Not unique on Mega2560 - Githubissues

haimiko commented 4 years ago

After flashing a few dozen Mega2560 I stumbled on duplicate serial numbers.

ricaun commented 4 years ago

Hello @haimiko

The Mega2560 does not have an official serial number on the datasheet.

So a repeat serial number could be should occur... 😞

Could you try the code below and share the result, the code shows all the signature bytes of the mcu on the serial monitor.

#include <avr/boot.h>

#ifndef SIGRD
#define SIGRD 5
#endif

void setup() {
  Serial.begin(115200);
}

void loop() {
  // put your main code here, to run repeatedly:
  Serial.println("");
  for (int i = 0x00; i < 0x40; i++ ) {
    if ((i % 0x10) == 0)
      Serial.println();
    byte b = boot_signature_byte_get(i);

    if (b < 0x10)
      Serial.print("0");  
    Serial.print(b, HEX);
    Serial.print(" ");
  }

  Serial.println("");
  for (int i = 0x00; i < 10; i++ ) {
    byte b = boot_signature_byte_get(0x0E + i);
    if (b < 0x10)
      Serial.print("0");  
    Serial.print(b, HEX);
    Serial.print(" ");
  }
  Serial.println("");
  delay(5000);
}

I appreciate if you could share the boards with the same serial number.

Thanks

haimiko commented 4 years ago

Only encountered 2 duplicates so far. Here is the output:

1st Mega2560: returned serial #: 6E756E6B776F02C

1e 9c 98 ff 01 ff ff ff ed ff ff f7 ff ff 6e 75
6e 6b 77 6f ff 00 02 0c 17 02 12 08 13 08 ff ff
ff ff ff ff ff ff ff ff ff ff ff ff 70 1c ff ff
ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff

6e 75 6e 6b 77 6f ff 00 02 0c

2nd Mega2560 returned serial #: 6E756E6B776F02C

1e 55 98 ff 01 ff ff ff ed ff ff f7 ff ff 6e 75
6e 6b 77 6f ff 00 02 0c 17 02 12 08 13 08 ff ff
ff ff ff ff ff ff ff ff ff ff ff ff 30 1c ff ff
ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff

6e 75 6e 6b 77 6f ff 00 02 0c

Thanks

ricaun commented 4 years ago

Cool! Looks the same and it's bad...

You found two chips with the same id 😆

The code is based on the https://loranow.com/uniqueid/ and this old datasheet of the atmega328pb

The Device Signature is the same: 1e9801 - Mega2560 RC oscillator Calibration byte: Is different and it's ok! The rest of the byte looks the same, except the byte 0x002E with the value 70 and 30 probably is not related to the supposed serial number.

Congratulation! You have two devices with the same serial.

Like I said It's not an official serial number and you have the luck and found two mcu really similar.

How many Mega2560 did you test? Could you share other output, maybe we could use another byte on the serial number o make this odd case different?

I have these two Mega2560 laying around the lab, and this is the output.

1e a5 98 ff 01 ff ff ff ed ff ff f7 ff ff 59 36 
35 30 32 37 ff 10 17 0e 17 08 12 07 13 c0 ff ff 
ff ff ff ff ff ff ff ff ff ff ff ff 70 1c 3f 0c 
a2 26 c3 02 a3 0a 90 01 ff ff ff ff ff ff ff ff 

59 36 35 30 32 37 ff 10 17 0e

1e af 98 ff 01 ff ff ff ed ff ff f7 ff ff 59 35 
39 30 30 30 ff 0c 0d 0d 17 08 12 07 13 07 ff ff 
ff ff ff ff ff ff ff ff ff ff ff ff 40 1c b7 0c 
ba 27 e4 02 33 0b d2 01 ff ff ff ff ff ff ff ff 

59 35 39 30 30 30 ff 0c 0d 0d

The byte 0x002E is a good candidate, but in the end, I guess is not unique and a big disclaim saying that is better them trying to make different using some unknown z-address.

Sorry for the long response... 😆

See yaa!

haimiko commented 4 years ago

That's what I was afraid of. As you said the datasheet makes no mention of this, so it was an assumption that they are unique. I've tested about 20 so far and only encountered this once. Thank you for sharing.

haimiko commented 4 years ago

Just encountered another duplicate. Mega2560's were not engineered to have unique serial numbers, unlike the mega328pb.

ricaun / ArduinoUniqueID

Not unique on Mega2560 #6