Fastino DIOT + HVAMP temperature stability estimate

[nkrackow]

We were thinking about stability/repeatability of Fastino-DIOT and HVAMP V2 recently so I performed the following estimate for a worst case tempco of the full system:

device	description	tempco	effective contribution to +-20 V outputs
LTC6655	main reference	2 ppm	80 uV
ADA4528	ref buffer offset compensation	15 nV	120 nV (aka negligible)
AD5542	DAC	0.05 ppm	2 uV
OPA2197	output amp/offset	2.5 uV	20 uV
OPA2197	output buffer	2.5 uV	5 uV
Vishay Y1747	output amp/offset resistors	0.5 ppm	20 uV
LT1028	HVAMP amp	1 uV	2 uV
Vishay NOMCA	HVAMP resistors	5 ppm	200 uV

Adding them all up to get the worst case: 329 uV/°C or 8.2 ppm/°C

Assumptions:

• LT3042 reference buffer is in loop with ADA4528 and therefore compensated
• Used "tracking" (aka relative between in-package resistors) tempco for all Vishay resistor networks

Is there anything obviously wrong with my estimate? The real devices will likely be a lot better than the worst cases from the datasheets and might even cancel out if we're lucky.

The Vishay resistor network for HVAMP is the only SOIC-16 10K one I found. Is there a better one? That is obviously the biggest contributor to thermal instability on paper.

I cannot think of any hysteresis in the system which would lead to repeatability issues apart from thermal.

We would like to perform our own tests, ideally even with a thermal chamber some time in the future. I'll post the results once we can show them.