Single lead ECG ( RA- LL )

Hello Brain ;

To summary what you suggest I draw the shematic please could you comment on it ?  

and Do you think I need pullup on the IN1P and pull down on IN1N ?

What is the kind of the Dc blocker capacitor) ceremic ones ?    what about the values  capacitors and resistors ? 

Thank you inadvanced 

ads1291.PDF

Hi Brain ; 

We discussed this with my team ; we already have a design based on ADS1291 ; so the aim now is to  use the blocking DC capacitor ; but that would create an headache  to detect pace signal since pace signal and AC lead signal will interface . And also we need to add a huge FIR filter ; we decide to add an external lead detect circuit with  comparator; 

Do you think my shematic now is ok ? thank you Brain 2350.ads1291.PDF

Hi Brain ; 

Sorry I was out the office the last 2 days I could not reply ; thank you for your effort to help us . 

I modified the drawing ; Do you think we need C26  for DM filter ? What is your comments about our anti alliasing filter and the values ? 

for RLD we set to VDD/2  ; do we need to  mux IN1P and IN1N to the RLD opamp ? 

For the lead off detect ; 

We will sample at 8khz ;  then  LPF at cut off 1800 hz and then pacemaker detect algorithm ;  ( FIR ) 

a HPF  or BAND PASS FILTER ( 1800-2200) ( FIR or IIR  )  for AC lead detect. 

7181.ads1291.PDF

Sorry my image did not attach properly. It is attached to this post.

Hello ; 

I am back after I get the prototype PCB on my office ; 

We could not make the circuit works ; 

we can see in the scope the 2khz AC lead signal but it is very difficult to analysis it in the digital world; 

we tried to send 6nA and 6microA ; but  it seems it s very noisy ; 

what we did is the bandpass filter about 2khz  ; and we calculate the  power of 8 samples ( at 8khz )  

in open circuit the power should be he highest ; in short circuit the power is the lowest  ; some times it works. 

when we attach it to human body it not stable . 

our VADD is 4.6V   RLDout is 2.3V;

internal gain PGA : 6 

internal REF : 4.033V 

the shematic of the prototype is here 

5001.ads1291.pdf

5001.ads1291.pdf

Please could you guide us to find the problem and solve ; we already tried everything 

Thank you in advanced 

Regards 

Hi Brain

i will post some graphics when i go ofice. I think we have problem in the algorithm of analysis the signal. we are calculating the power of the signal only. Do you suggest a diffirent algorithm?

Thank you

Hello

I am sorry for the delay ;

I post some FFT result of the signal ;

We collect 512 data at 8000 and we perform a 256 FFT ; and we connect the electrodes to human body; open and short and with one electrodes.

As we noticed  if the electrodes are open a good 2khz signal is reflected ; same for short electrodes with less magnitude;  but for real human body other frequencies apprears on the FFT ; also with one electrode connected .

please could you commend of the graphics ; and could you guide us to resolve this

Thank you in advanced for your effort

electrodes.pdf

Hey Hector,

I'm not really sure what the graph is showing. What is the y-axis unit? Where is DC? If it's possible you can just send me the time domain data and I can analyze it.

Regards,

Brian

Hello Brain ; 

Sorry for the Delay ; I needed time to go to the clinic inorder to connect the device to human . 

Here are  down I attach many result ; 

the first case are :  

we get samples at 8000Hz ; then (101 TAPs) low pass  120Hz FIR  filter then decimante to 250hz ; then again low pass  35Hz   (252TAPs) FIR filter ; then IIR filter to remove DC  with cofficient = 0.95 . . 

when we get the result from Fluke simulator the result is ok ; but when we connect to Human there is no ECG . 

the second case for impedance track ( the impedance of the electrodes ) we sample at 8000hz to any filter  I stored 1000 samples and I send put the data of each of : open circuit, shortcircuit , to 50ohm resistance and human body . 

8000hz_50ohm.xlsx

8000hz_human_2.xlsx

8000hz_human_1.xlsx

8000hz_open.xlsx

ecg_1.xlsx

I have other question please could you answer : is it a must to use AC lead detection method incase no RLD electrode ? even in battery powered application ? 

Thank you in advanced 

human_1.xlsx

noise_5mv_ECG_50hz.xlsx

Thank you Brain for your effort ;

I will test that in the clinic and  inform you about disable lead detect signal and measure ECG.

for impedance tracking you main I should use an HPF about 2Khz ; and track the power or the FFT ? Do you suggest any thing about the filter for example IIR with how many db ?

Thanks

In the previous posts I have uploaded 8000hz data without any filtering , on human , open electrodes, short electrodes and 50 ohm resistor ; please could check if I am on the right road ?
Thank you advanced Mr Brain

Hello Brain ;

I think the last result is mixed ;

I will go to the clinic now and after few hours I will send you  a new data :

open circuit electrodes

short circuit electrodes

50 ohm electrodes

1k ohm electoredes

and connect to human  data .

we  use only 20 bits according to data sheet noise free bits are 22bits ; so we read 24 and shift the data  to use only  20 bits .

#define FULL_SCALE         0xFFFFF // 20 bits fıll scale

#define  VREF                                          2.42f
#define LSB_weight                                 VREF/ 524287  
#define adc_gain    6                                // PGA gain
#define adc_total_gain                             (float)(1000*LSB_weight/adc_gain) // to get the volt directly

   if ((adc_data& 0x00080000) == 0x00080000)   // check if negative
   {
  
    adc_data_1=FULL_SCALE-adc_data;
    adc_volt_8000=((float)adc_data_1)*adc_total_gain*-1.0f;
    
    }
    else
    {
     
    adc_volt_8000=((float)adc_data)*adc_total_gain;
     }   

Today I will send better data from the clinic .

Thank you for support Mr Brain ;

Hello Brain ;

I have arrived from the clinic ; I could only connect to 4 people;

the data is  collected 22 bits as mVolt you are right ; in order to send the data to PC for debug purpose we multibly by 1M ;

we get data from ADS1291 by this code :

 for(i=0;i<6;i++)
  {
   spi_buffer[i]=readspi_data_adc();
   
   }
   adc_data=0;
   adc_data_1=0;
   adc_data=(signed int)((spi_buffer[3]<<16))|(signed int)((spi_buffer[4]<<8))|(signed int)(spi_buffer[5]);

   adc_data=adc_data>>4;
  
   if ((adc_data& 0x00080000) == 0x00080000)   // check if negative
   {

    adc_data_1=FULL_SCALE-adc_data;
    adc_volt_8000=((float)adc_data_1)*adc_total_gain*-1.0f;
    
    }
    else
    {
     
    adc_volt_8000=((float)adc_data)*adc_total_gain;
     }   
 

we first check if negative them we change it to mv ;

For the previous question I forget to answer about QRS  when we connect to fluke we could see the QRS very clear as in the previous attached, but for human body it is very noise

here the clinical data for today ;

7080.8000hz_human_1.xlsx

8156.8000hz_human_2.xlsx

8000hz_human_3.xlsx

8000hz_human_4.xlsx

8000hz_opencircuit_electodes.xlsx

8000hz_shortcircuit_electrodes.xlsx

Thanks a lot Brain

Hi Brain ; 

Thank you for your support and interest; 

the problem of not getting the QRS is the electrodes quality ; in few days we will go to teh clinic with better electrodes. 

However ; today we just wanted to see the power of the signal open short and one electrode conencted the right and the other to the left hand of human ;

we use order 4  bandpass IIR filter 1950-2050 HZ  . 

we noticed that we can see good 2000hz signal ; 

we read 200 samples and calculate the power of the signal ; we noticed that  the power of the open electrodes is higher than the power of short electrodes and that is expected. But with electrodes with right and left hand the power is even less than of short circuit electrodes. 

It is a little strange for us ; any idea about it ?  because we expect a power somewhere between open electrodes and short electrodes. 

Thank you inadvanced 

Infact in our application we want only 3 level indicator;  not connected electrodes; low contact electrodes and high contact electrodes;

the measuring of 5 impedances of human all  between 12k ohms to 40k ohms ; 

with open circuit electrodes it is : 65k ohms

short circuit: 56k ohms  (  closed to the open circuit ..)any idea?

one electrode connected to the human and other floating : 180K

Hello Brain ; 

Here is the shematic ; 

ECG_P= 4.8V 

RLOUT= 2.4 V 

Thank you inadvanced 4064.ads1291.pdf

low_pass.h

Hi Brain; 

Here the low pass filter it is 25hz ; 252 order fir ; sampling frequency is 250hz . 

Do you suggest an analog circuit that improve the single lead reading before ADS1291? 

Hello Brain ; 

Yes we are sampling at 8Khz  then a nyqest filter of 120hz ; then DECIMATOR by 32 to get 250sps ECG . 

Thank you again 

Hello Brain ; 

The 252 order FIR is after decimated ; it is at 250 sps data ;  but before decimated we use 120hz fir at 8khz data .