Random generator

[meena]

Hi I am using STM32G030K8.

 

     as there is no TRNG options I am doing the ADC noise values to generate a 8 bit number.The MCU is supplied with 2.2V supply .I am using both mains and battery for my operation.However if battery is  removed Im not getting the ramndom number with mains supply .Can you please help this out 

void ADC_seed()
{
	HAL_ADC_Start(&hadc1);
	HAL_ADC_PollForConversion(&hadc1, HAL_MAX_DELAY);
	 seed = HAL_ADC_GetValue(&hadc1);
	 srand(seed);
}
uint32_t generate8DigitRandomNumber(void)
{
	
	ADC_seed();
	
}
 serialNumber = (rand() % 90000000) + 10000000;
 }

[Andrew Neil]

---

@meena wrote:

However if battery is  removed Im not getting the random number with mains supply .

---

What do you mean by, "not getting" the random number:

• The ADC doesn't work at all ?
• You get a value, but it isn't "random" ?

I think your approach is flawed anyhow: you are just taking the whole ADC value - which is likely to be pretty much stable.

The more usual approach is to just use the least significant bit(s) - as that's where the noise will be most obvious.

What is your ADC connected to? Please show the schematic.

Is it possible that the "mains" supply is earthed, so you see less noise in that case ... ?

 

Is it mere coincidence that you're the second person asking this kind of question this week, or are you working with @undacmic ?

https://community.st.com/t5/stm32-mcus-products/read-uninitialized-sram-values/m-p/761569 

[meena]

Hi, 

   I am getting the value 656284467 for serialNumber which is correct but I am getting this only if I connected my battery .I am working on a gear firmware uses Mains supply and Battery.I need this to be generated initially if the mains powered without battery .My serialNumber  is 0 if battery is not there.What may be the reason?

 

I am working in  different may be a coincidence with the person mentioned .I am not aware about that

[Andrew Neil]

---

@meena wrote:

I am getting the value 656284467 for serialNumber which is correct 

---

Pardon?

If it's random, then it should not be a fixed number - surely?

EDIT: But, surely, a Serial Number needs to be unique - not random?

 

You didn't answer the questions:

• What is your ADC connected to? Please show the schematic.

• Is it possible that the "mains" supply is earthed, so you see less noise in that case ... ?

[Peter BENSCH]

@meena The approach of reading in something ‘random’ via an open ADC GPIO is widely used with Arduino, but has nothing to do with real randomness with sufficient entropy.

True random numbers are actually not a trivial task, which was also the reason for developing the TRNG (True Random Number Generator), as found in some STM32. Depending on the entropy requirements, however, an approach such as a PNRG (Pseudo RNG) or as proposed by @mƎALLEm may well make sense.

[edit] Please read the AN4230 for more information about random numbers, NIST and statistical background. [/edit]

Regards
/Peter

[Omnigen]

ADC noise based entropy can change significantly depending on the power source. When running only on mains, the supply is often more stable and filtered, which reduces measurable noise compared to battery operation.

You may want to:

• Check if ADC reference and ground are introducing enough thermal or quantization noise

• Try sampling a floating or weakly biased ADC pin

• Increase oversampling and mix multiple readings (with bit shifting or XOR)

• Add intentional jitter sources (RC noise, internal temperature sensor, etc.)

For quick testing or validation, I sometimes compare MCU-generated values with simple external random generators or word-based entropy tools during experiments. I’ve used this small utility set for reference when checking randomness behavior:
https://omnigentools.com/ 

Hope this helps narrow down the issue.