Background radiation of a Geiger Counter

I started a long awaiting project, to build a Geiger-Müller radiation counter. The Geiger tube is a SBT-11 russian tube capable of measuring alpha, beta and gamma radiation. The tube is quite sensitive and of a relatively small size. The picture at the right hand side shows the SBT-11 tube and the high voltage power supply in a plastic CCFL case. The power supply works with 3.3V. A small CCFL transformer boosts the 3.3V to about 390V. The switching regulartor IC is a MC34063.

The datasheet says the tube has 0.6 pulses/sec of own background. 0.6 pulses/sec are about 36 pulses/min. To get the prove I started to count the pulses.

The Measurement

The common method to measure pulses is a round robin array to save the pulses and the sum of pulses yelds the CPM.
The code below shows an Arduino example:

ISR(each_second) {
   int pulses_per_second = count - last_count;
   register unsigned int cpm_pos = it_cpm++ % 60;
   cpm[cpm_pos] = pulses_per_second;

   int sum_cpm = 0;
   for(int i=0;i < 60;i++)
      sum_cpm = sum_cpm + cpm[i];
   Serial.println(sum_cpm, DEC);
}

This can be surely simplified by changing the sum by 1/60 for one minute average:

uint16_t hist[80] = {0};
uint16_t sum_cpm = 0;
ISR(each_second) {
  uint32_t count0 = count;
  int pulses_per_second = count0 - last_count;
  last_count = count0;
  sum_cpm = (uint16_t)(0.5 + 59.0 * sum_cpm / 60.0 + 1.0 * pulses_per_second / 60.0);
  if (sum_cpm < 80)
    hist[sum_cpm]++;
  Serial.println(sum_cpm, DEC);
}

The Result

The diagram on the left hand side shows the histogram of about 20 hours measurement. The own background of the tube has an intervall of 0 to about 86 pulses per minute. It shows (as expected) a discrete normal distribution with a mean of 36-38 pulses per minute or 0.6-0.63 pulses per second. So what does it say? This may concern everybody who needs to measure low radiation or background radiation. Usually it is not needed to substract the 0.6 pulses from result, because it is more expressive to give an confidence interval as an result.