I'm building a data logger using a MKR ZERO in order to measure and record signals from a geophone. The geophone has a sensitivity of 28 V/m/s but I'm finding the signal to noise ratio still isn't good enough for my purposes, even using max gain on the ADC. I've wired it up according to this diagram except using the inbuilt ADC :
As per the diagram I'm using 1k resistors in series with both the positive and negative poles. I understand this protects the circuit in terms of an over-current but I'm wondering whether this could also limit the sensitivity of the geophone and so perhaps reducing the resistance might help? Also, I'm currently sampling at 12bit, would oversampling for a 14 or 16 bitrate help increase SNR?
Apologies, I'm fairly new with electronics so I only know the basics but any thoughts on how to improve sensitivity would be much appreciated.
Use an op amp to increase the signal. The DIY Nerdaq amplifier is specifically designed for geophones and Arduino, with a very low bandwidth (DC to about 50 Hz).
For better informed help, post a link to the geophone.
Pay attention to the output impedance of the converter and drive it with a compatible opamp. it would be appropriate to use a high impedance fet input. The impedance of geophones can be 1megaohm and above.
The opamp specs are mentioned here in the ardunio.cc header section, for example, I think a low impedance converter ne5534 and ad797 would fit
opa602 or opa627 for high impedance
Good to know that you are dealing with such sound and pressure waves.
Everything after the geophone ADDS to the noise! Signal and NOISE are all amplified the same amount. You need to use the minimum number of components and any active devices need to be selected for low noise contribution. Converting from analog to digital only adds to the noise by introducing distortion. You may have to test a bunch of amplifying components to find the one with the lowest noise addition.
Given the differential inputs, am I right in saying IN+ and IN- would just be the inputs from the positive and negative geophone poles respectively, and then the single output would be read by the ADC? Or would I need two op amps, one for each pole?
@johnerrington Yes this was a concern. Am I right in assuming that op amps are able to discriminate the signal somehow and increase the SNR? Because otherwise what would be the point, you could just multiply all the amplitudes by a constant in post-processing!
The TLV2462 op amp you linked needs to be in a circuit similar to the Nerdaq circuit, so it is still "DIY".
For that circuit, most op amps will work OK, but a rail to rail op amp that works well at 5V, like the TLV2462, is preferred.
Am I right in assuming that op amps are able to discriminate the signal somehow and increase the SNR?
No. The associated circuitry and data processing does that. "Low noise" is of course desired, meaning that the op amp does not add much noise. But at this point, it sounds like just getting something working will be a significant learning experience.
Okay, and presumably it does this by amplifying certain frequencies more than others, which correspond to the signal of interest? I guess what I'm trying to work out here is whether working an op amp into the design has any benefits over increasing the gain at post-processing stage, which can also be done selectively to certain frequencies with fourier or wavelet transformation. If there's no difference and actually the SNR could reduce because of added noise, then it sounds like my only option is a more sensitive geophone.
No need to guess, a low noise amplifier is standard procedure. With so little understanding of what you are doing, you really, really need to spend more time reading up on what others have done.
You might start by explaining the actual purpose of this project. If it is DIY earthquake detection and warning, there is a huge network of amateurs and professionals working on the general topic and sharing their expertise and results. If the project is to detect people walking across the floor, or driving up to your house, you don't need a amplifier.
I do have a lot to learn but, with respect, you're not answering my point on how SNR can be increased with an op amp. Others have stated that noise and signal are amplified by the same amount so if the problem is low amplitude signals are getting lost in the noise, how does this help?
The amplifier itself will introduce its own noise, along with power supply noise.
So it all adds up.
So the cleaner the signal you begin with the better.
Noise and signal are separated by a process called "filtering". The Ole Wolf circuit provides that filtering. You connect the geophone at JP1 and the output from pin 1 of the op amp connects to your ADC.
The TLV2462 will be suitable - but you need to build the circuit above - or similar - around it.
Low impedance coil is giving minimum noise, resistors and capacitors produce more noise. For protection you can use amplitude limiter, however the coil is producing max voltage much less than 5V, the sensor is similar to microphone . Remove resistors, show the software.
The sensor bandwidth is in hundreds of kHz, that is not good for your application, you can reduce it by adding 1uF capacitor parallel in to sensor or use amplifier/filter suggested in post #2
Presenting appropriate voltage levels to the ADC dramatically reduces the quantization error introduced by the digital conversion. A 3 mV peak to peak sine wave cannot be accurately represented by an ADC with 1 mV resolution, but a 3V p-p sine wave can be.
Also, filtering the input during the amplification reduces or eliminates sampling noise. Frequencies in the input that are higher than (sample frequency)/2 are aliased to lower frequencies, adding to the quantization error. There is plenty of high frequency noise in footsteps on a floor.
No geophone needed for that, a simple contact microphone will do.
Like a 1" piezo disc from a music toy.
Doesn't need a preamp or external A/D either.
Leo..
