Recording microphone / audio signals on OpenBCI for use as a trigger channel

Hello,

I am interested in performing evoked response measurements using OpenBCI (auditory steady state response, p300, etc). I have looked into various methods for doing this, such as:


I’m concerned about jitter in the trigger timing, so I’d prefer not to use the lab streaming layer if I can avoid it. I also I don’t have a WAV Trigger board.

The simplest solution I think would be to wire a microphone directly into one of the channels of the OpenBCI amp, to ensure that both the audio stimuli (e.g. the trigger) and the EEG data will be recorded in sync.

I tested a dry run of this (no EEG connected, only the mic). I connected the mic’s to leads to SRB2 and the bottom (N) pin of N1P, and disabled all other channels except Ch 1. I was able to pick up loud noises, such as if I clapped my hands. But the signal was quite weak. When I played auditory tracks (containing either click trains or pure sinusoidal tones), I could detect nothing. 

A few hypotheses for why this didn’t work:
1) Something was wrong with my set up (e.g., is it okay that I did not have anything connected to bias or AGND, for that matter?)
2) The sampling rate of OpenBCI is too low to pick up the sounds. However, the audio clicks were square waves, so I would think they would have a low frequency component. 
3) The amplitude of the microphone's signal is too low. However, I read elsewhere that mic signals should be on the order of millivolts (https://www.kfs.oeaw.ac.at/manual/3.8/html/userguide/461.htm; “a microphone level signal is more often in the range from 5 to 50 mV (millivolts)”). So I would think that this would be strong enough for OpenBCI to detect.

If anyone has recommendations or alternative suggestions, it’d be great to hear.

As a second attempt, I’m going to buy an audio cable splitter so as to split the audio line out of my laptop (running on battery). I’ll hook one end up to a pair of headphones that I'll wear, and the other end I’ll connect to the OpenBCI as above. I presume that directly capturing the line out will provide a stronger signal than what I had before from my Mic.

Is there any chance this could damage the OpenBCI? (Apparently the audio line out signals are on the order of volts). Does this sound like something that could feasibly work?

Thanks

Comments

  • wjcroftwjcroft Mount Shasta, CA
    Alexi, hi.

    Did you change the gain on the channel you were using? It's possible your clicks were over-driving the microvolts levels expected. See the Cyton tutorial for how the ECG test is setup with a lower gain.

    Yes, the headphones output voltage level is much higher than that of a microphone.

    Also, many 1/8" jack TRS mic's require power,


    Be very careful with driving the Cyton channel with the audio output from your headphone jack. Use the lowest possible setting.


    William

  • wjcroftwjcroft Mount Shasta, CA
    Also see this definition of Line Level, for those equipments that have a line level output jack.


  • Hi William,

    Thanks for your feedback. I actually ran a test driving with the audio jack before seeing your reply. Using this method, I was able to capture a very strong audio signal while simultaneously capturing EEG. My next step is to do the signal processing to convert this audio signal into a trigger.

    I used the Cyton ECG tutorial to use the Ch4 N and P channels to measure the line out audio jack voltage. I removed Ch4 from bias and SRB2, while also lowering its gain to 1x.

    When using the microphone, the amp definitely wasn't saturating. The lack of signal might be because, as you said, the mic was unpowered.

    A few other quick questions I had: 

    1) About this comment: "Be very careful with driving the Cyton channel with the audio output from your headphone jack. Use the lowest possible setting."

    I am quite concerned about damaging my Cyton board by doing this. Do you know what the maximum voltages are that you can apply to the input pins without causing damage?

    2) Since the line out signal is so strong, and the ERP signal in the epochs I will be averaging over is very weak, is it possible that the the audio signal could bleed through into the other channels, and affect the appearance of the averaged ERP in the EEG channels?
  • wjcroftwjcroft Mount Shasta, CA
    (1) http://openbci.com/forum/index.php?p=/discussion/1373/max-voltage-i-can-input-to-a-cyton-channel

    (2) you can use a "voltage divider" to reduce the signal to any desired level, even microvolts. Not sure about cross channel leakage. Could happen more likely if your EEG and trigger cables ran next to each other. Because they couple inductively that way. But if you divide your trigger signal down to say less than 200 microvolts, cross cable coupling would be less likely.


    Regards, William

  • Thanks for the info!
Sign In or Register to comment.