THANK YOU R. K. Johnson At 11:27 PM 6/30/99 +0200, you wrote: >OK ! Interesting thoughts ! Now I see what you mean with skin reactance. The >tric would most likely be quite sensitive, as this probably could detect >changes very early in the "sweating phase". > > >Sven > >-----Ursprungligt meddelande----- >FrŚn: Robert K. Johnson >Till: PICLIST@MITVMA.MIT.EDU >Datum: den 30 juni 1999 10:52 >€mne: Re: SV: Re: SV: Re: Help Researching Galvanic Skin Response Circuit >Designs > > >>At 11:00 PM 6/27/99 +0200, you wrote: >>>Best Robert, >>> >>>Most likely someone else wrote "...For one: You want either a WELL-isolated >>>power supply, if you're going to plug this unit into AC power (And I don't >>>mean a typical "Wall >>>Wart", I mean something that's certified for medical use!) - OR, you >>>want to use a deep-cycle lead-acid battery (probably a better choice!)..." >>> >>>which makes me suggest batteries is best bet for this application. Very much >>>so, as the application suggests electrodes attached to a human body. I am >>>sure you are familiar with these issues. Kindly understand that my comment >>>is aimed towards any reader who would make his own electronic circuits, and >>>attach it to a body. >>> >>>2nd comment: Well - I did read the original question - maybe a bit fast. The >>>reason I suggest accelerometers is not to detect a seizure; that is too late >>>- yes. That was the event to be avoided. What I mean is that there might be >>>other ways to detect a situation that would precede the seizure. Sweating is >>>one, but possibly the occurence would be preceded by an increase in "tossing >>>and turning", which would be easier detecable as an increase in >>>accelerometer activity over - say - a 3 minute period or so. Another guess - >>>but remember that this also is just a guess as we know very little of the >>>situation - is that signals from a microphone to a PIC could assist. >>> >>>I don't believe that 2 to 5 Volts DC would result in just a few microamperes >>>of current through electrodes, and I suggest that 2 to 5 Volts DC is way too >>>much for monitoring purpose over - say a whole night. I also suggest that >>>"just a few hundred nanoseconds" is too short time to measure anything but >>>reactive componens. Sweat is a lot of ions, and you should let them get a >>>move on to be able to measure by applying a small voltage for several >>>milliseconds. So I suggest that a fairly low frequency is applied for skin >>>counductivity measurements. BUT it may NOT be DC. The ions that are moved >>>under the skin, must go back again. Rephrased: the capacitor in series with >>>the output of the exciting oscillator (or whatever) is important; making >>>current resultant 0, like you said. >>> >>>Threshold of feeling is about 500 microamperes under 100 Hz. Currents >>>smaller are harmless, but must not have a DC component. This is all just >>>clearifying the DC issue, that you seem aware of. I am not at all an expert >>>on GSR, but hope that the presented problem can be solved without getting >>>into too much trouble with GSR. A change in GSR as an indicator ? well - if >>>applied right and the change is very distinct - maybe...:=) >>> >>>Anyway - most imortant is of course that this person can be helped. I hope >>>that the assistance progresses, and a solution emerges. >>> >>>Best Regards >>rest of message deleted for the sake of brevity... >> >> My suggestions were not intended as any method of making an absolute >>measurement, Rather as a way of detecting a change in conditions. As to >>the small pulses >>my intent was to make reactance measurements. I am not concerned with >>subdermal ionic activity in this suggestion. I do believe that there will >>be significant changes in skin conductivity when the body begins to sweat. >>This effect will change the capacitance of a pair of electrodes coupled to >>the subject. and seem to be a nonintrusive method of detecting the sweating >>condition using insulated non contacting electrodes. The µProc should >>generate a variable width pulse stream and adjust its pulse width to center >>the input at the detection threshold (logic level)... This will require a >>voltage follower with a fast rise time as the input mpedance of a typical >>PIC is in the order of 20K ohms with 20 to 50 pf of capacitance... >>(these figures are from memory so dont shoot me if I am off in my estimation). >> >> There are many possibilities here for a measurement system that >should be >>easy to code and cheap to build... The WDT in a pic16lc711 can be >>progbrammed to operate at a 2.18 secomd (approx) period turn itself on >>(housed in a watch type case) make a measurement, compare it against the >>previous sample... seems to me that the ram comtents are stable in sleep >>mode and if the error condition is detected could ring an LC circiut with >>short pulses to be detected by a close am radio... >> >> Robert K. Johnson >> rkj1@ix.netcom.com >>