I am working with the AD688 precision reference and am trying to understand how to calculate the maximum voltage error over temperature.

For discussion, the AD688ARWZ grade is sufficient. The data sheet specifies the +10V and -10V output voltage error is +/- 4mV. I assume this is characterized at ambient ~25C?

Regarding output voltage drift, for -40C to +85C it is specified as +/- 8ppm/C. Using the equation TC = [(Vmax-Vmin)/(Vnom*(Tmax-Tmin))]*10^6 and plugging in 8ppm for TC, 10V for Vnom, and 125 for delta_T, I calculate a delta_V (Vmax-Vmin) = ~ 6.5mV. Is this correct?

Does this stack on top of the output voltage error of +/- 4mV or is it the total error over temp and the +/- 4mV specified at ambient can be ignored?

Furthermore there is a +/- 1.5 mV tracking error. If I assume my above assumptions are correct, lol, and for example at max temp the output is +10.006 V, the -10V would still be within 1.5mV of that worst case value correct?

Thanks for any assistance as I am somewhat confused on how to perform Worst Case Anlaysis of the maximum output error of this reference.