using high speed ADC with noise

I would like to get help with ADC due to my limited knowledge.
Currently, I am using 16bit ADC LTC2368-16((https://www.analog.com/media/en/technical-documentation/data-sheets/236816f.pdf),) with Arduino Uno for making measurements at very short intervals, but having problems with noise.
First, let me show the connections to LTC2368 in the figure below, the first attached picture.

Following the datasheet, I set up the device to ‘normal mode, single device’ configuration, with CHAIN and RDL/SDI connected to the GND. When I connected 2.5V reference voltage to IN+ and Vdd and measured the output, peek-to-peek noise was at 50~100LSB (ex. OUTPUT = 10000 ~10050)
At rare moments, the peek-to-peek noise stays very low, below 8LSB, but if I restart Arduino several times or plug the cable at a different port and reconnect, then the noise increases by significant amount, back to 50~100LSB.

Strangely, when I connected 2.5V reference voltage to Adafruit’s ads1115 breakout board (ADS1115 16-Bit ADC - 4 Channel with Programmable Gain Amplifier [STEMMA QT / Qwiic] : ID 1085 : $14.95 : Adafruit Industries, Unique & fun DIY electronics and kits) for comparison and measured the output, peek-to-peek noise was kept at 2LSB, even after a long time.
According to the data sheets, both ADCs have INL of below 1LSB, and while ads1115 has
2.45 times larger input voltage per 1LSB (LTC2368 : 5/2^16, ads1115 : +-6.144/2^16) than ltc2368, LTC2368 has much larger peek to peek noise. Comparing with ads1115, 8LSB seems to be adequate level of noise, but it’s very rare as I’d mentioned above.
I tried connecting additional capacitors following the demoboard example on the datasheet in order to reduce noise but only marginal difference. As shown below, I added 0.1uf and 10uf capacitors to Vdd and OVdd, and 0.1uf and 47uf capacitor to REF PIN in parallel.

What could be the possible cause of large peak-to-peak noise? If I am not understanding something right, which parameter from datasheet should I look into?
Shown below are the Arduino code for measurement and the timing diagram of LTC2368. For Ads 1115, Adafruit’s library was used and the rest are the same.

Thanks in advance for your help
and have a nice day!

#include <SPI.h> 

const int adcCs = 9;   // CNV pin

float volt;
unsigned int val = 0;
unsigned int temp, tempmax = 0, tempmin = 65535;
byte spiBit[2];


void setup() {

pinMode(13, OUTPUT);   //  SCK
pinMode(adcCs, OUTPUT);
digitalWrite(adcCs, LOW);
SPI.begin();
SPI.beginTransaction(SPISettings(200000000, MSBFIRST, SPI_MODE2));
digitalWrite(13, HIGH);   //  idle when high

Serial.begin(9600);

}

void loop() {

ReadAdc();

val = spiBit[0];

//Serial.print(spiBit[0],BIN);
//Serial.print("   ");
//Serial.print(spiBit[1],BIN);
//Serial.print("   ");
//Serial.print( val<<8|spiBit[1] ,BIN);
//Serial.print("  AAA  ");

temp = val<<8|spiBit[1];
if (temp > tempmax) tempmax = temp;
if (temp < tempmin) tempmin = temp;
volt = (val<<8|spiBit[1])*0.000076295109;

Serial.print(temp);
Serial.print("  :  ");
Serial.print(tempmin);
Serial.print("   ");
Serial.print(tempmax);
Serial.print("   ");
Serial.print(tempmax - tempmin);
Serial.print("   ");
Serial.println(volt);
delay(10);

}

void ReadAdc(){

digitalWrite(adcCs, HIGH);

delayMicroseconds(1);   //  minimum conversion time 527ns

digitalWrite(13, LOW);
spiBit[0] = SPI.transfer(0);
spiBit[1] = SPI.transfer(0);


digitalWrite(13, HIGH);
//delayMicroseconds(1);
digitalWrite(adcCs, LOW);
//delayMicroseconds(1);
}

Please edit your post to add code tags, as described in "How to use this forum".

OP's images, posted properly:
Connection


Capacitor

timing

//jremington

thanks a lot, i modified
and I will be able to reply after 7 hours or more, because here is Korea, midnight.

What could be the possible cause of large peak-to-peak noise?

You seem to be trying to use this device on solderless bread board. That is probably the reason why you get poor noise performance.

To achieve low noise performance on a 16 bit A/D you need a good layout, a PCB and ground plane and proper shielding, not loads of superfluous little wire antenna picking up noise like you have now.

Also you don't seem to have connected the ground connections to the chip. What precautions have you taken to isolate the analogue side of the chip from the digital side?

Hi,
First of all, i have connected all the pins which illustrated with GND above in the body of this page to common ground, as i re-drew below

and as you mentioned, poor wire connection may be the part of cause of noise.
As i replace pin head wire with shorter jumper wire, signal became stable, especially, when wire for IN+ PIN is replaced. So, i am soldering with shield and finally will draw pcb design for this.

Unfortunately, because of my poor knowledge for electronics, I can not fully understand isolation of ground.
I think that separating ground you are saying may be like picture below

(Staying Well Grounded | Analog Devices)

Anyway, I have to learn more about ground separation, and your advice that loads of
superfluous little wire antenna pick up noise is right.

thanks a lot, you have saved my one week. I should have to met you faster
I appreciate it, have a beautiful day

staying well grounded.png

I suspect the distance from the chip to the decoupling cap is a factor - normally the datasheet will give a suggested pcb layout for the chip - use it if you want datasheet performance.