(Voltage references belong to which blog???)
I am testing the 10V-Reference AD587.
1.) In an temperature stabilized housing (40°C) enclosed is...
...Reference AD587 + 220µFpolym. on 15V_Vin + 2µF noise reduction + 3mA output Load (NO Trim, NO Boost)
=> output voltage is 10.00210V - measured with Keithley 2000. (+2,1mV above exact value - OK!)
=> most of the time the voltage fluctuates less than +/-20µV possibly for 30 minutes.
=> then, without any reason, suddenly starts a downslope in 3 - 5 minutes to 9,994V (down 8mV !!!) with fast wabbly values.
=> After that starts a even faster increase (1 minute) to regain 10.00180V.
=> Once this voltage level is reached, the fluctuations are smaller again (< 100µV).
I have added a trim potentiometer with 2 Resistors (Tk 50) in series to narrow the range of adjustment.
And I have put the reference in an temperature stabilized housing.
After waiting for the case temperature of AD587 to stabilize, I adjust ref. voltage to 10,0000V (measured with Keithley 2000)
Now the weired behaviour:
- short term: Output voltage fluctuates +/-30µV (is that OK?)
- suddenly the output begins to jitter and output voltage decreases by (sometimes over) 1mV for some time.
Then suddenly the output voltage jumps to 10,0V+20µV and then the fluctuations are small again.
What is this strange behavoir about???
It seems to be that internal auto-correction circuitry becomes unstable from time to time?
It seems that using trim increases the frequency of occurrence drastically.
Even without that erratic behaviour, the medium time stability (minutes to one hour) is only 300µV peak-peak = 30ppm!
That is in strong contrast to data sheet stating "AD587 output noise is typically 4 μV p-p"
I am deeply disappointed by such a "precision".
Is there a remedy to that issue????