DSADC (MCP3465R) Input vs. Vref

[8086]

I'm working on a RTD temperature measurement design with the main focus of low cost (for the electronics). Using PT1000 type RTD. 1C accuracy is fine.

The MCP3465R delta-sigma ADC came up on Mouser as the lowest cost option with Vref input for ratiometric measurement.

I have come up with the circuit attached which by calculation gives appropriate values. It's based on the circuit in section 2.1 of this TI appnote: https://www.ti.com/lit/an/sbaa275a/sbaa275a.pdf . R9 is 0.1% 25ppm. Please ignore the potentially redundant filtering.

My question is this - am I allowed to have the absolute value of Vref+ lower voltage than either of the ADC differential inputs (CH0, CH1) with this part? Or must Vref+ be higher than the highest ADC input? I have checked the MCP3465R DS and I think my circuit is valid but I wanted to double check with the forum.

Thanks in advance

[Kleinstein]

With most SD ADC it should be OK. Otherwise the neg input side would be essentially limited to ground to use the full range with a gain of 1.

One may want extra diodes or so for protection.

[Nominal Animal]

I'm working on a RTD temperature measurement design with the main focus of low cost (for the electronics). Using PT1000 type RTD. 1C accuracy is fine.

LCSC sells MAX31865ATP+T for 2.75€ apiece; a known good PT100/PT1000 resistance-to-digital converter designed for the purpose.  (At Mouser, 7.50€ apiece.  Ouch.)
If you create and assemble your board at JLCPCB, they sell MAX31865ATP+ for $2.44 apiece (but minimum 4 units) when assembled there.

In any case, you'll probably want to use 3/4 wire sensors, and implement the 12th and 9th order polynomials for accurate resistance-to-temperature conversion (see ITS-90); or, alternatively, approximate it either piecewise linearly or piecewise cubicly (with a table in ROM/Flash).  Although most like to use floating point for this, it can be done using 32-bit integer fixed point math too.