Well, an "auto range" meter (fluke or otherwise) isn't necessarily able to make a DVA type readings, you can however use a DVA adapter with an auto range meter (set to DC), is this what your doing?
If however you're not and you're just using a auto range meter set to VAC and those spec'd out readings you're looking for are in Vdva/Vp as per your manual, then there's definitely going to be some discrepancies.
If so, I think that is what is going on is, your 87V and 91V are in Vrms and are correct as your meter is just giving you a Vrms interpretations of non-sinusoidal waveforms. Again, they are most likely both correct, a DVA meter would show this to be the case.
TMI..............
Some (most'ish) AC meters are bandwidth limited (say 100 or 200 hz) AND they can't necessarily integrate non-sinusoidal waveform very accurately either (stator waveforms).
So instead of beating mech's heads up with values there equipment might not be able to read, DVA readings were used.
Now, any old meter (or simply a DVA meter) switched over to DC AND using a DVA adapter could be used to measure accurately what is typically calling out as sped'd DVA values.
These readings are nothing more then the very tip top peak value of an AC (typically non-sinusoidal) waveform...For example a saw tooth, spiking transient, flyback , triangle, etc., etc. all have peak values but, are non-sinusoidal. A DVA/Peak Hold meter (or adapter on a DC meter) can easily and accurately capture that peak value and display that peak value.
So, because the AC reading you took are of a fast'ish non-sinusoidal waveform, it isn't too accurate to then use a bandwidth limited, non DVA or peak hold meter on a any non-sinusoidal waveform where the meter is expecting to measure 60'ish hz sine wave and display it in Vrms.
Multiplying your readings by 1.414 will get you close but still off some because the stator's waveform is non-sinusoidal
