Double time stamps in raw data

katuto03

Hi, I'm using Ganglion board for a project and I realized that all sample index are duplicated in raw data and in brainflow csv, they have the same timestamp but different values, why is that?

I also noticed in raw data sometimes sample index doesn't include 0 and 99 or they only have one index 0. Also sometimes brainflow 2** index headers are repeated (image below). Are this normal recording issues or should I be worried?

On the other hand, I was comparing raw data with Brainflow csv data and they doesn't seem to be the same, they even have different timestamp values. Is the data stored differently, recorded separately or they have some transformation in between? I attached both raw and brainflow data from the same recording as example (I converted brainflow csv to txt because it was not supported here in the forum).
Sometimes I also found that recordings where only saved as raw data and not in Brainflow csv format, what is the trigger or condition tha GUI has to save the data in brainflow format also?

Finally I wanted to ask you how can I just record one channel values keeping all the other channels values at 0 in recordings, because even if I'm just using first channel, the other 3 keep getting some data values out of nowhere.

Thanks in advance and sorry for the number of questions in just one discussion but they are slightly all related.

wjcroft

The duplicate timestamps on pairs of sample packets, is because the Ganglion radio protocol packs TWO samples into each packet.

https://docs.openbci.com/Ganglion/GanglionDataFormat/

While you are using the GUI, you can easily switch off unused channels by 'toggling' the channel number button at the start of each channel trace in the Time Series window.

re: "...data out of nowhere..." Since EEG is microvolt level sensitive, room environmental EMF noise can be picked up on unwired channels.

katuto03

So in case I want to take this data to python for analysis and plotting I should read them as separated like 0-0 are going to be sample 0 and 1 with different timestamps (if sample rate is 200Hz it should be 1/200s difference between them right?). And what about the different values in raw data and brainflow csv?

wjcroft

Please just ignore the 'packet number' column. The samples are always sequential in the file. Also realize that the time stamps are NOT assigned inside the Ganglion device, but only when the radio packets are received by Brainflow library in the laptop. So can be various delay factors due to buffering, processor load, other simultaneous processes demanding CPU, usb buffering delays, etc.

Sample rate is 200 Hz, so nominally 5 milliseconds separate samples.

GUI raw recording is using Brainflow behind the scenes. So whatever differences you see are not important.

https://docs.openbci.com/Software/OpenBCISoftware/GUIDocs/#ganglion

Richard @retiutut, do you have any comments on his statement:

@katuto03 said:
...
Sometimes I also found that recordings were only saved as raw data and not in Brainflow csv format, what is the trigger or condition tha GUI has to save the data in brainflow format also?

retiutut

We attempt to save to both OpenBCI CSV (or BDF+) and BrainFlow CSV. This is set in the control panel currently.