there is no specification in the datasheet.
ADR444 is based on ADI XFET technology which includes a trim terminal that allows output voltage adjustment over a ±0.5% range. For more information regarding output voltage adjustment please see applications information located on page 16 of the ADR444 data sheet.
While trimming please make sure to add series resistance R1 followed by potentiometer or fixed resistors. Please see figure below.
What trim output voltage you are trying to achieve and initial accuracy?
indeed I do not use that pin, but leave it open.
We have Problems here with the damp heat test and I wondered, whether the trim or the testpins could be sensitive to creeping current on the pcb surface.
Indeed it seems not so. When I cut off the pins, I still see the sensitivity of about 20ppm to humidity rising from 50 to 85% at 40°C within 8h and the reversal, when I lower the humidity. The Voltage was measured directly at pin6 of the ADR444B.
How can I get better results? I ordered samples of ADR4540BRZ and will make the test with them.
Do you have other suggestions?
I really appreciate for sharing your test data. Based on comparing humidity step response data of ADR444 with ADR4540, both should perform similarly. Both ADR444 and ADR4540 integrated circuit is enclosed within plastic package. Plastic package do absorb moisture which will induce mechanical stress on the die.
Option 1: Continue to use ADR444.
Apply conformal coating (humiseal) on PCB board in order to protect against humidity variation.
NOTE: Before apply any coating please makes sure PCB board is dry and clean.
Option 2: Transition to LS8/Ceramic/hermetic package voltage reference.
LS8 packaging helps to improve stability of voltage references.
Recommended parts with similar performance as ADR444:
LTC6655BHLS8-4.096: initial accuracy = 0.025%, Vout Tempco = 2ppm/*C
LTC6655CHLS8-4.096: initial accuracy = 0.05%, Vout Tempco = 5ppm/*C
LT6654AHLS8-4.096: initial accuracy = 0.05%, Vout Tempco = 10ppm/*C