Russell McMahon wrote:

>I wish to measure human pulse rate in an exercise environment.
>
>Known options include (but are not limited to) ecg direct contact
>measurements using various sensor locations (chest (eg polar), hand to hand
>etc) and blood flow variation (typically thumb or ear clip.)
>
>I want to try to produce good results using an ear clip sensor but may in
>due course end up using some other method.
>
>As ear clip sensors (IR LED to photo-detector via ear lobe & blood) are
>common enough one might expect that effective circuitry would be easy enough
>to develop. Getting reliable signals is more annoying than I expected :-).
>Interestingly (or annoyingly) the sorts of problems that I experience match
>those reported by users when I search the web. So maybe nobody ever really
>sorted this out well. A major factor is the wide variation in optical
>transmission through the ear lobe,  time to time, and user to user, and in
>session. Sensor movements or variations in clip tension on earlobe lead to
>shifts in the DC operating point as effective ear transmissivity varies
>which is seen as a substantial signal level change. As the AC levels are low
>(millivolts) such changes swamp the detector circuit until the DC level
>again stabilises. Driving the detector via a feedback loop to keep its
>operating point stable helps but is not a perfect solution. The received
>signal must be reasonably aggressively low pass filtered to remove 50 (or
>60) Hz hum artefacts. The low frequency of the desired signal leads to large
>capacitor and resistor values (around the 1 uF and 1 to 10 Mohm range) which
>makes DC operating point susceptible to capacitor leakage (so electrolytics
>no good) and offset bias currents.
>
>I have a few more ideas to try (eg feedback modulate the IR LED rather than
>changing the detector operating point) but I think it's liable to prove an
>annoying application still.
>
>Interestingly, I was unable to find a single circuit diagram on web although
>there are many mentions of equipment using ear clips
>Did turn up the complete circuitry for a pulse oximeter along the way :-)
>
>SO - anybody tried this or can suggest any useful circuit ideas?
>
What you need is quiet simple: DC correction based on let say 1 or 2
seconds of signal (I think it's impossible to realize it be linear networks)
Having a period of 1 sec, you catch a complete waveform from the heart,
so you have the maximum and minumum, and can correct the DC value
according. If you're talking about bloodflow, 50 Hz sampling is far
enough, so you only need a buffer of 50 samples. The problem of course
is to get enough dynamic range, but with a 12 bit AD converter it should
be just enough.
About the hum of 50 or 60 Hz, if you circuit is well designed it isn't
coming from electrical sources, but from light sources in the room ;-)
So a good remedy would be to use an IR filter, and to shield the sensor
from light.

Measuring in an exercise environement makes it more complex.
You could use a Polar band and catch the signal with a coil (5 kHz puls
modulation).
In exercise environement I would go for good electrodes (Ag-AgCl from
3M, low DC drift) on the chest (Einthoven-II), combined with carbon
shielded cables (no movement noise) and a amplifier with a good CMRR (no
hum). You can even achieve this with (almost ;-)) standard Lego RCX, see
here
 http://oase.uci.kun.nl/%7Emientki/Lego_Knex/Lego_electronica/BioSensors/ECG_sensor.htm

Stef Mientki

>
>
>
>        Russell McMahon
>
>--
>http://www.piclist.com hint: The list server can filter out subtopics
>(like ads or off topics) for you. See http://www.piclist.com/#topics
>
>
>

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics