Other Parts Discussed in Thread: ADS1299, OPA2188, INA118
Hello,
being in the conceptual and design phase of an EEG device using the ADS1299, I scanned through (almost) all posts in this forum regarding the ADS1299 for EEG measurements to find information on pre amplification and filtering of the signal.
(thread with highest match so far: http://e2e.ti.com/support/data_converters/precision_data_converters/f/73/t/267314.aspx )
To state the problem briefly:
I found many reports on successful design of (private/non-professional) EEG hardware using the ADS1299. I understand, that with 1µV pp noise RTI and 24Bit ADC Sigma-Delta resolution, the surface (scalp) EEG signal can "in most cases" (as I read several times now) be acquired without any gain higher than the internal PGA gain G=1-24.
However, I would like to use the ADS1299 for a design that is as close to a "professional" / high quality commercial EEG as possible: Thus, I am looking for an approach with low-noise pre-amplification/filtering to further enhance SNR/ the resolution of the actual EEG Signal in the 0-4.5V span of the ADC.
a) Do you agree with my opinion, that this approach can yield an even higher performance using the ADS1299? If not: Has there been an approach to evaluate the raw ADS1299 EEG signal quality in comparison to high-performance EEG solutions that can support this?
b) Looking for low noise instrumentation amplifiers for the pre-amplification step, I found TI products INA118, OPA2188 and the competitor's Analog Devices AD8222. However, it is not totally clear to me, how a pre-amplification block would be designed for best results with ADS1299: How would you suggest to design the differential input stage? If the input stage consists of differential instrumentation amplification (using one of the ICs above), how to feed the differential inputs of the ADS1299?
If the input stage consists of single non inverting high precision opamps, don't we introduce several sources deteriorating the CMRR performance of the ADS1299 (resulting e.g. from resistor and gain tolerances) in each (+ and -) input channels?
c) Even without pre-amplification, for anti-alias filtering the above problem is again true - is it not?
If I introduce a (passive, yet better active, e.g. 3rd order) bandpass to reduce the bandwidth of the signal (and thus also the total noise power), again - the CMRR performance might suffer! In a conservative EEG amplifier approach, the filtering unit would be placed behind the instrumentation amplification (and common mode reduction): Obviously, this is not possible using the ADS1299.
d) As soon as any active pre-conditioning is introduced, the ADS1299 lead-off detection/impedance measurement features are not applicable anymore?
To summarize my questions: The ADS1299 is an Analog Front End. Does this truly mean, that additional analog signal pre-conditioning is not necessary or would even be unwise for the above reasons (also excluding Anti-Alias Filtering and pre-Amplification)?
If yes: just add patient detection circuits - and that's it?
If signal conditioning can further improve the EEG signal, how would you recommend to introduce the components for best interaction with the ADS1299?
Thank you very much for your time and reply.
Alexander