Hi,
I've gone though all the posts that mention AD1298 and EEG, however I have a few unanswered questions.. I apologize if some of these are basic, but any input would be appreciated.
In looking at table 1, with a PGA gain of 12 and a sampling rate of 500 (more than sufficient for EEG), the noise is listed as .5uVrrms / 2.9uVpp. There are many posts that discuss the need for a preamp - but these figures seem reasonable. From my understanding, the EEG signal is about 10-100uV at the scalp. Is this listed figure still too high? Are there other noise sources that I'm missing?
I'm also a bit confused about the reference/RLD implementation. In a number of posts, people discussed how to generate a RLD signal from a differential pair. However, this means I would generate a RLD for each sensing electrode. Would I then average all the RLD outputs (akin to a Wilson central terminal) and use that?
In looking at figure 23, it looks as if the output of the PGAs can be configured to sum-up (averaged) and output on the RLD_INV pin. If this is passed through an external inverting amplifier (one post mentioned a 7uV noise floor for that amp which is too high), why couldn't this be used as the RLD signal, even with the preamps in the signal path?
The EEG "helmet" I am trying to interface has a GND and a REF electrode. I would guess that the RLD output connects to the GND terminal, and the REF electrode is connected to the negative terminal of all the pre-amp IAs? And the IA output would connect to the the positive inputs to the AD1298, and the negative terminal would be grounded. Would this be the ideal configuration to use?
Finally, given how many posts discuss the need for a preamp, has anyone found a preamp (by any manufacturer) that supports more than one channel per IC? The main advantage of the AD1298 is the small form-factor. Adding 16 additional IAs along with passives would negate this benefit.
Thanks so much,Reza Naima
Hi Reza,
We currently do have customer's doing EEG with the ADS1298 using a gain of 12 and sample rate of 500SPS. For some applications, the noise level is fine and they can get what they want. For others, they need additional gain at the front end to be able to see more details of the EEG signal. Keep in mind that the noise numbers are input referred meaning that 3uV peak to peak in a gain of 12 corresponds to about 36uV once gained up by 12. We have a few INA parts we have recommended for a discrete solution for customer's who need higher gains. If interested, I can pass the part numbers along.
In basic ECG applications, customer's use the averaged point from the inputs as the RLD signal to set the common mode bias point for the patient. You can give this app note a read talking about the benefit of using the RLD in this configuration -
http://www.ti.com/lit/an/sbaa188/sbaa188.pdf
I am not familiar with the EEG helmet and cannot really comment on configuration. I can only recommend and comment on using the chip with an ECG/EEG simulator.
Regards,
Tony Calabria
Tony,
Thanks so much for the prompt reply. I would love to see the recommended part numbers. What do you think of the INA333, with it's integrated RFI/EMI filter? And I take it you still have no multiple IN amps in a single package? I want to apologize again if some of these questions are basic, but I want to make sure I understand how to best implement the ADS1298 in an EEG application and to avoid making any stupid mistakes. (Would love to see that EEG app note be released that I've heard so much about)
As to the other question, I was essentially wondering if the typical implementation is to use a single reference electrode as the negative input to all the channels? And in the case of using a preamplifier, if it is sufficient to configure the ADS1298 to use a non-differential input (assuming a non-differential output of the preamplifier).
I'm not sure what the 'averaged point from the inputs' is on the schematic. Can you point out a pin or component on the eval board that you are referring to?
In the case of using preamplifiers, the other question is if a single RLD is generated from each of the preamplifiers (as in http://openeeg.sourceforge.net/doc/modeeg/modEEGamp-v1.0.png which someone posted) and averaged, or if the RDL_INV output can be used even with preamplifiers? I've looked at the sbaa188.pdf document, but it doesn't seem to answer this question.
And, this is perhaps a bit obvious, but I want to make sure: the RLD connects to a discrete electrode on the scalp.
According to sbaa188, connecting Vref (which I'm assuming is (VREFP-VREFN)/2) through a resistor helps with the CMR as well. Have people implemented this strategy successfully?
I suppose the simplest thing to do is hook up the eval board to the EEG electrodes and see if sufficient detail is visible which I'll try shortly.
Thanks again,
Reza
1 - INA114, INA115, INA131, and INA333 are low noise devices we have recommended in the past to customers. All of which are single amps per package.
2 - Your second question depends on what you want to measure. If the measurement is a differential measurement, then you would use two electrodes to record the voltage that is read back. If you are interested in a single ended measurement, for example a chest lead for ECG, the WCT is used to route to the negative terminal as a reference point. The reference point (WCT) for the ADS1298 can be routed internally using the ADC's internal mux unless you are in an application where you would need the preamplifier (INAs), requiring you to route pre-amp.
3 - You will want to use the RLD_INV and close the loop around the internal RLD amplifier to isolate the internal averaging from the RLD. Otherwise, you will create a voltage divider between the series resistances from the averaged inputs and the load on the RLD.
4 - It is hard for me to comment on the performance when hooking it up to a patient, you will have to ask the other innovators on the forum. All of our testing has been done with patient simulators and we are not legally allowed to comment on and help with designing using a human.
Are you with a company working on an EEG solution? What is the end product you are developing?
Once again, thanks for the reply. I looked over the amps you recommended, and the best fit would be the INA333 for it's package size and built-in RFI filter, however would really prefer using a multiple-amp per package device.
Has anyone looked at the intersil ISL28470? It's the only one that I can find with 4 amps in a package. It's noise is slightly higher (60nV/rt(Hz)) vs the INA333 (50nV/rt(Hz)), but it looks OK otherwise. Any other people on this forum look at it?
With regards to my second question, I suppose there are two ways to implement the system. One would be to use a single electrode as a negative terminal reference for all the preamplifiers, or to do something similar to the WCT and average the signal from all the electrodes through a set of resistors and feed that to the negative terminal of the preamplifiers. I was wondering if one method was preferred to another. Either way, I would imagine that the input to the AD1298 would be single-ended after the preamplifiers, so the negative terminals of the inputs to the AD1298 would be Vref (mid-rail).
As for the RLD, I'm a bit confused by your reply. My question is if I can use the signal form RLD_INV to generate the RLD signal using a low-noise amplifier, or if I would need to do so using the preamplifiers.
I'm working at a startup, and this project is to make a wifi-enabled small (ambulatory) EEG device to monitor people while they are active.
The more I've been looking at the design, the less the ADS1298 looks like it would be a good fit for EEG applications - the bulk of the benefits it provides are for ECG functionality. I've started looking at using a high gain amp + your 16-channel 24-bit ADC. The INA826 looks quite promising at high gains (~.5uV of RTI noise). Have you recommended that amp to your customers for EEG applications?
Thanks,