.
You can probably buy modern equivalents for D3,4 and 5. which will reduce the effect, but probably not by much.
The engineers in those days did know their stuff. I worked with some of them.
The main reason I've considered using modern voltage regs/refs is that I figured the engineers knew their stuff and did the best with the available tech, and that aside from putting things back into spec, the greatest opportunity for long term improvement was in leveraging modern ICs that no one had access to or even conceived of in the late 1960s when this design was probably conceived.
It's interesting to me that Sony calls out a standard resistor where an NTE thermistor was actually installed. SM oversight or last minute attempt to fix drift that other compensation methods (the series diodes mentioned) failed to cure?
I honestly think the primary advantage of quartz lock or other purely frequency based control schemes is less in absolute accuracy (for to my ear, standard servo referenced tables can have perfect pitch to the human ear) but in partially decoupling the feedback loop from variances in components, whether from age or initial tolerances. As long as things don't drift so far out of spec that the circuit ceases to function, the reference frequency and corresponding output frequency from the table are nearly fixed reference points that remain unaffected by temperature drift. If the components in the servo drive drift over time or temp, the servo feedback loop can simply compensate to keep the frequency comparator happy.
Cheers,
Nathan