Life Science

cancel
Showing results for 
Search instead for 
Did you mean: 

Low Frequency Detection (DAQ)

We are trying to develop a vi that is able to measure the respiration rate of a subject.
 
We are using a piezo electric film that changes voltage with strain to the chest and gives a somewhat clean sine wave when the subject is stationary. However, this sine wave is of very low frequency. The average respiration rate is around 15 breaths per minute, which results to a frequency of 0.25 Hz, or an event every 4 seconds.
 
 
We tried using Extract Single Tone Information and outputting the frequency. However, this value is jumping around because of the small noise that is present in the signal. The small noise is taking precedence over the small frequency it seems.
 
When signal simulation with noise is used, the frequency counter (from Extract Signal Tone Information) correctly counts the frequency for frequencies greater than 10 Hz. Frequencies lower than this value are chaotic. This indicates that nothing is wrong with our measurements and DAQ but with low frequencies the frequency counter does not work.
 
To sum it up: Is there anyway to detect frequencies of very small values? Perhaps a method of filtration? We also tried using the SINAD but we weren't sure what it was doing.
0 Kudos
Message 1 of 6
(10,177 Views)
Have you tried amplifying the signal you're sending? It would require an instrumentation amplifier (IC), but is doable. Also, doing this, you would need a low pass filter to eliminate the noise. I suggest you do all this electronically rather than programmatically.
0 Kudos
Message 2 of 6
(10,168 Views)
We tried amplifying the signal but doing so amplifies the noise as well, which throws off the frequency counter even more. I don't think amplifying the signal electronically will work because we are already almost hitting the limits on our DAQ range.
 
We're trying to implement a low pass filter now, but we're not sure what "toplogy" we should be using. Do you know of any references that would help explain the difference between them?
0 Kudos
Message 3 of 6
(10,164 Views)

I would try a low pass filter with 3-10Hz ** and than use a threshold detector to measure the time between two breaths. Or track the time of the last n breaths and do some more math....

The peak detector also is worth a try, if you define the peakwidth big enough your noise shouldn't trigger a false peak

**) Ups, I assume that DC offsets where already eliminated. if not define a band pass of 0.01 to 3Hz ....

For such low frequence signals, it can help to implement a simple RC alaising filter and choose a not to high (>1kHz) sampling rate. On the other hand: If you noise is spiky a median filter can help a lot.   

 

Message Edited by Henrik Volkers on 10-02-2007 02:31 PM

Greetings from Germany
Henrik

LV since v3.1

“ground” is a convenient fantasy

'˙˙˙˙uıɐƃɐ lɐıp puɐ °06 ǝuoɥd ɹnoʎ uɹnʇ ǝsɐǝld 'ʎɹɐuıƃɐɯı sı pǝlɐıp ǝʌɐɥ noʎ ɹǝqɯnu ǝɥʇ'


0 Kudos
Message 4 of 6
(10,146 Views)
How long is the sample set you are sending to the tone extraction VI? If I recall correctly, that VI uses FFT internally. If you are giving it less than 3-4 full cycles of the frequency you want to measure, the results may not be very good. I have done some informal studies which found that FFT-based frequency estimators do not work well with less than the amount of information mentioned above. >10 cycles is better. This means that you should pass at least 15-20 seconds of data to the VI. It also means that if the respiration rate is varying during that time, the result will be some sort of average. Even nurses like to have 30 seconds or a minute of counting time to measure respiration.

If you need to track the rate on a breath by breath basis, you probably need to use time domain techniques, such as period measurements.

Lynn
0 Kudos
Message 5 of 6
(10,140 Views)

The reason I suggested amplifying your signal is because I had the same proiblem when implementing an EKG using a DAQ as well. Of course the noise would also be amplified, that's why you'd require an instrumentation amplifier, not just an opamp. Plus a Low Pass filter, as someone else has already suggested you. There's a useful article in Scientific American that describes it more thoroughly. I scanned the article for my project (if you're interested, I can send them to you). The reference is " Home is Where the ECG Is" by Shawn Carlson. Scientific American Magazine, June 2000.

As for the filters, the book "Power Electronics" by Rashid is what I would recommend. I have some more info, but I'm afraid it's in Spanish. Might that be useful to you?

Cheers.

Ali

Message Edited by Ali84 on 10-02-2007 09:00 AM

0 Kudos
Message 6 of 6
(10,131 Views)