AD 8237 in precision low cost multichannel biological amplifier (ECG, EMG, EEG) circuit - questions.

Dear colleagues,

I am currently designing circuit to measure biological signal - ECG, EMG, (and simple EEG if possible). Due to

battery powering and low cost, I am planning to use AD8237 instrumental amplifiers. I have some question I want to

ask you before ordering and designing PCB.

I read the datasheet carefully, and spotted there scheme for ECG circuit. I made some notes on the scheme:

 

 

My questions:

 

1) Why 47nF capacitor was used in parallel with gain setting resistor?

2) Why 110kOhm/22nF lowpass filter is placed on the output? Only for filtering?

3) Why a capacitor is used in parallel with resistor setting Vmid voltage. What should be its value?

4) Is it possible to use different values of the elements in lowpass integrator? F.e 3MOhm/1uF?

5) is it possible to power whole circuit with single Li-Ion 18650 (4.2-3.3V)? I want to use single battery and do

not want any switching regulators (due to noise it generates).

 

6) I want to make 3-channel universal biological amplifier. How should I connect the op-amps?

As far as I understand the scheme, electrode C (right leg) is for dealing with noise and DC component (thats why I

am using inverting integrator) - so I need only one electrode like this (C) for all the channels.

Electrode A and B are simply non-inverting and inverting input of the amplifier, so I can say, that A is a value

of the measurement, and B is the 'reference voltage' (A is measured referred to B). So if I want to have more than

one channel, I should connect all inverting inputs to electrode B and place electrodes A (A, A', A'') to points on

the body I would like to measure, am I right?

For the next channel I need only the next AD8237 + gain setting resistors?

And I connect all REF pins of all AD8237's to the output of inverting integrator?

 

7) I want to use shielded cables due to better noise immunity. Due to the discussion:

https://ez.analog.com/message/167615#167615

I assume, that I need to take the signal from 'between' the inputs and drive it to the shield? Due to the costs I

want to use one shield potential for all of the cables. I have written it on the schematic using blue color - is

it correct (the shield will be connected to the SH pin)?

About the filtering: I know what filters to use, so please focus on the questions. I will use single passive RC

highpass with tau (RC) = 3 (cutoff = 0.05 Hz, medical standards) and double passive RC lowpass (with cutoff

frequency = 15kHz) as I use sigma-delta ADC with 1MHz modulator.

I am waiting for your help as I study electronics in medicine and I am strongly interested in construction of

low-cost biological amplifier

 

best regards,

Mateusz Winkowski

 

PS: Is it possible to buy some AD8237's together with the samples? They are not expensive, but the shipping cost

is huge, and I can not get them in Poland (I can sample 2 pieces per two weeks, but I do not want AD to waste

money for the shipping it many times, so I would like to buy additional pieces and ship them together with the

samples).

Parents
  • I hope the following helps:

     

    3) The larger the capacitor value, the better, sort of. Capacitors have an equivalent circuit that is a series RLC, shunted by a series RC, where this RC is composed of a very large resistance and a small capacitance. The values depend upon the chemistry of the capacitor and the manufacturing process. There is a lot going on here, but the most significant information is that the capacitor will have an impedance that increases quite a bit above AND below a certain frequency. I suggest you use a ceramic capacitor of about 0.01 to 0.1 microfarad (noise is high frequency, of course). If you are concerned about lower frequencies, solve for a capacitor to give the desired cutoff frequency. I like to use a value that is 4 to 5 times this calculated capacitor value, when practical. To avoid the use of very large capacitors, you can split the upper resistor (the one connected to the supply) into two values and add a capacitor at the node between these resistors and ground. However, the real intent of this the capacitor in the figure is to reduce the noise generated by the resistive divider. A better circuit to limit low frequency noise would be to use a voltage reference with good PSRR to drive the AD8607 input. Most voltage references can tolerate a capacitor on their output. Voltage references generally have tighter output voltage tolerances than most LDOs.

     

    4) There is no problem using large resistors. However, large resistors are noisier and, to avoid a voltage change at the op-amp input, that op-amp must have a very low input bias current.

     

    6) I do not see the need for 9 cables. The differential signal between nodes “A” and “B” is amplified by the AD8237. These are “sensing” leads. The “driven” lead goes to Node “C” and is called “driven” because it is driven with the Vmid voltage. This provides a mid-point reference around which the differential signal between “A” and “B” swings. The human body can pick up power line and other electromagnetic noise that becomes a common mode signal. With node “C” in place, the circuit rides on this common mode signal, enhancing the common mode rejection of the system. Since the AD8237 has high impedance inputs, I do not understand the need for further buffering. Also, Vmid is driven by an AD8607, so I see no need to further buffer it. I would wait for ADI’s response on this.

    For the AD8237, the input signal will internally swing around the Vref pin voltage. In the Figure 77 circuit, Vref is the same potential a Vmid. Any low frequency signal at the AD8237 output is integrated, greatly reducing that signal. Because each AD8237 could have a different low frequency noise signal, I would use a separate integrator (REF-pin-driver) for each AD8237.

    7) I am not sure how to drive the shields, but I would think that they should be driven by Vmid. Wait to see what ADI says.

     

    8) While I am not positive of this, it seems to me that setting Vmid to 1.25V should be O.K. What you need to avoid is having the AD8237 output distorting as the output signal swings towards it supply rails of zero and Vcc. You can check the data sheet to see how close to the rails the output can swing without distortion. Using the Vcc potential for the ADC reference is a good option, but there are two concern. First, be sure the chosen ADC can use Vcc as a reference. Second, Any noise on the reference pin will come directly out of the ADC without any attenuation. And, yes, it is true that you are more concerned with the potential between electrodes “A” and “B” rather than the exact voltage of either or both of these electrodes.

    9) I cannot answer this question. Please wait for ADI to answer.

Reply
  • I hope the following helps:

     

    3) The larger the capacitor value, the better, sort of. Capacitors have an equivalent circuit that is a series RLC, shunted by a series RC, where this RC is composed of a very large resistance and a small capacitance. The values depend upon the chemistry of the capacitor and the manufacturing process. There is a lot going on here, but the most significant information is that the capacitor will have an impedance that increases quite a bit above AND below a certain frequency. I suggest you use a ceramic capacitor of about 0.01 to 0.1 microfarad (noise is high frequency, of course). If you are concerned about lower frequencies, solve for a capacitor to give the desired cutoff frequency. I like to use a value that is 4 to 5 times this calculated capacitor value, when practical. To avoid the use of very large capacitors, you can split the upper resistor (the one connected to the supply) into two values and add a capacitor at the node between these resistors and ground. However, the real intent of this the capacitor in the figure is to reduce the noise generated by the resistive divider. A better circuit to limit low frequency noise would be to use a voltage reference with good PSRR to drive the AD8607 input. Most voltage references can tolerate a capacitor on their output. Voltage references generally have tighter output voltage tolerances than most LDOs.

     

    4) There is no problem using large resistors. However, large resistors are noisier and, to avoid a voltage change at the op-amp input, that op-amp must have a very low input bias current.

     

    6) I do not see the need for 9 cables. The differential signal between nodes “A” and “B” is amplified by the AD8237. These are “sensing” leads. The “driven” lead goes to Node “C” and is called “driven” because it is driven with the Vmid voltage. This provides a mid-point reference around which the differential signal between “A” and “B” swings. The human body can pick up power line and other electromagnetic noise that becomes a common mode signal. With node “C” in place, the circuit rides on this common mode signal, enhancing the common mode rejection of the system. Since the AD8237 has high impedance inputs, I do not understand the need for further buffering. Also, Vmid is driven by an AD8607, so I see no need to further buffer it. I would wait for ADI’s response on this.

    For the AD8237, the input signal will internally swing around the Vref pin voltage. In the Figure 77 circuit, Vref is the same potential a Vmid. Any low frequency signal at the AD8237 output is integrated, greatly reducing that signal. Because each AD8237 could have a different low frequency noise signal, I would use a separate integrator (REF-pin-driver) for each AD8237.

    7) I am not sure how to drive the shields, but I would think that they should be driven by Vmid. Wait to see what ADI says.

     

    8) While I am not positive of this, it seems to me that setting Vmid to 1.25V should be O.K. What you need to avoid is having the AD8237 output distorting as the output signal swings towards it supply rails of zero and Vcc. You can check the data sheet to see how close to the rails the output can swing without distortion. Using the Vcc potential for the ADC reference is a good option, but there are two concern. First, be sure the chosen ADC can use Vcc as a reference. Second, Any noise on the reference pin will come directly out of the ADC without any attenuation. And, yes, it is true that you are more concerned with the potential between electrodes “A” and “B” rather than the exact voltage of either or both of these electrodes.

    9) I cannot answer this question. Please wait for ADI to answer.

Children
No Data