• State Resolved
  • Locked Locked
  • Replies 23 replies
  • Subscribers 47 subscribers
  • Views 2723 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

ADS1192 - noise in case of short with IN1P, IN1N and RLD

Hirotaka Matsumoto
Guru 27525 points
Other Parts Discussed in Thread: ADS1192

Hi all,

would you mind if we ask ADS1192?

<Question>
Please refer to the file "20150418_AFE-Circuit.pdf".
20150418_AFE-Circuit.pdf

In case of CN1-1pin(IN1P) and CN1-2pin(IN1N): No noise
In case of CN1-1pin(IN1P) , CN1-2pin(IN1N), CN1-3(RLD): It occurs noise ±60uV
Noise cycle was 350ms, ±60uV.

If you have some advice, could you let us know?

Kind regards,
 

  • 0
    Prodigy 230 points
    From the circuits. you should check the 'COM' (maybe I think it is the brief of "COMMON") signal.
    1. Who creates 'COM' ?
    2. Do you make sure that the COM signal(or reference) is noiseless?
    3. Isn't 'COM' floated from Analog Grount?
  • 0 in reply to Jang Ho Cho
    Guru 27525 points

    Jang san,

    Thank you for your reply.

    At the first, we mentioned "COM" as following;
    "COM" connects to RLDOUT pin and RLDIN/RLDREF(via R32).

    <Reply about your question>
    1. What does it mean "Who creates "COM""?
        COM connects to RLDOUT pin and RLDIN/RLDREF.

    2. Our customer tries to measure the only "COM" without no connecting(without short to any inputs). 
        After that we can judge whether internal voltage causes noise or not. 

    3. "COM" is not connected any pins. it means that COM floats.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Hirotaka Matsumoto
    TI__Mastermind 38745 points
    Hello Matsumoto-san,

    The RLD amplifier is very configurable. Is your customer using the RLD as a common-mode cancelling amplifier or is it simply buffering the chip analog mid-supply voltage?

    Regards,
    Brian Pisani
  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your prompt reply!

    We will have meeting with our customer tomorrow.
    After that, we will feed back you some information.

    By the way, we would like to confirm followings;

    1. Does "common-mode cancelling" mean common mode rejection(noise cancle)?
    2. Does "analog mid-supply voltage" mean (AVDD + AVSS)/2, doesn't it?
    The RLD amplifier is very configurable, so I don't understand all of function completely yet.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Hirotaka Matsumoto
    Guru 27525 points

    Brian san,

    We got some setting of resister and additional information.
    We attached it.(We seem it as common mode rejection(right leg rejection).)

    Today, we confirmed that our customer doesn't use RLD function.
    In case of no use, should PDB_RLD at Adress06h(RLD_SENS) be disabled?
    By disabled PDB_RLD=0, will the noise be decreased?
    (And then, the power sorce is Coin type lithium battery(2.7V to 3.XV). So, it is used Back and Boost converter to make 3V for AVDD.)

    20150421_ADS1192_RLD.pdf

    We always appreciate your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Hirotaka Matsumoto
    TI__Mastermind 38745 points
    Hello Matsumoto-san,

    The RLD amplifier can be configured in a few ways and it's actually helpful to look at the diagram you attached. The goal of the amplifier is to supply a stable voltage to the patient so that the common mode of the inputs does not drift away from the valid range for the device and clip the converter. One way to do this would just be to provide a buffered common mode voltage to the patient by making RLDREF_INT = 1 and make all RLDxP/N = 0. From the diagram, you can see that in that case the amp is just a buffer of the (AVDD+AVSS)/2 and there is no feedback from the analog input pins. For the specific test your customer is doing, I'd imagine that is how they would eliminate the noise.

    In the case where any of the RLDxP/N bits are set, this will provide negative feedback to the amplifier from the inputs i.e. if the common mode starts to drift higher, the amplifier's output to the RL electrode will pull low (or vice versa) in an effort to actively stabilize the input common mode to mid-supply. For an application, the RLD output would travel through the electrode to the body and its feedback effect would be felt through the impedance of the body and those electrodes and so it will not be as pronounced as if one were to connect the RLD output directly to the inputs. One could imagine a situation where the feedback network had a low enough impedance that the output would tend to oscillate. I think this is what is happening with your customer.

    Is there a reason they are performing this test specifically? If they wanted to test the RLD's ability to stabilize the common mode while still using the feedback capabilities, I'd recommend testing it through some sort of simulated body contact. Hopefully this explanation makes sense.

    Regards,
    Brian Pisani
  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your excellent reply and your excellent kindness!
    We got about RLD amplifier almost.
    However, we'd like to confirm one point.

    <Confirmation>
    You mentioned as the following;
    "One could imagine a situation where the feedback network had a low enough impedance that the output would tend to oscillate.
     I think this is what is happening with your customer."

    We can not understand,
    if the feedback network had a low enough impedance, why would the RLD-output tend to oscillate(pulsate)?
    Could you let us know it?

    As you mentioned as the following, is it relation to the negative feedback to the amplifier?
    "If the common mode starts to drift higher, the amplifier's output to the RL electrode will pull low (or vice versa)
     in an effort to actively stabilize the input common mode to mid-supply."

    Furthermore, we heard from our customer that RLDOUT-IN1P-IN1N shorted test are defined 120sec no noise as the rule of heart rate monitor.
    That's why they had to pass this test.

    We appreciate your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Hirotaka Matsumoto
    TI__Mastermind 38745 points
    Matsumoto-san,

    The phenomenon I was describing was "overshoot". A small change in the input causes a larger change in the opposite direction in the output which, since it's tied directly to the input, causes another larger change in the input which is then corrected at the output. The effect is an oscillating negative feedback system. If you do the math on the amplifier, you'll find that the strength of the response is dictated by the feedback components R_EXT and C_EXT. The magnitude of the gain relates directly to the magnitude of the impedance created by the parallel combination of R_EXT and C_EXT. Decreasing that impedance should decrease the gain of the amplifier which should in turn decrease the overshoot. I'm not sure exactly which values to place there, but I'd recommend some experimentation to find what works best.

    Regards,
    Brian Pisani
  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your prompt and excellent reply!

    OK, in order to supress overshoot, it is needed to decrease RLD's gain. 
    So, we got that we have to try to experiment what works better to supress overshoot.

    We would like to confirm your mention about R_EXT and C_EXT.

    Does these - R_EXT, and C_EXT mean followings at the user manual P51 of ADS1x9xECG-FE Demonstration Kit, doesn't it?
    (ADS1x9xECG-FE Demonstration Kit: http://www.tij.co.jp/jp/lit/ug/slau384a/slau384a.pdf)
    R_EXT: R35, R36
    C_EXT: C23

    We need your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your excellent reply!
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention.

    Do R_EXT and C_EXT mean followings?

    <For example from ADS1x9xECG-FE Demonstration Kit>
    R_EXT: R35, R36
    C_EXT: C23
    http://www.tij.co.jp/jp/lit/ug/slau384a/slau384a.pdf

    We need your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Brian Pisani
    Guru 27525 points


    Brian san,

    Thank you for your excellent reply!
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention.

    Do R_EXT and C_EXT mean followings?

    <For example from ADS1x9xECG-FE Demonstration Kit>
    R_EXT: R35, R36
    C_EXT: C23
    http://www.tij.co.jp/jp/lit/ug/slau384a/slau384a.pdf

    We need your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Hirotaka Matsumoto
    Guru 27525 points

    Hi Brian san,

    Thank you for your excellent reply!
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention.
    Do R_EXT and C_EXT mean followings?

    <For example from ADS1x9xECG-FE Demonstration Kit>
    R_EXT: R35, R36
    C_EXT: C23
    http://www.tij.co.jp/jp/lit/ug/slau384a/slau384a.pdf

    We need your help.
    Kind regards,
    Hirotaka Matsumoto

  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your excellent and prompt reply!
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention.

    Do R_EXT and C_EXT mean followings?
    <For example from ADS1x9xECG-FE Demonstration Kit>
    R_EXT: R35, R36
    C_EXT: C23
    http://www.tij.co.jp/jp/lit/ug/slau384a/slau384a.pdf

    We need your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your excellent reply!
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention.

    Do R_EXT and C_EXT mean followings?

    <For example from ADS1x9xECG-FE Demonstration Kit>
    R_EXT: R35, R36
    C_EXT: C23
    <http://www.tij.co.jp/jp/lit/ug/slau384a/slau384a.pdf>

    We need your help.
    Kind regards,
    Hirotaka Matsumoto

  • 0 in reply to Brian Pisani
    Guru 27525 points
    Brian san,Thank you for your excellent reply!
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention.Do R_EXT and C_EXT mean followings?
    <For example from ADS1x9xECG-FE Demonstration Kit>R_EXT: R35, R36
    C_EXT: C23www.tij.co.jp/.../slau384a.pdfWe need your help.
    Kind regards,Hirotaka Matsumoto
  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your excellent reply!
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention.

    Do R_EXT and C_EXT mean followings?

    <For example from ADS1x9xECG-FE Demonstration Kit>
    R_EXT: R35, R36
    C_EXT: C23
    http://www.tij.co.jp/jp/lit/ug/slau384a/slau384a.pdf

    We need your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your excellent reply!
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention.

    Do R_EXT and C_EXT mean followings?

    <For example from ADS1x9xECG-FE Demonstration Kit>
    R_EXT: R35, R36
    C_EXT: C23

    We need your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your excellent reply!
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention.

    Do R_EXT and C_EXT mean followings?

    <For example from ADS1x9xECG-FE Demonstration Kit>
    R_EXT: R35, R36
    C_EXT: C23
    www.tij.co.jp/.../slau384a.pdf

    We need your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Brian Pisani
    Prodigy 20 points

    Brian san,

     

    Thank you for your excellent reply!

    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.

    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.

    We'd like to confirm about your mention.

     

    Do R_EXT and C_EXT mean followings?

     

    <For example from ADS1x9xECG-FE Demonstration Kit>

    R_EXT: R35, R36

    C_EXT: C23

    http://www.tij.co.jp/jp/lit/ug/slau384a/slau384a.pdf

     

    We need your help.

     

    Kind regards,

     

    Hirotaka Matsumoto

  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your excellent reply!
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention.

    Do R_EXT and C_EXT mean followings?

    <For example from ADS1x9xECG-FE Demonstration Kit>
    R_EXT: R35, R36
    C_EXT: C23
    http://www.tij.co.jp/jp/lit/ug/slau384a/slau384a.pdf

    Followings show resister setting.

         0x01,      //Adress 01 CONFIG1  250sps
         0xA0,      //Adress 02 CONFIG2    (E0->A0)
         0xF0,      //Adress 03 LOFF
         0x00,      //Adress 04 CH1SET (PGA gain = 6)
         0x00,      //Adress 05 CH2SET (PGA gain = 6)
         0x23,      //Adress 06 RLD_SENS  RLD connected IN1N/1P
         0x00,      //Adress 07 LOFF_SENS  (0F->00  Lead-OFF Disable)
         0x00,      //Adress 08 LOFF_STAT
         0x02,      //Adress 09 RESP1
         0x02,      //Adress 0A RESP2 RLDREF is fed internally
         0x0C       //Adress 0B GPIO

    If CH2 is not used, should it be disabled?(Address05 CH2SET 0x80?)
    Is it relation to cause of overshoot?
    And then, on the hardware, our customer treats as following;
     IN2P-3.0V
     IN2N-GND

    We need your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your excellent reply!

    At the first, because of network failre, we unintentionall contributed multiplely.
    And then, we tried to delete multi contribution, but we could not delete and showed message "unknown error".

    By the way,
    In order to supress overshoot, it is needed to decrease RLD's gain with changing R_EXT and C_EXT.
    So, we got that we(and our customer) will try to experiment what works best to supress overshoot.
    We'd like to confirm about your mention and some contents.

    Do R_EXT and C_EXT mean followings?

    <For example from ADS1x9xECG-FE Demonstration Kit>
    R_EXT: R35, R36
    C_EXT: C23
    http://www.tij.co.jp/jp/lit/ug/slau384a/slau384a.pdf

    Followings show resister setting.
         0x01,      //Adress 01 CONFIG1  250sps
         0xA0,      //Adress 02 CONFIG2    (E0->A0)
         0xF0,      //Adress 03 LOFF
         0x00,      //Adress 04 CH1SET (PGA gain = 6)
         0x00,      //Adress 05 CH2SET (PGA gain = 6)
         0x23,      //Adress 06 RLD_SENS  RLD connected IN1N/1P
         0x00,      //Adress 07 LOFF_SENS  (0F->00  Lead-OFF Disable)
         0x00,      //Adress 08 LOFF_STAT
         0x02,      //Adress 09 RESP1
         0x02,      //Adress 0A RESP2 RLDREF is fed internally
         0x0C       //Adress 0B GPIO

    If CH2 is not used, should it be disabled?(Address05 CH2SET 0x80?)
    And then, on the hardware, our customer treats as following;
     IN2P-3.0V
     IN2N-GND

    We need your help.

    Kind regards,

    Hirotaka Matsumoto

  • 0 in reply to Hirotaka Matsumoto
    TI__Mastermind 38745 points
    Hello Matsumoto-san,

    No worries about the multiple replies. In reference to your question regarding the feedback components for the RLD amplifier, only R36 and C23 constitute the feedback components. R35 is not since the RLDREF is fed internally and the RLDREF pin will be disconnected from the amplifier internally under that circumstance. For your question regarding the unused channel settings, it shouldn't affect the RLD amplifier output since the RLD_SENS is only connected to the channel 1 pins, but in general we recommend powering down unused channels and shorting their inputs internally to save power. For that, the CH2SET register should be 0x81.

    Recalling something I mentioned in one of my first responses. I believe that if the RLD_SENS register was set so that no inputs were used by the RLD amplifier, that the noise would disappear. I'm not sure if the use of the common-mode cancelling feature is a requirement for your customer, but it might be worth trying simply to validate my theory. To be clear, I'd like to try making the RLD_SENS register equal to 0x20 (RLD enabled with no input feedback) and see if that eliminates the noise altogether.

    Regards,
    Brian Pisani
  • 0 in reply to Brian Pisani
    Guru 27525 points

    Brian san,

    Thank you for your prompt reply!

    <Reply to your request>
    -Our cutomer uses the common-mode cancelling feature.
    -OK, we(our customer) will try testing with no input feed back as the following. 
     "To be clear, I'd like to try making the RLD_SENS register equal to 0x20 (RLD enabled with no input feedback) and see if that eliminates the noise altogether."

    And then, now our customer is checking the noise.
    They mentioned that it maybe be caused by voltage drop(power source).
    That's why, we recommended that our customer has to clarify the problem at the first. 

    If we get update, we will inform you soon.

    Kind regards,

    Hirotaka Matsumoto