anyshake
Documentation: Geophone + Datasheets
Documentation for the geophones seem to be missing.
- What geophones are used?
- Where can we find the datasheet for it?
I think I found it. It seem to be this one:
LGT-4.5C (LGT-20D4.5C)
3-axis 4.5 Hz velocity geophone (100 V/m/s sensitivity)
- http://www.longetequ.com/geophone/183.htm
However, there still is no proper data sheet showing it's frequency response, and the sensitivity seem a bit low compared to high sensitivity ones with: 1 Hz @ 280 [V/m/s].
Either way, this should be added to the BOM if correct.
As long as the manufacturer provides the necessary detector parameters, the frequency response can be accurately derived using the detector's transfer function.
I have just created a simulation model using TINA, which is available in the hardware/simulation.TSC. This model demonstrates the frequency response of our system, showing that the sensitivity effectively doubles after compensation, with the bandwidth extended to 0.5-27 Hz.
We have just completed a round of refactoring, so some documents are still being prepared. I'll do my best to complete them as soon as possible, thank you for your support.
Hi @bclswl0827
Great, Thank you!
I have just created a simulation model using TINA ...
I'm not familiar with TINA. However, the graph is still only a simulation so it's very likely not accurate below below 0.5 Hz, which is the region I'm interested in.
This model demonstrates the frequency response of our system, showing that the sensitivity effectively doubles after compensation
(a) What kind of compensation are you talking about?
(b) So the other question is, why are you not using a more sensitive geophone for this project?
I'm not familiar with TINA. However, the graph is still only a simulation so it's very likely not accurate below below
0.5 Hz, which is the region I'm interested in.
We have performed vibration shaker testing on the device and verified that AnyShake Explorer does achieve gain flatness from 0.5 - 27 Hz. I will upload the test result soon.
(a) What kind of compensation are you talking about?
The natural frequency of LGT-4.5C is 4.5 Hz. According to the frequency response graph above, the gain of the detector drops rapidly below 4.5 Hz due to the mechanical structure of the detector. Therefore, a circuit is required to amplify the signal below 4.5 Hz while retaining the gain above 4.5 Hz.
After the seismic waves propagate, the high frequencies are often attenuated, leaving only the low frequencies. Compensating the low frequencies of the detectors can increase the AnyShake Explorer cycle, thereby monitoring earthquakes farther away.
(b) So the other question is, why are you not using a more sensitive geophone for this project?
Since this product was designed as an open-source alternative to closed-source solutions like the Raspberry Shake, the goal was to match their performance as closely as possible. As a result, a higher-sensitivity detector was not selected. The choice of the LGT-4.5C geophone was driven by design-for-manufacturing (DFM) considerations, ensuring a balance between performance, cost, and ease of production. We can also accept customized business if you need.
@bclswl0827 Great answer. Thank you.
I will upload the test result soon.
Looking forward to see that.
Finally, is the hardware able to use a LF geophone of higher sensitivity?
For example, one designed for 0.05 Hz @ 800 V/m/s?
one designed for 0.05 Hz @ 800 V/m/s
Please note AnyShake Explorer is currently not compatible with this type of geophones. These geophones typically have extremely high sensitivity, resulting in a very limited measurement range. To extend their usable range, a external resistor divider network must be carefully designed. Additionally, such geophones often include built-in signal conditioning circuits (e.g., force balance feedback, pole-zero compensation, etc.), which means they are intended to be connected directly to an ADC chip without the need for further signal compensation.
If you plan to use this kind of geophones, we would recommend designing your own data acquisition circuitry and interfacing it with the AnyShake Observer via our data protocol (a friend of mine did this and successfully connected a BBVS-60 to AnyShake). This will allow you to integrate seamlessly into the AnyShake ecosystem. The data protocol will be published in the near future (we currently have many TODOs, so please be patient).
The following two pictures show the effect of connecting AnyShake to the BBVS-60 and a minor earthquake occurred on May 21. (The scaling factor of helicorder has been significantly reduced because this equipment has a period of 60 seconds.)
Currently, the recommended detector models have been added to the BOM. https://github.com/anyshake/explorer/blob/416c5766cd737a8b4ad6d4060de3ecbc7288cee6/hardware/Explorer.csv#L30-L31 Thank you for your issue, it clearly answers the questions that many users may have.
If you plan to use this kind of geophones, we would recommend designing your own data acquisition circuitry and interfacing it with the AnyShake Observer via our data protocol (a friend of mine did this and successfully connected a BBVS-60 to AnyShake). This will allow you to integrate seamlessly into the AnyShake ecosystem.
Would be great to see how this was done.
I was looking at the filtering circuitry, and not sure what's are the thought about it. Seem to be missing the design white paper or whatever background research criteria the project was based upon.
I was looking at the filtering circuitry, and not sure what's are the thought about it. Seem to be missing the design white paper or whatever background research criteria the project was based upon.
In our system, the "raw" signal is passed through a VCVS second-order filter compensation network to enhance or suppress different frequency components precisely. This compensation stage consists of:
- High-Pass Filter (HPF) – Removes DC offset and low-frequency drift (Omitted in our design because the response is almost constant over the entire effective frequency band)
- Band-Pass Filter (BPF) – Emphasizes signal features in the frequency band of interest
- Low-Pass Filter (LPF) – Suppresses high-frequency noise and enhances signal clarity
All three filters share a common second-order system structure, expressed as:
$$ H(s) = \frac{N(s)}{s^2 + 2\zeta \omega_c s + \omega_c^2} $$
Where:
- $s$ is the Laplace variable
- $\omega_c$ is the system's center angular frequency (rad/s)
- $\zeta$ is the damping ratio (which controls bandwidth)
For band-pass filter (BPF), transfer function is:
$$ H_{\text{BP}}(s) = \frac{2\zeta_0 \omega_0 s}{s^2 + 2\zeta \omega_c s + \omega_c^2} $$
Center frequency:
$$ f_0 = \frac{\omega_c}{2\pi} $$
Bandwidth:
$$ BW = \frac{2\zeta \omega_c}{2\pi} $$
Lower and upper cutoff frequencies:
$$ f_L = f_0 - \frac{BW}{2}, \quad f_H = f_0 + \frac{BW}{2} $$
For low-pass filter (LPF), transfer function:
$$ H_{\text{LP}}(s) = \frac{\omega_0^2}{s^2 + 2\zeta \omega_c s + \omega_c^2} $$
Cutoff frequency:
$$ f_{\text{LP}} = \frac{\omega_c}{2\pi} \cdot \sqrt{1 - \zeta^2} \quad $$
In our implementation, we chose:
- Center angular frequency: $\omega_c = 2\pi \cdot 0.5 = \pi , \text{rad/s}$
- Damping ratio: $\zeta = 0.707$
With these parameters, we can compute the actual frequency characteristics of each filter stage.
Afterward, you may modify the previously mentioned simulation schematic to observe the frequency response of each stage individually and verify the effectiveness of the design.
Those are indeed beautiful equations, but without being put in terms of the physical component values, I'm afraid they are nothing but theory.
Have you checked you equations and values? (See #9 ) Please check the f_c's for each of the filter stages using the schemtics values...
All component values have been carefully checked and verified through both simulation and small-scale closed testing. The design has passed all internal validation procedures.
We're currently in the crowdfunding phase — you're very welcome to follow the project. And if you still have doubts, we encourage you to get a unit and evaluate it yourself. Nothing speaks louder than real-world results.