Question about unprecise messurment

[MADxHAWK]

I recently bought a ads1220 module and tested it connected to an ESP32 (Heltec Wifi Kit32)
As an input i used a 100k potentiometer between VDD and GND where the output is connected to Ain0.
Using the default example compared to a precise Benchtop Multimeter i have a quite large difference between the messurment of the ADC and the Multimeter:
50mV (Multimeter: 50.34mV - ADC: 49.06mV) ~ -1.5mv difference
100mV (Multimeter: 100.36mV - ADC; 97.90mV) ~ -2.5mv difference
200mV (Multimeter; 200.15mV - ADC: 195.03mV) ~ -5-6mv diffenrece
500mV (Multimeter: 501.52mv - ADC: 489.11mV) ~ -11-12mv difference
1000mV (Multimeter; 1000.67mV - ADC 977.05mV) ~ -23-24mv diffenrece
With rising Voltage the difference between the ADC and Multimeter gets bigger, wihle the ADC show a to low voltage.

The ADC uses the +3.3V and GND of the ESP32 as supplyvoltage.
I think for a 24Bit ADC this diffenrence is a quite high between the messurments of the ADC and the Multimeter with an accuracy of 4 1/2 Digit 20000 Count +/-0.03% +3
Even the 16Bit ADS1115 giving me a more accurate reading then the 24Bit ADS1220.
Anyone alreay noticed something similar or am i doing something wrong or did i even might got a fake chip?

[dottoreD]

Can you exclude for sure that you don't have a systematic issue with your test setup? Starting with VDD is assumed to me a not good enough source to generate error free test voltage.

[MADxHAWK]

Sorry havent been around for quite some time.
Well the supplyvoltage for the ADC is quite stable at about 3.3V +/- 50mv or so, but with some noise from the powersupply.
Its not a pressision voltage source anyway.