03-16-2010 07:16 AM
I am using Laview 8.5.
i have square wave signal from the skin electrose and it contail some peaks. how do i filter out this peaks ontop of my square wave signal so that it doesnt affect my readings.
please see attached picture.the red signal is my required signal and you can see the peaks.i want to reduce or eliminate the peaks.the first signal is the imput voltage to the skin.
thanks
03-17-2010 03:22 AM - edited 03-17-2010 03:30 AM
Looks like an square wave excitation and a a disturbed square wave input.
For n periods build a histogram and identify the peak values to eliminate.
Take a look at the waveform measure vis to measure pulse attributes ..
Or
Use Sine excitation and use a bandpass filter (and/or a MSE sinus fit for amplitude and phase, it's quite robust)
(I hope you are not stimulating a living human skin with a 10V square waves for skin resistance..... I have in mind that about 0.5V works best for this , AND USE APPROPRIATE ISOLATION!)
03-18-2010 07:03 AM
sorry Henrik
i try all that and it doesnt work.
any other suggestion pleas?
03-18-2010 08:25 AM - edited 03-18-2010 08:26 AM
Unless you have variable DC offset on the red signal, you may just coerce to the maximum/minimum allowable value (they seem to be 1 & -1 on your red chart).
Regards
03-18-2010 09:48 AM
I don't know how you tried, since you didn't provide code and hardware information.
At a second look your samplerate looks quite low...
03-18-2010 04:16 PM
another problem i am facing is the sample rate. maybe if i can change the sampling rate.then maybe i can avoid the part of the signal that is having the leaks.
i will upload the Vi and also the picture of the block diagram .
the red signal is the signal from the skin.i know it has to be noisy but how can i eliminate the noise(peaks) it affect my reading.
the hardware am using is from Quancom and am using their Vi. The Ch0V1 is the signal i put in from fungtional generator and Ch1V2 is the red signal am interested in.
please can you show me the a small picture of what you mean.
thanks
see attached
03-19-2010 04:01 AM
Some point so far:
place the content of sequence 0 and 3 outside the loop.
use the error in-out and the handle to determin dataflow .
I don't know your device, but currently you are doing (not even) software timing (one value is requested per loop interation) .
Read the manual (RTFM) if the DAQ device support a hardware timing to read more data faster and with a defined samplerate.
OK, I personally try to avoid express vis.... but try to wire from left to right .. 😉
If you don't need the Chart , just delete it (your userinterface need a clean up !)
Get these write to file vis out of your data read loop! (System calls like writing to disk tend to cause undefined delays)
The solution here is a producer consumer architecture. one loop get the data feeding a queue. A second loop read the queue, does the calculation, display and store the data (in high speed/datavolume applications the datastorage is again an extra loop, however I thing you don't need that )