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.

ADS1298 AC Lead Off Detection for EEG

Other Parts Discussed in Thread: ADS1298

Hi Ti,

I'm currently researching wether the ADS1298 is suitable for our EEG application. An important requirement is that we have constant information about the connection of the electrodes, without having to stop our measurements. If I understand correctly, the ADS1298 can do this, though only with AC lead off detection, not DC. Am I right?

I believe using this method, the ads1298 injects a high frequency signal into the positive or negative electrode of a channel. Then by converting the incoming signal, the output of this signal can be analyzed and the state of the electrodes can be determined. This must be done however externally. Therefor, I have several questions.

1. In our EEG measurements, 18 signals are measured in comparison to 1 central electrode. Thus, to be efficient, I thought I would just connect this last electrode externally to each of the inamps of the other channels. Another possibility would be to use this 1 electrode as the input for my RLD signal and then reroute this RLD signal internally to all other channels. I'm now wondering how this could be combined with the lead off detection.

In the last case, I have got no idea how lead off detection could be implemented. Can I do this by measuring the RLD signal and injecting a high frequency signal on each of the other inputs?
In the first case, will I be able to get the information I need, just by putting a high frequency signal on each of the other inputs and measuring the first input?

Another thing to take into consideration, is that I will need 3 ADS1298's to have enough inputs for EEG. This might give problems with the RLD solution.

2. Maybe this issue could be resolved by using DC lead off detection, if it doesnt take to much time to process it. I need 250 SPS, can I do lead off detection with DC between two of these samples, without possible capacitive or inductive distortion of the data? But then again, I don't really see how i should connect my inputs, since DC mode uses the differential electrodes of each channel, and I plan not to use all of these electrodes.

3. If part 2 is possible, do I lose the ability to use RDATAC mode, to continously run my conversion? Actually I'm not really sure what the difference is between using continuous mode and just iteratively using pulsed mode. It seems to me in both cases you still need to send a START signal, only in continuous mode, it's not possible to send opcodes.

I realise I'm probably asking alot here, but I hope you can help me out a bit. Thanks in advance!

Pieter Willemen
Antwerp, Belgium

 

  • Hi Pieter,

    Your assumption is correct, the ADS1298 can distinguish 'lead off' via AC methods, but an electrode being off can only be sensed through DC means.  Let me look into this for you a little and see if we can come up with a solution for you.

  • Hi Tom,

    I was just wondering if you were still going to give a reply here? I don't want to be pushy or anything: If you have other work to do or something, I completely understand. I can only hope you can find some time and maybe give me some advice here.

    Pieter Willemen,
    Antwerp, Belgium

  • Hi Pieter,

    My apologies for the delay - we have an application note underway for the various lead off scenarios you might encounter and that should help answer your questions. 

  • Hi Pieter,

    Depending on what you are trying to achieve with the lead-off detection, you may be able to get by using DC lead-off. If what you care about is to find out if the electrode is connected or not connected DC lead-off will work. If you care about a little more detail and possibly what the impedance is of the electrode, etc. AC lead off may be what you will want to do.

    The way that the DC lead off detect works is that it using an internal comparator to test to see that when a small internal DC current is enabled (a few nA) it reports back a DC voltage created from the body's impedance within the threshold limitation of the comparator. So, the way to think of it is more of a electrode lead off detection to test to see if the electrode is connected or not. This happens independently of the differential voltage that the ADS1298 uses for conversion. One more important aspect about DC lead off is that it can be enabled and constantly running in the background. If you take a look at page 32, you will notice that in the STAT byte (really it is 24 bits so 3 bytes) which appears prior to the 8 converted channel results contains individual bits showing the lead off status for each electrode. Reading these 24 bits back, you will be able to know if the lead is connected or not connected without having to stop your conversion process to run a lead off detection. If you are using AC lead off detect, you cannot use this 'on the fly' lead off detect method and, instead, would have to stop the conversion to and perform an AC lead off separately.

    So, assuming that the inputs are not AC coupled and you are just interested in knowing if the electrode is connected or not connected, I think that DC lead off should work for you.

    Please let me know if I did not answer any of your questions and I will try to help where I can.

    Regards,

    Tony Calabria

  • Hi Tony,

    thanks for your reply. I still have some questions though.

    First of all, I got a bit confused. You said I can continuously run conversions using DC detection, but I need to stop them using AC detection. I actually had the axact opposite in my mind the whole time. I thought this actually made sense, because using AC detection, you simply convert the voltage and digitally filter out the out of band signal afterwards. Why do I need to stop conversions using AC detection? How can DC detection work, without stopping converting? On page 63, it says that AC detection "can be accomplished simultaneously with the ECG signal acquisition".

    Secondly, in a reply above, Tom said that "an electrode being off, can only be detected through DC means". Why is this the case? If the electrode is off, the sensed impedance will simply be very high. Which is something I think should easily be detectable, even with AC detection. Am I wrong?

    Could you give an example of how a setup of electrodes would look like? For example IN1P and IN1N connected to the patient using DC current source mode detection. Because I am planning to use RLD or WCT, in which case not all inputs of the inamps are connected to the patients body. I don't understand how this lead-off detection should then be implemented. I have said this in more detail in my first post.

    Thank you.

    Pieter Willemen
    Antwerp, Belgium

  • Hi Pieter,

    First off, DC lead off allows you to compare each individual input channel to a threshold voltage using an internal comparator that runs in parallel with the PGA/ADC. We include a block diagram in Figure 59 which shows how these are running in parallel. The individual comparators for each channels use the result from a 4 bit DAC which will set the DC lead off threshold limit. Therefore, when the electrode is off, the sensed impedance may be very high leading to a high voltage when a current is excited and would push it beyond the threshold limitation which set by the comparator to report a lead is off. Using this DC lead off method, the converter will know without any post filtering if an electrode is off, and if it is, which electrode is off. This will be reported back in the STAT bits as a conversion result is read back. DC lead off works without stopping conversions as the comparator runs in parallel and updates the lead off status bits prior to the conversion result. The only downside of using DC lead off is that it will turn on the small current source, in order to measure to see if the electrode is connected, and leave it on during conversions which may result in a slight increase in offset.

    Now, with AC lead off, you could also use this method while running continuous conversions and use some post filtering to read back the AC lead off frequency (fDR/4). However, this will take some more work on your end requiring you to create a band pass filter to pull out the AC lead off detect frequency. Customer's usually prefer to use DC lead off instead of AC lead off as AC lead off will require you to look at the results from multiple leads to detect which electrode is off. For example, if you if you are performing AC lead off across the LA and RA on a human body for ECG and the signal is "missing" post filtering you will not know exactly which electrode is off. You will know that the lead is not correct and that one of the electrodes (either RA or LA) is missing. You will need to look at an adjacent channel (either LL and RA or LL and LA) to see if the signal is missing to narrow down exactly which electrode is off.

    In theory, you are able to use either AC lead off or DC lead off in real time. However, if you decide on using AC lead off, you will need to use some post filtering to determine if the signal is there and the lead is connected. A lot of ECG customers stay away from AC lead off because the AC lead off frequency signal bandwidth meshes with the bandwidth of the PACE signal from a pacemaker making it difficult to distinguish if a pacemaker is present. Additionally, if  respiration is used in the ADS1x9xR devices, the modulated signal that is output could be within the same frequency bandwidth as the AC lead off signal (depending on the DR selected). 

    Regards,

    Tony Calabria

  • Hi Tony,

    this really helped me alot! Thank you.

    But there's one thing that remains a mystery to me. I'm probably missing something silly, I guess:

    I understand how the lead-off detection would work if you attach electrodes to both inputs of one of the inamps. Then it would look like this:

     

    What I don't understand, is how it would work, when you would use a WCT or RLD. Because then it would like this:

     

    Do you see my question? Where is the current going?!

    I know it has to be working, because the EVM is using WCT and it is also doing DC lead-off detection. I just don't see how?

    Thanks for your reply's!

    Pieter Willemen
    Antwerp, Belgium

  • Hi Pieter,

    The more questions that you ask, the better as I am in the process of writing a lead off detect app note explaining how it works. Take a look at this regarding the lead off with the RLD - 

    http://e2e.ti.com/support/data_converters/precision_data_converters/f/73/p/124924/447760.aspx#447760

    I am not sure how you are doing WCT lead off detection. The WCT output is routed back around into the input of the ADC.Therefore, there is no reason to do lead off detection with the WCT.

    If in the pictures above you are trying to show the amplifier as the internal PGA of the ADC, then with DC lead off detection, the only current source that is turned on is for the (+) input. You do not turn on the lead off LOFF bit for the (-) terminal. The current from the (+) input is sunk through the body to create a voltage which is compared to a threshold using a comparator running parallel to the internal PGA.

    Regards,

    Tony

  • Hello again,

    it's nice to know you're working on extra info. You have again told me alot and I learned something new, but my original question remains actually, sorry ;).

    I asked where the current was going in the picture. The amplifiers indeed represent the PGA's. I'll try to rephrase my question :). Let's say I have the following setup.(This is an extract from figure 59 on page 63.)

    Now I want to know whether electorde IN-P is connected or not. Using DC lead-off detection, the datasheet says a current will flow through the electrode, through the patients body, coming from AVDD. But here's my question then. Current can only flow in a closed loop. Where does the current go, once it has reached the patients body?! [QUESTION 1]

    You said: "The current from the (+) input is sunk through the body." But where to?

    Here's another question. In my understanding the detection works by measuring the voltage drop on IN*P, caused by the current going trough the resistor. I don't see how this can be done without interfering with the measurement of the actual input signal. If you put a current of 6nA through those 150kOhms, you get an extra 9mV on the positive input of the PGA. And if you only do lead-off detection on the positive electrode, and not on the negative electrode, these 9mv are not neutralised, and the AD convertor measures an extra Voltage [9mV multiplied with the gain]. At least, that's the way I see it. [QUESTION 2]

    Maybe it happens between two measurements or something? So the AD never actually sees the extra voltage? [QUESTION 3]

    About the WCT, I wasn't interested in lead-off measurement of the wct signal. I wanted to measure lead-off for the positive electrode, while routing the WCT signal to the negative input of the PGA. Because this is what my setup is going to look like. Like this:

    I first thought, the current would come out of AVDD, go through the positive electrode, onto the patient, and then come back through the negative electrode and into AVSS. That's why I didn't understand, how it would work while using WCT, because then there wouldn't be a negative electrode, to allow the current to come back. So how does it work then? [QUESTION 4]

    I think without understanding this, it's not possible to choose how you're going to set up your inputs and how your going to route WCT and/or RLD. And without this setup, you can't move on with your design.

    I'm looking forward to some of your answers and also the app note.

    Pieter Willemen
    Antwerp, Belgium

  • Hi Pieter,

    Thank you for the description and diagram, this has been very helpful to me. I now understand the confusion and what aspects of lead off detect need to be better explained -

    You are right that there needs to be a path to the ground of the chip in order to complete the loop for lead off detect. For that reason, you need to have the RLD connected in order to both set the DC operating point for the body and provide a route to ground for the current to flow. The RLD line has a driving amplifier which is capable of sinking the current to ground.

    You should notice that if you try to use DC lead off and disconnect the RLD electrode, the lead off results may begin to read back incorrectly. In this case, depending on the configuration of the electrodes and chip, the currents may try and find their way back to 'complete the loop' through leakage paths at the individual PGA inputs and capacitive leakage paths throughout. 

    The detection method works as you have explained where you are sending a small current across a resistance to create a voltage. With the 6nA current through a 150k impedance you will get an extra 9mV offset error. This offset error is very minor when you consider the fact that the offset error of a typical electrode could be somewhere around 300mV. 

    Regards,

    Tony Calabria

  • Hi,

    you said the more questions I ask, the better. So here it goes :)

    For the positive electrode, the current for DC lead-off comes from AVDD and goes to the RLD. Why is this inverted at the negative input? Do you then just measure a negative voltage? Why can't it just be the same way?

    Regards.

    Pieter Willemen
    Antwerp, Belgium

  • Hi Pieter,

    With Lead 1, 2, or 3, which derive the Einthoven Triangle, the DC lead off is designed to source the current from the positive electrode (+) and sink the current through the negative electrode (-). You will notice that their is a LOFF_FLIP bit which allows you to switch the source and sink current between the positive and negative ADC inputs. The chest leads which refer the negative leads to the WCT terminal do not have this clear path to ground and are forced to use the RLD electrode as the return path for the current. In theory, it could have been designed to have all the inputs just source current for lead off detection and rely on the RLD amplifier to sink all the current sources.

    Regards,

    Tony Calabria