negative SampleIndex? repeating: -29 to 0

mittenchopsmittenchops
edited December 2014 in Software
I'm at the hackathon now, and having some trouble parsing the output from OpenBCI_GUI_simpler/SavedData

Most of the files are empty and look like this:
%OpenBCI Raw EEG Data
%
%Sample Rate = 250.0 Hz
%First Column = SampleIndex
%Other Columns = EEG data in microvolts

Some of them do something weird where the indexes start at -29 and move to 0, before going back to -29.

These look like this:
%OpenBCI Raw EEG Data
%
%Sample Rate = 250.0 Hz
%First Column = SampleIndex
%Other Columns = EEG data in microvolts
-29, 84792.34, 108628.17
-28, 65161.95, 83489.14
-27, 59106.45, 75716.46
-26, 68452.14, 87705.66
-25, 70581.50, 90436.42
-24, 62344.30, 79895.67
-23, 57696.71, 73947.23
-22, 56784.04, 72788.88
-21, 54148.61, 69429.67
-20, 52055.78, 66751.60
-19, 50621.60, 64909.21
-18, 49421.29, 63394.99
-17, 47789.21, 61300.32
-16, 45793.13, 58751.91
-15, 44741.04, 57391.16
-14, 44267.51, 56809.25
-13, 42779.54, 54899.25
-12, 39156.68, 50277.70
-11, 38010.88, 48804.92
-10, 35913.93, 46130.87
-9, 34333.75, 44112.01
-8, 33476.85, 43011.57
-7, 30334.22, 38988.25
-6, 29897.27, 38436.17
-5, 29216.10, 37566.84
-4, 27743.09, 35682.45
-3, 26431.25, 34002.05
-2, 25857.90, 33265.88
-1, 25258.36, 32508.02
0, 24070.52, 30983.09
-29, 22853.65, 29433.04

The final format that looks more reasonable, is what's contained in: openBCI_raw_2013-11-15_09-55-27.txt

%OpenBCI Raw EEG Data
%
%Sample Rate = 250.0 Hz
%First Column = SampleIndex
%Other Columns = EEG data in microvolts
0, -39.12, 0.00
1, 263.82, 0.00
2, 160.04, 0.00
3, -287.82, 0.00
4, -42.94, 0.00
5, 120.30, 0.00
6, 147.72, 0.00
7, -4.52, 0.00
8, 262.81, 0.00
9, -31.87, 0.00
10, -51.43, 0.00
11, 27.14, 0.00
12, -190.66, 0.00
13, -31.16, 0.00
14, 78.61, 0.00
15, -57.94, 0.00
16, -60.13, 0.00
17, 300.54, 0.00
18, -27.67, 0.00
19, 88.13, 0.00
20, 148.35, 0.00
21, -64.19, 0.00
22, 84.82, 0.00
23, 190.55, 0.00
24, 21.12, 0.00
25, 4.38, 0.00
26, 10.28, 0.00
27, 112.43, 0.00
28, -32.30, 0.00
29, 78.39, 0.00
30, 51.74, 0.00
31, 79.80, 0.00
32, 66.79, 0.00
33, 133.06, 0.00
34, -91.13, 0.00

Comments

  • chipaudettechipaudette
    Hi,

    This is chip. I wrote the code that makes the text output, so I'm responsible for any weirdness. Sorry!

    I looked at the relevant code ("OpenBCI_GUI_Simpler/OutputFile") in the method "writeRawData_txt". The line the writes the sample index is this one:

    <code>output.print(Integer.toString(indexOfLastValue - nsamp + i));</code>

    I can definitely see how it can write negative values...the "- nsamp" would cause the output to be negative if "indexOfLastValue" were not big enough. In my mind, though, indexOfLastValue should always be equal to or larger than nsamp. I'm looking into the main program now...maybe the problem is there...

    Chip
  • chipaudettechipaudette
    Wow, looking back at the main program ("loop()") and at that text-writing function, I had some crazy thinking going on with regards to how the data is buffered and how I keep track of what sampleIndex it's at. I wrote it and it's hard for me to untangle. Wow.

    Sadly, I can't seem untangle it right now. I'll hit it again this evening.

    Chip
  • chipaudettechipaudette
    OK, I figured it out. I think that the Processing code is correct...if the Arduino's software were written as intended...

    The Processing code assumes that the Arduino is keeping track of the number of EEG samples that it has collected. In the transfer of data from the Arduino to the PC, there is a field that I set aside so that the Arduino would tell the PC which sample is currently being sent. The Processing code has been written to use that sample value as the basis for the SampleIndex that gets written to the TXT file.

    Great.

    The problem is that I had modified the Arduino code to NOT transmit its current value of the sample counter. Instead it just sends a value of one. I did this because I had problems with my Processing code every time the Arduino's sample counter incremented to the point where one of its bytes was the same as the header byte or footer byte for the data packet. So, I just avoided the problem by not writing the actual sample counter anymore. Easy.

    Unfortunately, you're now seeing the side-effect...the sampleCounter getting written to the text file is all wrong. You see it go from -29 to zero because the Processing code only writes to the text file every 30 data points. Since the sampleIndex from the Arduino is "1", and since the Processing code think that's the sampleIndex for the latest data point, the Processing code writes -29 to zero. So, yes, there is an "off by one" error here, but the bigger problem is that the Arduino code doesn't send the actual sampleCounter.

    I'll look into the best way of fixing this. It's probably just best to ignore the first column in the text file for now. As far as I can tell, the EEG values themselves should be correct and in the right order. I'll try to confirm this experimentally later.

    Chip
  • mittenchopsmittenchops
    Hey Chip, another brilliant dude at the hackathon (Robert) made a really helpful barebones repo that might be helpful in sorting out the python issue?

    https://github.com/strob/openbci-python
Sign In or Register to comment.

© Vanilla Keystone Theme 2026