Hello,

I'm working on a human pulse detector, which must be applied to read the pulse at the radial artery (wrist).
For this I'm basically using a LDT0-028K piezo sensor from Measurement Specialties, Inc. The output of the piezo sensor is connected to a charge amplifier whose simulation circuit is shown in the figure.
The amplifier has been tested on breadboard and seems to be working fine. The sensor works fine as long as it is vibrated manually. However there is no output to the wrist pulse, I've tried different configurations of placing the sensors;clamps, wrist straps, etc. But no matter how much tight i hold the sensor there is no pulse pick up.
Could you help me troubleshoot?
I have very little time to complete this project...

I'm wondering if the sensitivity of the sensor is low. However, there are a couple of project papers where they have used the same sensor.
Can you give any suggestions for better sensors? Or the configuration so that noise due to human motion is reduced and pulse pick up increases?

Thank You.

 

R1 and R2 are really 100MΩ !!??

The sensor works fine as long as it is vibrated manually. However there is no output to the wrist pulse, I've tried different configurations of placing the sensors;clamps, wrist straps, etc. But no matter how much tight i hold the sensor there is no pulse pick up.

When you manually test it, I suppose you are tweaking it more rapidly and/or violently than a small pulse?

 

It looks like to me that your circuit has very low ac gain. The voltage gain will be approximately the impedance of the C1 divided by the impedance of C2. This is way less than 1. You need to eliminate C2 (or drastically reduce it's value) to get voltage gain.

The transducer has a very high impedance at the frequencies of a human pulse thus the need for the 100 Meg ohm feedback resistor. The data sheet implies that the transducer's capacitance is 480 pF. This calculation confirms table 3 in the data sheet: Xc = 1/(2 pi f C), Xc = 1 /(2 * pi *3.3 Hz * 480 pF), Xc = 100 Meg ohm.

In the past, I used a non-inverting amplifier with positive ac feedback ("bootstrap") to get a very high input impedance for the transducer. The positive feedback gets rid of the need for extremely high value resistors like the 100 Meg ohm one.

See section 3.1 in the application note I have uploaded for information on the amplifier I am talking about.

 

Just yesterday I went to see my vascular team for a followup checkup. When listening to and monitoring pulse and blood flow they always use a conductive gel with the pickup heads they use. I haven't a clue exactly what transducer you are using or the sensitivity but have you tried using any sort of acoustic conductive gel?

Ron

 

It looks like to me that your circuit has very low ac gain. The voltage gain will be approximately the impedance of the C1 divided by the impedance of C2. This is way less than 1. You need to eliminate C2 (or drastically reduce it's value) to get voltage gain.

The transducer has a very high impedance at the frequencies of a human pulse thus the need for the 100 Meg ohm feedback resistor. The data sheet implies that the transducer's capacitance is 480 pF. This calculation confirms table 3 in the data sheet: Xc = 1/(2 pi f C), Xc = 1 /(2 * pi *3.3 Hz * 480 pF), Xc = 100 Meg ohm.

In the past, I used a non-inverting amplifier with positive ac feedback ("bootstrap") to get a very high input impedance for the transducer. The positive feedback gets rid of the need for extremely high value resistors like the 100 Meg ohm one.

See section 3.1 in the application note I have uploaded for information on the amplifier I am talking about.

Yes the amplifier has low gain and that is why i have used a second stage for just amplification... I dont think the amplification circuit has any much problem (unless Im missing something) cuz Im getting an output voltage of almost 9V (saturation of opamp) for any kind of small vibrations...

 

Just yesterday I went to see my vascular team for a followup checkup. When listening to and monitoring pulse and blood flow they always use a conductive gel with the pickup heads they use. I haven't a clue exactly what transducer you are using or the sensitivity but have you tried using any sort of acoustic conductive gel?

Ron

Ohh... I never thought about it. I should try.
Thanks a ton

 

Yes the amplifier has low gain and that is why i have used a second stage for just amplification... I dont think the amplification circuit has any much problem (unless Im missing something) cuz Im getting an output voltage of almost 9V (saturation of opamp) for any kind of small vibrations...

Hey but this acoustic gel i just read abt, is used for ultrasound applications. Will it work for a piezo sensor? Reading the principle, It doesnt look quite positive

 

R1 and R2 are really 100MΩ !!??

When you manually test it, I suppose you are tweaking it more rapidly and/or violently than a small pulse?

It is not reading small pulses :/
Only with a large bend on the sensor.

 

Im getting an output voltage of almost 9V (saturation of opamp) for any kind of small vibrations...

It is not reading small pulses :/

So is it working or isn't it ?

 

Hey but this acoustic gel i just read abt, is used for ultrasound applications. Will it work for a piezo sensor? Reading the principle, It doesnt look quite positive

Merely a suggestion. Again, I haven't any idea exactly what sensor you are using? Do you have maybe a link to a data sheet?

Ron

 

U2 has a gain of 55 dB. That is way too much for something that has no offset voltage management and is DC coupled to something that is not GND. 10 mV of offset at the output of U1 will saturate U2. At a minimum, put a large capacitor in series with R9 to reduce the offset and bias errors at the output. Maybe a CMOS chopper stabilized opamp would give you the combination of input impedance and DC stability you need.

Separate from that, I don't see how a sensor designed to detect bending will capture the extremely small skin movement during a pulse. Maybe a true pressure sensor would work better. What about a solid state air pressure sensor? Any $40 blood pressure monitor reads the pulse rate through the cuff, and that is based on air pressure.

Also, is it a specific requirement to determine the heart rate via the wrist pulse, or can you use other methods. Electrical contacts on the chest will do this very reliably.

ak

 

A small electret mic might make a suitable sensor (or is that what your 'piezo sensor' is already?) .

 

Also, is it a specific requirement to determine the heart rate via the wrist pulse, or can you use other methods. Electrical contacts on the chest will do this very reliably.

Like those little fingertip IR pulse rate detectors too, those are pretty slick.

Ron

 

U2 has a gain of 55 dB. That is way too much for something that has no offset voltage management and is DC coupled to something that is not GND. 10 mV of offset at the output of U1 will saturate U2. At a minimum, put a large capacitor in series with R9 to reduce the offset and bias errors at the output. Maybe a CMOS chopper stabilized opamp would give you the combination of input impedance and DC stability you need.

Separate from that, I don't see how a sensor designed to detect bending will capture the extremely small skin movement during a pulse. Maybe a true pressure sensor would work better. What about a solid state air pressure sensor? Any $40 blood pressure monitor reads the pulse rate through the cuff, and that is based on air pressure.

Also, is it a specific requirement to determine the heart rate via the wrist pulse, or can you use other methods. Electrical contacts on the chest will do this very reliably.

ak

What do u mean by R9? And what is the problem with the opamp going into saturation?
I'll look into the solid state air pressure sensor specifications.
Moreover, Yes I do need to pick up from the wrist, my aim is to observe the pulse pressure waveforms (requirement of an ayurvedic doctor)

 

Looking at your data sheet a big part of the problem could be in the mounting of the sensor. Setting aside pulse have you just tried applying stress according to the data sheet? If you know how much stress movement happens as a result of a pulse then using again the data sheet amplifyng the signal should be relatively easy. These units have a relatively high output.

Ron