GQ Electronics Technical Support Forum
.


GQ Electronics Technical Support Forum
Home | Profile | Register | Active Topics | Members | Search | FAQ
Username:
Password:
Save Password
Forgot your Password?

 All Forums
 GQ Electronics Forums
 2.GQ Geiger Muller Counter
 Regular recurrent patterns in per second recording
 New Topic  Reply to Topic
 Printer Friendly
Author Previous Topic Topic Next Topic  

matiahi

6 Posts

Posted - 05/20/2021 :  22:52:50  Show Profile  Reply with Quote
I have recently bought a GMC-320plus detector and I am not a specialist in these things.

Currently I am living in my home country in the middle east.
I have recorded the radiations since 30.04.2021 in my living apart
which is also the place that I use to work on my master research project.

Now I have the following three questions in this regard which are very important to me and they are the only reason that I bought this device and they are a matter of huge concern for me.

1. What is the nature of these radiations (X-ray, Beta, ...)?

2. Are these radiations natural or man-made? I ask this because I have noticed a high level of correlation between the noises that I hear from my neighbor and the intensification of the level of radiations. Also I have noticed a lot of regular and recurrent patterns and symmetries in the detected signals recorded "per seconds", such as:

Lots of:
0,1,1,1,1,1,1,1,0 (reminds me of Fluoroscopy imaging during angiography)
1,2,2,2 or
2,2,2 or
2,3,2 (which I call searching impulses to find my position in my ~190 square meter apartment)
0,1,0,2,0,1,0 (hand gesture)
0,0,0,0,1,0,1,0,1,0,0 (relaxation phase to keep CPM low so as not to arise suspicion)

Please give priority to the data recorded between 21:30 and 03:00 in your analysis.

3. If they are man-mad, how dangerous is it for my family to live in this apartment for years while being subjected to these type and levels of radiation?

Thank you very much in advance.




Reply #1

ullix

Germany
602 Posts

Posted - 05/21/2021 :  05:42:08  Show Profile  Reply with Quote
First, nothing in your description suggests that you live in any kind of danger zone, and all things you see seem to be of natural origin.

I suggest you switch your counter to the CPM (counts per minute) mode, because at the low levels which you observe, the numbers will be a bit easier to comprehend.

The essential part in any Geiger counter is the Geiger-Müller tube in it. The GMC-3XX series has a glass tube, possibly type M4011, in it. This tube is capable of registering Beta and Gamma radiation. Alpha cannot be registered. X-rays also belong to the Gamma radiation; they can be registered, provided their energy is high enough.

The counts which you see is the natural background radiation. It is made up of two sources: one is cosmic radiation. When you take the counter into an airplane (is allowed) you will see that the background rises. The other is radiation from earth. Both building materials and anything outside does contain naturally radioactive substances. Two prominent ones are Uranium and its daughter nuclei, and Potassium, of which the isotope K40 is radioactive.

You actually have a third source: the counter creates some electronic noise, which is also counted. GM specifies this as :
quote:
Instrument Background: < 0,2 pulses/s

equal to CPM=12.

The background which you will measure depends on your location, altitude, rock composition, building material, and other. In my area the typical background read from the counter is CPM=20. Thus this includes the electronic noise of CPM=12, so the radioactive origin is only CPM=8! One typically provides only the read-off numbers, and omits the correction. At this low counts it is immaterial anyway.

The fluctuating numbers result from the statistical property of the decay of radioactive material. While being random, they are actually very precisely following a Poisson statistical distribution.

If you want to dig deeper, you find more details on counters in the manual of my GeigerLog software (open source, free to download and use): https://sourceforge.net/projects/geigerlog/

For some experiments I suggest to use Potassium, which you find in garden fertilizer and other stuff. More on that in my "GeigerLog-Potty Training for Your Geiger Counter" article, here: https://sourceforge.net/projects/geigerlog/files/Articles/

(In case you tried to attach data: there aren't any!)

Go to Top of Page
Reply #2

matiahi

6 Posts

Posted - 05/24/2021 :  23:48:08  Show Profile  Reply with Quote
Thank you for your reply and sorry for the delayed response.
Actually what I am concerned with is the pattern of variation of the signals registered by the detector, not the value of readings.
The pattern seems to be artificial and non-natural/non-random.
To show that I have to attach the relevant data files. But I do not know how to upload the data files here. It only accepts images.
Go to Top of Page
Reply #3

ullix

Germany
602 Posts

Posted - 05/26/2021 :  00:14:29  Show Profile  Reply with Quote
Such is statistics! All is fine.

You still didn't mention what CPS or CPM values you see. From the few numbers you showed, I guess it will be something near the CPM=30, or CPS=30/60=0.5 range.

For CPS=0.5 here are a few data created synthetically from a Poisson distribution using GeigerLog (4th column is CPS):

#HEADER ,2021-01-01 00:00:00, SYNTHETIC data: WhiteNoisePoisson_CPSmean=0.5
#LOGGING,2021-01-01 00:00:00, Start: cycle 1.0 sec, device 'SYNTHETIC'
       0,2021-01-01 00:00:00, 0, 0
       1,2021-01-01 00:00:01, 0, 0
       2,2021-01-01 00:00:02, 0, 1
       3,2021-01-01 00:00:03, 1, 0
       4,2021-01-01 00:00:04, 1, 1
       5,2021-01-01 00:00:05, 2, 0
       6,2021-01-01 00:00:06, 2, 0
       7,2021-01-01 00:00:07, 2, 1
       8,2021-01-01 00:00:08, 3, 0
       9,2021-01-01 00:00:09, 3, 1
      10,2021-01-01 00:00:10, 4, 0
      11,2021-01-01 00:00:11, 4, 0
      12,2021-01-01 00:00:12, 4, 1
      13,2021-01-01 00:00:13, 5, 0
      14,2021-01-01 00:00:14, 5, 3


When you look through a long enough set of such data, you will find your data.

You can read your data into GeigerLog and do a Poisson test. If you don't know how to do that, upload the data (best is a hist download from your counter) to the GeigerLog web site, topic Discussion, as others have done it already. https://sourceforge.net/p/geigerlog/discussion/ I'll do a test
Go to Top of Page
Reply #4

matiahi

6 Posts

Posted - 05/26/2021 :  01:23:07  Show Profile  Reply with Quote
Thank you for the directions. I uploaded the files in the https://sourceforge.net/p/geigerlog/discussion/ forum.

Regarding the CPS and CPM, I have to say that even though the system is set to record EVERY SECOND, the CPM values
are also recorded in the *.csv files.

The maximum recording is on 5/20/2021 6:01:00 PM which goes as high as 0.377 uSv/h or 58 CPM.

But as I have clarified in my first post, it is the regular pattern of the pulses that I am concerned with.

Go to Top of Page
Reply #5

ullix

Germany
602 Posts

Posted - 05/26/2021 :  03:05:14  Show Profile  Reply with Quote
Thanks, Matiahi. If anyone is interested in some genuine, nice counter data, I encourage you to download and take a look.

I used one of the *.bin files which is just short of 1 million records, covering more than 11 days.

The counter had been set to CPS recording. CPM is calculated from CPS, both in GeigerLog as in the creation of CSV files.



The upper left portion shows the complete time course as read from the counter. Magenta is CPS, the original data, Blue is CPM, derived from CPS.

I always recommend to show the Moving Average, the yellow double-thin line wiggling in the middle of the blue. Clearly, no long term trend is visible; the data are constant over time.

There are two exceptions in this graph: one is near 3 days marked with a red 1, the other is near 11 days, marked with a red 2. The two smaller pics show what it is: @1 looks like the counter had been reset, which resets CPM to zero, and so it takes 60 sec before it reaches proper CPM again. @2 is a spike in CPS of some 20+ counts. This looks like some artifact, perhaps electronic noise impacting the counter.

So I limited the analysis from t=2.9d ... 11.4d to exclude those artifact regions from the CPM analysis, and from t=0 ... 11.4d for CPS.

For analysis simply press the Poiss button. GeigerLog calculates for a few minutes (!, these are big files) and shows the bottom left picture for CPM.

On first glance you see the data from a near perfect Poisson distribution. R2 (=r-squared) is exactly 1 in the best case, and is 1.000 here. CPM average is 20.12, and CPM var is 20.80. In the perfect case, these two numbers are identical. Further, the residuals are as close to the bottom line as you may ever see.

In the CPS case all numbers are perfect, but this is less meaningful because you have only 6 different bins to fit: 0, 1, 2, 3, 4, 5. CPS avg in 0.33. As a rule-of-thumb: look at CPS only when its average is >= 10.

Overall: this is as good as it gets; data are in a quality ready to be put into a text book. And not the slightest hint of anything unusual with respect to radioactivity (and perhaps you have an idea where this supposed electronic spike might have resulted from; but those spikes do happen with counters).

Counts are random, though strictly governed by Poisson. Strange sequences of numbers may result. Just convince yourself by loosing some money at the Roulette table ;-)

An intro on Poisson in my "Potty Training" https://sourceforge.net/projects/geigerlog/files/Articles/GeigerLog-Potty%20Training%20for%20Your%20Geiger%20Counter-v1.0.pdf/download

Beyond that ask Google.




Go to Top of Page
Reply #6

matiahi

6 Posts

Posted - 05/28/2021 :  03:28:47  Show Profile  Reply with Quote
Thank you Ullix for your analysis.

There are some parts in your description that are not clear to me:

1. By Poisson, do you mean the Gaussian distribution?
2. What is r^2? Is it variance of the distribution? And why is it the best case if the radiation has an r^2 equal to one?
3. Why CPM average being equal to the variance indicates a perfect case of radiation detection?

Edited by - matiahi on 05/28/2021 03:32:11
Go to Top of Page
Reply #7

ullix

Germany
602 Posts

Posted - 05/29/2021 :  00:29:43  Show Profile  Reply with Quote
Gaussian distribution, Bell-curve, Normal distribution are different names for the exact same thing. I prefer the name Normal dist.

A Normal distribution is fundamentally different from a Poisson distribution!

A Normal dist is valid from minus infinity to plus infinity, and for every number in between. Poisson is valid from 0 (zero) to plus infinity, but ONLY for integer values, so like: 0, 1, 2, 3, ...

Normal is often useful to describe the error in measurements of smoothly changing parameters, e.g. temperature, voltage, speed. Poisson only applies when you can count it: car crashes on an intersection, whale sightings in a bay, counts from a Geiger counter.

A Normal dist is described by 3 parameters: average, width (=std.dev., 'standard deviation'), and height. A Poisson needs only 2 parameters: average, height. The width is already known as std.dev. = square_root(average). As 'variance' is the square of the std.dev., it follows that for a true Poisson dist average = variance! If this is not fulfilled, it is not Poisson, and a Geiger counter measurement is not valid!

When we graphically depict these distributions, they may look like this:



On the left side the dist are drawn with average=20. This implies that std.dev for Poisson is sqrt(20)=4.472 and so Normal is drawn with this std.dev! One is hard-pressed to see a difference between these two curves. However, keep in mind that the Normal is valid on all points on the black line, while Poisson is valid ONLY at the dots. It does not exist between them. The thin read line is drawn only to guide your eyes from one valid point to the next.

On the right side, drawn with average=2, you see where Normal even fails graphically: not only is the difference between the two curves more pronounced, but the hatched area for Normal shows negative counts, which obviously cannot exist.

R-squared (here as r2) is a statistical property routinely used to describe how well your model for the data fits the data. It has nothing to do directly with Poisson's or any other distribution's property. By the way r2 is calculated, it is 0(zero) when there is no agreement at all, and 1 if there is perfect agreement. (it can be negative when the data are really crappy!)

Here you have the data from your counter, and your model is the Poisson curve. In my above Reply#5 you have in bottom-left of the picture your data, the blue columns, and the thick red line the Poisson model. Already visually they do agree, and R2=1.000 is just a confirmation. With avg=20.12 and var=20.80 it is close but not perfect. As a 4th thing to do, look at the residuals. They should wiggle around the bottom line with no discernible pattern. Considering all 4 things, this is a measurement of text book quality. It will not get any better.

Here an example of a very recent GMC counter firmware "improvement", which got terribly wrong, ruining otherwise good data:




Apply your 4 criteria from above the see why it is awful.

A Poisson test of your data is possible only with my GeigerLog software https://sourceforge.net/projects/geigerlog/ and in GeigerLog easily achieved by just pressing a button.


Edited by - ullix on 05/29/2021 01:16:57
Go to Top of Page
Reply #8

matiahi

6 Posts

Posted - 05/29/2021 :  10:40:19  Show Profile  Reply with Quote

Thank you Ullix for your thorough and clear explanation.

What I am thinking now is that isn't it possible that one can use an X-ray equipment to do imaging/spying on his neighbors by operating the
X-ray machine for 30 seconds as he wishes and then switching over the control to a smart system which would then generate patterns for the next 30 seconds in a way that would satisfy the Poisson distribution?

What I am trying to say is that even though your analysis is very interesting and convincing, but it merely analyzes the information in the magnitudes of the data samples and ignores the information contained in the time variations of the magnitudes. I am not a mathematician,
but I think that there should be some significant correlations in the pattern of variation of the samples over time.


Go to Top of Page
Reply #9

Damien68

France
516 Posts

Posted - 05/29/2021 :  12:30:41  Show Profile  Reply with Quote
an Xray machine would make a lot of more clicks and would be Poissonian no need to spoof it.

there are not many numbers (0, 1, 2 and 3) so patterns can repeat themselves only by chance and it is still purely random, at no time you can predict in advance what will happen.
if you try to predict what will happen at the next CPS you will have around 1 chance over 4 to be right and 3 chances to be wrong. which is normal. if you choose to predict a 0 or a 1 you will have more chances to be right because there are more than 2 or 3.
The dice are in the place.

Mastery is acquired by studying, with it everything becomes simple

Edited by - Damien68 on 05/29/2021 13:58:19
Go to Top of Page
Reply #10

ullix

Germany
602 Posts

Posted - 05/30/2021 :  23:50:16  Show Profile  Reply with Quote
So your neighbor is bugging you with a high performance, pulsed X-ray machine mimicking Poissonian counts in your counter? Only one solution: put on the tin-foil-hat!
Go to Top of Page
Reply #11

Damien68

France
516 Posts

Posted - 05/31/2021 :  02:02:38  Show Profile  Reply with Quote
now the martians exist, OK for the moment they are still only robots (rovers, helicopter, ...)

Mastery is acquired by studying, with it everything becomes simple

Edited by - Damien68 on 05/31/2021 08:21:01
Go to Top of Page
Reply #12

matiahi

6 Posts

Posted - 06/01/2021 :  00:45:28  Show Profile  Reply with Quote
Thank you all for your contributions.
By the way, tin foil may be suitable for another range of the electro-magnetic spectrum. For x-ray one needs lead cover.
Go to Top of Page
  Previous Topic Topic Next Topic  
 New Topic  Reply to Topic
 Printer Friendly
Jump To:
GQ Electronics Technical Support Forum © Copyright since 2011 Go To Top Of Page
Generated in 0.31 sec. Snitz's Forums 2000