I am using ADR435 as a reference for a 18 Bit ADC(AD7631). The specification demands a detailed calculation of error introduced by the reference. I have done the error analysis in following way. Please confirm that the calculations are correct and acceptable.
Part No :ADR435
It is assumed that the initial accuracy is calibrated.
Temperature Variation : 3 ppm/DegC * 20 DegC = 60 ppm ( temperature variation is from 25 DegC to 45 DegC)
Line regulation : Consider +/-10% variation in input 15V, i.e 3V.
Vout = 20ppm/V * 3V = 60 ppm
Load Regulation : Rout = 15 ppm/mA = 75 mohm; Demanded current is 50uA. So, 75mohm * 50uA/5 = 0.75 ppm
Noise : 1/f noise ==> 8 uVp-p/6.6 = 1.21 uV rms /5V = 0.242 ppm
Temperature hysteresis : 20 ppm
Long term stability : 6 months ==> 6 * 30 *24 = 4320 hours ; 40 ppm * 4320 /1000 = 172.8 ppm
Total Variation in output after 6 months = 60 ppm + 60 ppm + 0.75 ppm + 0.242 ppm + 20 ppm + 172.8 ppm = 313.792 ppm
1. While calculating "Long term stability" of this reference I have used the given specification linearly for every next 1000 hrs. But in some datasheets I have seen that the variation in Long term stability reduces in each cycle of 1000 hrs.
It is not provided in this datasheet. Could you please provide that information.
2. Please let me know if there is any wrong calculation.
3. Please provide me documents/application_notes related to voltage reference error calculations with respect to high precision ADC.