First, I have to admit that I messed up building the conditioning circuit. The first RC network after the first op-amp stage should have had a 6.8kΩ resistor. I accidentally used a 68kΩ resistor instead.
My next error was in the RC network just before the second op-amp stage. The circuit called for a 0.01µF capacitor. Instead, I used a 0.1µF capacitor.
These errors are a big deal because they throw off the desired frequency response of the entire circuit. The bode plot below shows the measured response of the incorrect circuit in blue, the simulated correct response in red and finally the simulated incorrect circuit in yellow:
The newly simulated circuit has a response that is similar in shape to the actual (incorrect) circuit. However, the actual circuit seems to have better gain performance than the simulated one.
The unfortunate news was that once I corrected the values in the circuit, another problem cropped up.
Once I powered on the circuit, there was an odd sinusoid that would not leave even when after the geophone was left untouched for a long period of time:
With an oscilloscope setting of about 2 ms/cm and 100 mV/cm, the sinusoid measured about 200 mVpp at around 120 Hz.
At first, I suspected that the wires connecting to the geophone were acting as an antenna and were picking up the second harmonic of the 60 Hz power in the house. I made a ground plane for the circuit to see if that would help at all:
I’m not overly sure if I’ve even used the ground plane properly but I hoped that the rogue signal would attenuate a bit. Unfortunately, that wasn’t the case.
I then thought that maybe the circuit was self-oscillating. There is a positive feedback loop at the second op-amp stage which might be causing grief. I never thought that I would have to go back and refresh myself on the Barkhausen criterion. After looking at the simulated bode plot, I couldn’t see a reason why the circuit would oscillate. There wasn’t any frequency where the conditions (gain of 1 and phase shift of 0 or 2πn) were satisfied.
After some thought and experimenting, I noticed that when I added the proper capacitor (0.01µF as opposed to 0.1µF), problems would arise. The 68k/0.1µ RC network has a cutoff frequency of about 23 Hz. Could it be possible that the geophone was picking up this 120 Hz signal this whole time but was unseen because of the filtering? With the correct capacitor (68/0.01µ), the cutoff frequency is at about 234 Hz, wide enough to include the 120 Hz signal. Was there a 120 Hz vibration? If so, where was it coming from?
It turns out that the oscilloscope was the source of all my troubles. I moved the geophone to another surface and just like that, I saw the 120 Hz signal disappear:
Because the geophone was picking up this signal, that meant that the oscilloscope was vibrating at 120 Hz! My theory is that the transformer in the oscilloscope vibrates (hums) due to the alternating current – much like how you can hear the larger power transformers buzz loudly. From the Wikipedia article on mains hum:
Magnetostriction is a second source of vibration, where the core iron changes shape minutely when exposed to magnetic fields. The intensity of the fields, and thus the “hum” intensity, is a function of the applied voltage. Because the magnetic flux density is strongest twice every electrical cycle, the fundamental “hum” frequency will be twice the electrical frequency. Additional harmonics above 100 Hz or 120 Hz will be caused by the non-linear behavior of most common magnetic materials.
Well, this theory sounds entirely plausible. It could also be possible that the table has a resonant frequency close to 120 Hz and aids in the transmission of the vibrations.
That brings us to a couple of conclusions. First, the seismometer should be kept well away from any power mains. Second, this observation means that the geophone is crazy sensitive! You can’t feel any vibration whatsoever on the table but the geophone had no problem picking it up!
This was pretty fun and frustrating detective work but it’s time to move along and put the finishing touches onto the conditioning circuit. Next week, I will focus on changing the circuit to fully take advantage of the geophone’s features, then, we will start the steps to digitizing the data.