Welcome in the forum!
Can you please post your complete project, so that we all can have a look at all of your settings. To do so, use
Creator->File->Create Workspace Bundle (minimal)
and attach the resulting file.
Attached is a cut down version of using a 3 channel digital mux. The DVDAC is just a test signal. The timer is just is my attempt to let the mux stabilize after switching signals. The TestTimer_1 just provides a basic 10 ms clock count. There is a warning that there is no direct path for the 1.024 ref to the mux. I am not sure how that will affect the results.
There is absolutely no need to measure the unbuffered 1.024V reference. Delete that from your topdesign.
Set the ADC's reference to "Internal Bypassed on P0_3" and provide a blocking cap (1µF) to GND on P0_3. Otherwise you will never get 20bit resolution.
I couldn't get a good value for either the buffered or unbuffered VREF. The message that there was no direct signal path from VREF to the MUX probably means that the input was floating. That explains why that value is bad.
The reason the buffered value was floating and yielding bad results is I DID NOT START THE OP AMP in main.c MY BAD!!!!!!
Documentation indicates that the bypass of PO.3 is needed for high frequency. I did put the bypass in (after I started the op amp) and there is a difference n values obtained from the SIGDEL.
Now I have to address the accuracy
I set the input to mux and corresponding pins to different sources and used my DMM to measure the voltage on the pins as the program was sending data.
Source DMM Sig Del Ratio
DVDAC set for 1000mv 0.9975 1.0234 1.025
Voltage Regulator 3.3553 3.4535 1.029
Buffered Vref 1.0233 1.0496 1.026
My tendency is to believe the DMM and conclude that the Mux/SigDel yields results that are 2.6% high and I should scale the data from other channels accordingly.
Thanks for your response.