Calculation for Plate-to-Plate Load Impedance for a Pair of 833As??

Don't worry about it. It is not critical. The manufacturer's recommended p-p load is what gives the absolute maximum output at a given plate voltage without exceeding any of the tube's parameters like peak plate current, or plate dissipation. You can vary quite a way from the figures in the tube manual with no ill effect, except perhaps not being able to quite get as much output from the tube at the plate voltage you plan to use.

The a.c. voltage to the primary of a transformer from a pair of tubes in class-B will have a maximum peak that corresponds the lowest instantaneous plate voltage of one of the tubes. The exact figure is twice the difference between that value and the DC voltage from the supply. Each tube furnishes half the audio cycle to the transformer. Typically, at saturation, the tube will draw enough current to pull the plate voltage down to about 20% the power supply voltage. So the total peak voltage across the primary is 160% the DC plate voltage. Therefore, a 1.6:1 turns ratio transformer can be expected to furnish just enough peak audio voltage to modulate 100%. If you want more headroom, you have the options of increasing modulator plate voltage while leaving the PA voltage the same, reducing the PA plate voltage while leaving the modulator plate voltage the same, or reducing the step-down turns ratio of the modulation transformer.

Using less step down, and therefore running the tubes into a lower than recommended plate-to-plate load impedance will give higher peak output capability, but it will also increase peak plate current (up to the saturation point). Because of the higher peak output capability and greater head room, the modulator will run at lower efficiency at less-than-peak signal levels. This is exactly the same reason why AM linears have such poor carrier output efficiency.

If you are using a common power supply for modulator and final, a 1.6:1 total turns ratio at the modulation transformer will give approximately 100% modulation capability, and the output is likely to flat-top if you try to exceed this. Something like a 1.35:1 turns ratio will give good positive peak capability up to at least 125% modulation. My HF-300 rig uses an RCA broadcast modulation transformer with 1.55:1 ratio, but I run 2600 volts on the modulator and only 2000 volts to the final, to achieve some positive peak headroom.

The main thing to consider as far as impedances are concerned is whether or not the transformer runs well at the impedances that it is transforming. For example, a transformer with nominal rating of 10K to 5K should work well within a range from 20K to 10K, to 5K to 2.5K. But if you try to run it at 30K to 15 K you would probably begin to see degradation in low end response, depending on the quality of the transformer. If you tried to run the same transformer at 3K to 1.5k you would probably see a degradation in high frequency response. This is the principle used for multi-match mod transformers like the VM-5 and CVM-5. Look at the charts. There is a limited set of turns ratios available with the recommended tap connections, but the pattern keeps repeating itself over a wide range of nominal impedance ranges from the high end on the chart to the low end, so that the transformer may be useful over a 10:1 range or more of absolute impedances.

I would try to keep the tubes and transformer within +/- 50% of the nominal value, but I rarely ever pay attention to the exact impedance load, except as explained above.