• State Resolved
  • Locked Locked
  • Replies 11 replies
  • Answers 3 answers
  • Subscribers 47 subscribers
  • Views 4118 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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 - preamplifier needed for EEG apps?

Jim Boyd
Prodigy 110 points
Other Parts Discussed in Thread: ADS1299

I’m trying to evaluate the ADS1299 for use in an EEG application using the EVM-PDK. Although EEG signals are in the range of 1-20uV, the particular type of EEG signal of interest in my project is actually sub-1uV, making results even worse.

 

The datasheet gives an equation for ADC resolution vs VREF which implies

LSB = 539nV for VREF=4.5V (internal reference)

LSB = 298nV for VREF=2.5V (external REF5025 2.5V reference)

 

If the max PGA gain is 24 and EEG signals are in the range of 1-20uV then, without a preamplifier, the input of the ADC for EEG signals is 24-480 uV resulting in, according to my calcs, a range of:

44pk to 890 pk for VREF=4.5V (internal reference) ---- approx. 5-10 bits resolution

80pk to 1610 pk for VREF=2.5V (external REF5025 2.5V reference) ---- approx. 6-11 bits resolution

 

 

The first question is why isn’t a larger preamplifier gain included in the front end of an IC advertised as an EEG system solution?

 

The second question is how low can I set VREF to expand the conversion range? The external precision reference ckt in the datasheet shows the REEF5025 which is 2.5V but the Electrical Specs merely state both internal and external reference voltages as 4.5V typical.

 

No external preamplifier is mentioned in the applications docs or datasheet. Is there a recommended preamp device for EEG applications. It would seem that for my application, let’s call it 1uV pk-pk, I would need an overall gain when using a VREF of 2.5V of 2.5V/.000002V = 1.25M!!!! Given the PGA at x24 this still yields an external gain of 52,000.

 

Comments????

 

Jim

  • 0
    TI__Mastermind 38745 points

    Hi Jim,

    Thanks for your interest in the ADS1299! Regarding your first question, we appreciate any feedback from customers about our products. We are constantly looking ways to improve and so your input is truly valuable to us. The limit of the gain on the ADS1299 exists due to a tradeoff that exists between power and performance for all amplifiers.

    In reference to your second question, you can make the external reference voltage as small as you’d like to increase your voltage resolution. There is no minimum on the reference. You’re right; you need a large gain to resolve signals of the amplitude which you are speaking. There is no recommended amplifier to drive the inputs since the PGA will drive the modulator, which is the tough part.  Recommendations are included for the reference driver because the sampling circuit makes that selection much more critical. However, I’d be willing to help you determine the right amplifier if you’d like.

    Regards,

    Brian Pisani

  • 0 in reply to Brian Pisani
    Prodigy 110 points

    Yes, Brian. I would like to connect with you for some direct off-line conversation about properly configuring the ADS1299 for my app. Personally, I find the e2e forum itself as a truly sub-optimal mechanism for customer design support. Beyond the obvious privacy concerns, the time delays and unpredictable nature of the response makes me hesitant about considering TI parts in general.

    I previously posted a question about the undocumented part#s for the external precision reference which is present but unpopulated on the EVM board. I was told there is no BOM info for this and that the parts in the PCB layout were not specific (????). This was complicated by the fact the suggested schematics present in the datasheet and EVM User Guide do not match and do not contain complete part# info. Yikes...

    So I will plan on an external preamp with a gain of 20000 or more (the legacy circuit I'm trying to outperform has a front end gain of 20,000) and probably an external reference voltage of 1.25V (any low noise 1.25V suggestions??).

    Please send me a direct email and I will respond with my full contact info.

    Jim

  • 0 in reply to Jim Boyd
    TI__Mastermind 38745 points
    Hey Jim,

    I sent you an invitation to connect so we can share contact information.

    Regards,
    Brian
  • 0 in reply to Brian Pisani
    Prodigy 45 points
    Hi Brian

    I have a similar situation as the one described above, to make it real short in my application using 2.5 VREF and a max input signal of about 2mV i would require about 200x preamp gain (12 gain setting on the PGA) to get the full 24 bit otherwise i get 16bit tops (wich is not bad at all but is suboptimal)

    If you can make a similar reccomendation on preamps and lower voltage VREF it would be appretiated! We can discuss this offline if it works best.

    Regards
    Bruno
  • 0 in reply to Bruno Rosa
    TI__Mastermind 38745 points
    Hey Bruno,

    What is your target SNR? The gain used and type of amplifier will likely depend on that. In addition, how do you intend to measure the signals. Will they be single ended with respect to some reference electrode or will they be measured fully differentially?

    Regards,
    Brian Pisani
  • 0 in reply to Brian Pisani
    Prodigy 45 points
    Hi Brian

    I mistakingly flagged your reply as spam.. sorry for that. My mouse hanged on me over the 'more' button. Please ignore it.

    As to your questions i have no figure so far for SNR, it should be as good as possible. the 'source' impedance is of a few megaohms and generates differential voltages from 2uV to 2mV . The measurement is made differentially across the source through Ag/AgCl electrodes. I will be measuring two independent sources of the same type.

    Regards
    Bruno
  • 0 in reply to Bruno Rosa
    TI__Mastermind 38745 points
    Hey Bruno,

    Ok that is helpful. Another thing I need to know is your bandwidth. The amount on noise that is contributed by each component can be limited significantly by band limiting your input and taking advantage of the digital filter on the ADS1299. Do you know the highest frequency you will need to convert?

    Regards,
    Brian Pisani
  • 0 in reply to Brian Pisani
    Prodigy 45 points

    Hello Brian

    The Bandwidth should be limited (single pole bandpass) with Fclow = 0.15Hz FcHigh=34Hz i remember seeing a similar application (without ADC). that used an LMC6084, there was not much detail but i guess a differential setup similar to the one on page 6 of datasheet was used.

    I still have some doubts on wether to preamplify or to just feed the signal straight into the ADS1299 and just use the digital filter and decimator, i would loose some resolution but in the end the circuit would be less complex, maibe with a better noise performance feel free to throw in your opinion on this...

    thanks for the help!

    regards 

    Bruno

  • 0 in reply to Bruno Rosa
    TI__Mastermind 38745 points
    Hey Bruno,

    You're correct; adding another component will increase the amount of noise present in the output. Given the limited bandwidth of your input, I'd recommend using the lowest output data rate on the ADS1299 of 250 SPS. The cutoff frequency for the digital filter scales with the data rate so your noise performance will be the best at the lowest data rate.

    For that data rate, input referred ADC noise is on the order of less than a microvolt. Let's say you round up and the noise is equal to 1 uV. Assuming the Johnson noise of other components is negligible (for your limited bandwidth it likely would be), your 2 mVp signal would appear 66 dB above the noise floor; plenty of breathing room in my opinion.

    The noise actually looks better with a higher PGA gain. If you plan to AC couple the inputs, I'd say gain it up using the PGA setting of 24. If your inputs are not AC coupled, determine the maximum amount of electrode offset and see how high you can take the PGA gain without clipping the reference.

    For perspective, many ECG systems measure signals in the singe-digit millivolt range. I've personally been able to measure them perfectly fine employing only the on-chip PGA as a gain stage. In my opinion, there is no need for an extra gain stage and the internal reference on this part will be sufficient.

    Regards,
    Brian Pisani
  • 0 in reply to Brian Pisani
    Prodigy 45 points

    hello again Brian

    Yes those were my thoughts exactly.  So you think there is no resolution benefit on using a 2.5V external reference? One other question, would it be possible to exchange contacts in order to continue these discussions over email? I'm not sure if it's your policy to work that way but it would be more convenient for me. 

    I haven't still got down to layout the circuit itself, but I am pretty sure issues will come up on all kinds of details and this forum thread would not be the proper place to pour down my tears.

    Thanks again

    Regards 

    Bruno

  • 0 in reply to Bruno Rosa
    TI__Mastermind 38745 points
    Hey Bruno,

    There is a resolution benefit to a smaller reference, but at this point I don't see it as being necessary and I can think of a few reasons not to use an external reference. I will request to be friends with you on the forum with my contact information.

    Regards,
    Brian Pisani