Today I took a picture of the scope traces from the F-Scan2 during a DIRP at 1360 Hz. The F-Scan2 reportedly shows spurious peaks at 1360Hz, 2720Hz, 5080Hz and so on, every 1360 Hz. This is caused by aliasing of the ADC (analog-to-digital converter) in the F-Scan2 and I will show you how that happens:
The top trace is the sampling signal for the ADC in the F-Scan2 (scope channel connected to the RESET pin on the ADC). Below is the end of the F-Scan2 frequency burst signal as measured from the F-Scan2 output. The horizontal deflection is 2 ms/div.
You can see that these two frequencies match - this causes the F-Scan2 to give spurious peaks during a DIRP scan.
Recording a signal with frequencies larger than half of the sampling frequency (or Nyquest frequency) causes 'aliasing'. Here the F-Scan2 samples the signal from the finger probe 16 times (shown by the 16 peaks on the upper trace) and probably averages over them for noise reduction. It compares these averaged ADC values of two consecutive frequencies and displays the difference on the screen. If the phase of the first signal just happens to record 16 times the upper peak and on the next scan 16 times the bottom peak of the 1360 Hz signal, then it will show a strong peak in the DIRP display. By the way, the F-Scan2 also samples 16 times during the silence between two consecutive bursts - it uses this for its continuity value (CV).
The F-Scan2 uses frequency burst of 12Vpp with a DC offset of 6V. The finger electrode picks up a small portion of the DC and with it a small amount of the AC burst. When the ADC samples the 'biofeedback' signal from the finger electrode, which should (theoretically) only be a slow varying DC signal, the AC signal that leaks through causes the F-Scan2 to show these spurious peaks on the DIRP screen.
This should convince everybody.
So much for these spurious signals. But has the F-Scan2 any value? Are there any peaks that do not fall into this category?? In order to find out I encourage everybody that owns an F-Scan or F-Scan2 to add a low pass filter to the input for the finger electrode. The finger electrode has two contacts, one is grounded and that you can easily verify with an Ohm meter. The other should show a resistance of about 100 kOhm - this is the ADC input.
On my F-Scan2 this input is the pin of the finger adapter connector where a red wire is connected (inside the F-Scan2). Desolder the red wire from this pin (the pin from the connector where the finger adaptor is plugged in - not the connector that goes to the printed circuit board) and insert a 1/8 Watt 1kOhm resistor (shorten the leads) between the red lead and the pin where the red lead was before. Now connect a 1uF capacitor (e.g. Radio Shack 1uF multilayer ceramic capacitor, do not use an electrolytic cap) from ground to the red lead. Ground is on the pin where the black lead is connected. Be careful not to shorten anything and ground yourself when you solder, have the F-Scan's power supply disconnected. You can ground yourself by wearing the finger electrode and have the finger adaptor connected to the F-Scan! So the signal from the finger electrode first passes through the resistor and then to the capacitor and to the red lead, do not reverse this!
This R-C network (the network of the resistor and the capacitor) will cut-off frequencies above f = 1/(2*pi*R*C) = 159 Hz (this is the 3dB point). At 1360 Hz this filter will have an attenuation of ~19dB, or by a factor ~9 and at 2720 Hz it will be attenuated by ~25 dB or a factor ~17 and so on. An active filter would be better but that will be left for people with greater electronic expertise. The RC network should not affect the sampling of the biofeedback signal since the F-Scan2 steps through the frequencies every 0.75 seconds or at a rate of ~1.3 Hz. That frequency is much, much lower than the 3dB point of the RC network (159Hz) so it will pass the body's response through to the ADC. I really wonder why the manufacturer didn't implement this filter - maybe there won't be anything left to show during a DIRP?
I did this with my F-Scan2 and after that the DIRP signals 'flat-lined', there was nothing to be seen. Zip-zero, dead silence. I removed the filter and the spurious signals were back (so I didn't kill the F-Scan2)!!
Correction (05/01/05): I looked through my past data and the spurious peaks were still present with the filter in place. However the amplitude was greatly reduced!
Now I have read through half of the 2,100 posts on the Yahoo F-Scan forum so far and some people have nonetheless achieved remarkable results using the F-Scan or F-Scan2 in therapy mode after finding some 'peaks' and zapping themselves at those frequencies. So why did they get better? I think it is despite the DIRP, not because of it.
However, there was one Excel spreadsheet posted on the Yahoo forum that showed one test subject where the CV value (the continuity value) had changed substantially during a DIRP. This could indeed be a real response - however that scan was not repeated and we are left to wonder whether this was just a faulty connection of the finger electrode (or the conductive patches that were worn instead of the finger electrode) or the real signal?? Such a real response would pass through the RC low pass filter without attenuation, not doubt about this!
Please read also my recent follow-up post on my earlier F-Scan evaluation: