Splitting two AC signals for the human pulse

Discussion in 'General Electronics Chat' started by Hurt_it_Circuit, May 26, 2014.

  1. Hurt_it_Circuit

    Thread Starter Member

    Oct 2, 2012
    53
    0
    Hi,

    I am building a circuit that is measuring the pulse inside the body. I need to measure pulse wave velocity. To do so I have two sensors at two separate locations and from this I can get distance which will then give me velocity. The issue I am having is that I need a high gain (10 Million) in order to measure the pulse on the foot for example. I am able to get the signal but, it appears that the pulse is being coupled into the other circuit and showing in its output as well. I was wondering if there is an AC coupling trick that I could do to get rid of this effect. Currently I have capacitors across the power rails to store charge if the power rail fluctuates. My thought is that for these two circuits the power rails are all connected, so the pulse may get sent into the other circuit and get gained up. Would having more stages but the same gain help with this or is there something else I can do?
     
  2. wayneh

    Expert

    Sep 9, 2010
    12,120
    3,046
    Can you share your schematic and maybe a photo of your build? It should not be too hard to isolate the two sensors from each other.

    Are you sure you really need 10^7 gain for a pulse signal? You can feel someone's pulse with your fingers, so I find it odd that you would need much gain.
     
  3. #12

    Expert

    Nov 30, 2010
    16,284
    6,797
    I'd guess we're talking about strain gages, but guessing doesn't work very well.
     
  4. Hurt_it_Circuit

    Thread Starter Member

    Oct 2, 2012
    53
    0
    The last stage is a 4th order Sallen Key filter with a cutoff frequency at 3Hz. Prior to that are two inverting gain stages. Note that the other circuit has the same 4th order filter but the gain is less since the pulse on the neck does not need as much gain. We are using an accelerometer to sense the pulse. Also for each op amp we have added DC block caps for each of there terminals in case the power supply fluctuates.
     
  5. #12

    Expert

    Nov 30, 2010
    16,284
    6,797
    I'd say you have a gain of 10^8 but that isn't the problem. Crosstalk is the problem, but I don't have an idea to fix it right now.

    Edit: Maybe you should be looking for the higher frequency of the contraction and discarding below 3 Hz instead of passing below 3 Hz. Can you still pick that frequency up at the extremities or does the elasticity of the arteries damp that out?

    Another...add 4 aluminum electrolytic capacitors across the batteries to keep the voltage of the supply from, "dancing". I drew you a picture.

    Another, make the gain of each stage equal to each other. It makes their bandwidth the same if you do that.

    Another: Do you need a 100 ohm load on the sensor? The non-inverting configuration has much higher input impedance. With an input impedance of a million ohms or more, you might get a lot more out of the sensor and need less gain. Problem is, I don't know about accelerometers.
     
    Last edited: May 26, 2014
  6. Hurt_it_Circuit

    Thread Starter Member

    Oct 2, 2012
    53
    0
    Not quite sure what you mean by your first point about discarding below 3 Hz. I have not tried the diode with capacitor for the rails, I will give that a shot. As for the accelerometer, this is my first time using one. I will give this a shot as well and let you know.
     
  7. #12

    Expert

    Nov 30, 2010
    16,284
    6,797
    Sufficient to say I am unsure about your frequency of choice. It seems you are only measuring below 3 Hz. True?
     
  8. Hurt_it_Circuit

    Thread Starter Member

    Oct 2, 2012
    53
    0
    Yes, I am only interested in below 3 Hz. I think I may try larger capacitance across my rails to better stabilize them. If all else fails I will just run them off of two separate power supplies.
     
  9. Lestraveled

    Well-Known Member

    May 19, 2014
    1,957
    1,215
    A few suggestions:

    I suspect that your 3Hz filter is attenuating the response you want to detect. I am guessing that the pulse travel time is in the 10s of milliseconds. Your 3Hz filter implies a 333 millisecond response time. Responses in the 10s of milliseconds are now in the dirt.

    Put an electrostatic shield around your first high gain op-amp and components. This shield can be anything from a small square of thin brass (connected to ground) from the hobby store, to several wires connected to ground that "surround" your circuit. If your circuit needs this shielding, anything conductive and connected to ground will make a big impact.

    Why the NE5532s so late in the amplifier chain? These are exceptionally low noise amplifiers but this feature is lost when they are used so late in the circuit. They could be replaced with the cheaper LF357.

    This is a tough measurement to make. You will have to tune your parameters very carefully to make this work.
     
    #12 likes this.
  10. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,538
    1,251
    The 10 ms travel time is the time between two sensors, not the time of a single pulse from one sensor (and amplifier). So while 3Hz does seem kinda low, the time difference between two separate 3Hz band-limited signals 10 ms apart should be measurable. However, the risetime might be so slow that getting an meaningful result will be tough.

    Back to the beginning - what is the risetime of the pulse signal that is traveling? A better understanding of this can be used to optimize the filter section.

    Also, what is the sensor and what are its output characteristics (voltage, current, source impedance)?

    ak
     
  11. Lestraveled

    Well-Known Member

    May 19, 2014
    1,957
    1,215
    AnalogKid nailed it. Your filtering should be based on the rise time of the pulses you are trying to detect/analyze. You are trying to determine the time difference between event A and B. The leading edge of the pulse contains the data you want to measure. Any filtering applied must be higher than the bandwidth of that rise time otherwise you attenuate the data you want to measure.
    In other words, open up your filter band width. Since you have a reoccurring pulse you can use integration to suppress noise. You could even do statistical analysis on the pulses.
     
  12. Hurt_it_Circuit

    Thread Starter Member

    Oct 2, 2012
    53
    0
    The sensor is an accelerometer and its output characteristic is voltage. The rise time varies since it is a pulse.
     
  13. wayneh

    Expert

    Sep 9, 2010
    12,120
    3,046
    Have you looked at the output of both sensors on an oscilloscope? Seems to me that seeing what you're doing would help a lot. I believe the oscilloscope would also be able to directly give you the delay time you are looking for. You would tell it the time offset needed to match up the two traces as close as possible.
     
Loading...