This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ADS1299: Question about data collected on EEG electrodes

Part Number: ADS1299

Hello I have a few problems with data collected on eeg electrodes.

I took over the project from another student and I don't understand the signals I get.

The low-pass filters for the eeg electrodes were set with a cutoff frequency of 56.98Hz. And I set the frequency of the bias filter to 100Hz.

I get the following signals:

Maybe there is a high frequency noise.

Does anyone have any insight to suppress or at least attenuate this noise?

  • Hi,

    Do you happen to have/use a ADS1299 evaluation kit/board(EVM) to test?

    May I ask what are the signal sources you are injecting? And, what are the frequency band of the signal you are interested in?

    What was the original bias filter or RC values? Did you check/compare before you change it?

    The bias filter is not just a filter; it's actually an integrator circuit sum and integrate the common mode noise and then drive back to human body to suppress the common mode noises. I suggest you take a look of R,C values in the datasheet and schematic in EVM user guide to start with. e.g. data sheet 9.3.2.4 ECG and EEG Specific Features  and/or evm schematic.

    Did you try compare the acquired signals between the different cutoff frequencies, e.g. 56.98Hz and 100Hz? And, why do you want to increase to 100 Hz, which causes the 50/60Hz Main noises interfere with the signals.

    By the way, electrodes materials, patches materials and cables/wires are also very critical/important that people need to try/trials&tests&V&V.  We see many customers getting signals improved by trying many different combination of electrodes materials, patches materials and cables/wires.

    You will also need to do some FFT analysis to understand the noise frequencies and sources; if it's 50/60Hz, please take a look at this thread to improve -

    https://e2e.ti.com/support/data-converters-group/data-converters/f/data-converters-forum/842529/faq-does-ti-have-an-application-note-that-covers-how-to-improve-common-mode-rejection-remove-50-60hz-noise-using-the-right-leg-drive-amplifier-rld/3115361#3115361

    Thanks

  • Hello, I don't have a ADS1299 EVM.

    The signals are EEG signals from electrodes I put on my forehead. The frequency band I'm interested in is [0 ; 45] Hz.

    The original Bias filter has a cutoff frequency of 10Hz. I compared before/after it seems the noise dropped.

    The 56.98Hz is the cutoff frequency for the low pass filter on the electrodes that are collecting eeg signals.

    The 100Hz is the cutoff frequency for the Bias.

    This is the FFT analysis of the signals in my previous post. As you can see there is a pike at 50Hz and then again at 100Hz.

    I think the parameters I used for the ads1299 are properly set.

    The data rate is set at 250SPS.

    The channels 1 to 4 are used for measuring the eeg signals with gain of 24.

    The channels 5 to 8 are disabled.

    BIASN and BIASP are at 0xF.

    I generated a PWM signal at 5Hz and 400mV in amplitude that I connected to an electrode.

    When I trace the data collected, I get the same signal.

  • Hi,

    Appreciate for the analysis and feedback.

    It seems you are using Figure 73. Example Schematic Using the ADS1299 in an EEG Data Acquisition Application, "Referential Montage", correct?

    Where do you place the Bias electrode?

    May I ask why you set "daisy_EN" daisy to chain mode on? How is DAISY_IN pin connected?

    Yes, it's good to turn off all the lead off detection features/functions for now.

    BIAS_SENSP and BIAS_SENSN are the combinations that you will need to play around/adjust to find the appropriate combinations/configurations for your designs and applications whose Signals integrity and performances heavily depend on, but not limited to the following sources and factors -
    environment noises, EMI, electrodes, cables, gels, skin conditions, external circuits, board designs, RC/integrator circuit tuning, experiment setup,
    body, breathing, air in lung, walking/talking, skin conditions, environmental surroundings ,etc. A product design depends on many factors, variation, electrodes, subjects' skin conditions, mechanical, enclosures, materials, surroundings and environments. Any of these could affect how you adjust/set the BIAS_SENSP and BIAS_SENSN that you need to do tests&trials&V&V.

    Based on "This is the FFT analysis of the signals in my previous post. As you can see there is a pike at 50Hz and then again at 100Hz.", the noises are from the Main 50/60 Hz power line, environment and their harmonic.

    Can you try internal short test and acquire the signals? i.e. datasheet CHnSET: Individual Channel Settings (n = 1 to 8) (address = 05h to 0Ch) (reset = 61h) bits 2:0 MUXn[2:0] set to 001 : Input shorted (for offset or noise measurements) and observe whether the 50/60 Hz mains are there?

    "I generated a PWM signal at 5Hz and 400mV in amplitude that I connected to an electrode." 

    400mV is a strong signal that hides/makes the 50/60 Hz noises less obvious. What are the typical EEG voltage signal range? Can you generate something in that voltage range? Any EEG simulator?  If EEG band and magnitude are similar or overlap with ECG, could you try a ECG simulator?

    Can you also try 9.3.1.1.2 Test Signals (TestP and TestN) Setting CHnSET[2:0] = 101 provides internally-generated test signals for use in sub-system verification at powerup. This functionality allows the device internal signal chain to be tested out. Test signals are controlled through register settings (see the CONFIG2: Configuration Register 2 subsection in the Register Maps section for details). TEST_AMP controls the signal amplitude and TEST_FREQ controls switching at the required frequency.

    How is/are your system powered and also computer/laptop? Cables/wires lengths?

    There are multiple ways/approaches/methods to reduce the 50/60Hz interference -

    1. See if you can use battery to power up your system and PC/laptop(i.e. remove wall outlet adapter) and observe/compare whether the results improve?

    2. shorten and shield any wires and/or cables.

    3. you may need to play/adjust different combinations of the BIAS_SENSP and BIAS_SENSN

    4. consider setting the RC integrator back to some values close to what is shown in the datasheet or the EVM schematics.

    Thanks 

  • Hi,

    Indeed I use the referential montage. The bias electrode is place behind one ear and the reference electrode behind the other.

    Since I took over I never changed the daisy chain mode parameter.

    I'm going play with BIAS_SENSP and BIAS_SENSN to find which combination is better suited for my project.

    I haven't yet measured the internal short test, but I will post the results when I have them.

    Typical EEG signal range is only few tens of microvolts. I'll try to generate a signal with that amplitude. I haven't got an EEG or ECG simulator but I can generate the signal.

    I already tried 9.3.1.1.2 Test Signals here's the result:

    My system is powered by laptop, when I collect data I make sure the laptop charger is not connected.

    The system is linked to the laptop via usb cable. The electrode cable length is 30 cm.

    I'll play around with the RC values shown in the EVM schematic.

    Thank you for input. 

  • Hi,

    If the internal test signals look good, it seems is,  then, most likely the interferences/noises are from external, PCBAs, components, cables, electrodes, environments, etc.

    some more comments -

    USB cable - see if you can find&use USB cable with ferrite bead and usb isolator

    Electrode cables/wires - try longer(e.g. 1 m) and shorter(15 cm) to see/observe how the results/outcomes change

    yes, take a look ADS1299 EVM user guide and schematic, there are analog LPF along the electrode path.

    Also, the evm user guide section 7 EEG Specific Features

    And, take a look of data sheet section 10.2 Typical Application and 10.2.3 Application Curves and Figure 75.  "... . Given this measurement setup was a single-ended configuration without shielding, the measurement setup was subject to significant mains interference. A digital low-pass filter was applied to remove the interference."

    Thanks

  • Hi,
    Since I did not hear back from you, I believe my suggestions answered your questions.
    I will close this post and if you have any pending questions, feel free to post them here or open a new thread.
    Thanks and have a great day!