Hello DigiKey TechForum community,
I would like to introduce a new hobby electronics project that I am beginning to develop: Sikwal BeanSense.
The objective is to build a compact, battery-powered device that estimates the moisture content of a green coffee bean using a custom capacitive sensor. The measured value will be processed by a microcontroller and displayed as a percentage on a small OLED screen.
This project is currently in the concept and sensor-prototyping stage. The sensor geometry, calibration equation, schematic, and PCB have not yet been finalized.
Proposed Hardware
Custom capacitive sensor with copper electrodes
Texas Instruments FDC1004 capacitance-to-digital converter
ATMEGA328P-AU microcontroller for the first prototype
1.3-inch I²C OLED display
Push button to begin a measurement
3.7 V rechargeable lithium battery
3.3 V power regulation
ISP programming connection and T56 programmer support
Proposed Measurement Method
A single green coffee bean will be placed inside a repeatable plastic holder between two insulated copper electrodes.
The initial measurement sequence will be:
Measure the capacitance of the empty holder.
Insert the coffee bean.
Take multiple capacitance readings using the FDC1004.
Calculate the capacitance difference.
Convert the result into an estimated moisture percentage using calibration data.
The FDC1004 was selected because it provides high-resolution capacitive measurements, an I²C interface, offset-capacitance compensation, and active shield outputs that may help reduce interference and parasitic capacitance.
Main Engineering Challenges
Variations in bean size, density, and orientation
Creating a repeatable sensor holder
Minimizing parasitic capacitance and external interference
Compensating for temperature and environmental changes
Producing a reliable calibration curve
Comparing the prototype with a trusted reference moisture meter
The first goal is not to claim laboratory accuracy. It is to determine whether the sensor can produce stable and repeatable readings that correlate with known coffee-bean moisture values.
Development Plan
Study and explain the FDC1004 datasheet.
Design and test several copper-electrode geometries.
Measure the empty sensor and different coffee samples.
Build the preliminary schematic in EasyEDA.
Develop the ATMEGA328P firmware.
Create the PCB and enclosure.
Calibrate and document the results.
Share the schematic, PCB, firmware, and test findings as the project develops.
I would appreciate feedback from the DigiKey TechForum community, particularly regarding sensor geometry, shielding, FDC1004 PCB layout, reference measurements, and practical calibration methods.
This will be a learning-focused project combining two of my interests: electronics and coffee.
Regards,
Sikwal
Bean orientation and distance from the electrodes could easily introduce more variation than moisture itself. Characterize the empty holder first and track temperature during measurements. You can test multiple beans across several known moisture levels. Repeated measurements for each sample would help quantify repeatability. Comparing the results against a trusted moisture meter should also reveal the practical accuracy.
Thank you, these are very useful points. I agree that bean orientation, electrode spacing, temperature, and holder capacitance can all affect the reading. I plan to characterize the empty holder first as a baseline, then test repeated measurements using multiple beans at known moisture levels while recording temperature. I also plan to compare the results with a trusted commercial moisture meter to evaluate repeatability and practical accuracy. This should help determine whether the capacitive approach is reliable enough for the BeanSense project.
Hey sikwal 
,this is a pretty in-depth project idea, very ambitious. I am not well versed in sensor geometry or shielding. There are a few spots on our site that can help you with research that we find helpful. If you hover over Resources at the top of our Main Page:
I have put marks next to the spots that are most helpful. I know that you posted on the Forum, but we have a lot of posts on here that could help you in development and research. Maker.io might be very useful in finding references for prototype ideas.
As for the FDC1004, I don’t recommend reading just the datasheet. I’d recommend looking into its development boards too:
Sensor Evaluation Boards | Development Boards, Kits, Programmers | DigiKey
Finding specific layouts of integrated circuits is usually quite difficult. I always recommend looking into the manufacturer’s site for more in-depth Engineering content too.
You’d have to figure out reference measurements. That kind of depends on the sensor you use. As for calibration, that is a little more complicated. You have to determine how to adjust your sensitivity with whatever sensor you find. Most of that involves system development first. Calibration could be analog or digital. Sensor datasheets usually give parameters on operation where you can figure out what is “adjustable” and what isn’t. If a sensor has a given output that is unchangeable, you may have to include a way to vary the output to your liking. For resistive systems, for example, one common method is to use a Wheatstone bridge. Here is a good Texas Instruments documentation on bridge measurements for example:
A Basic Guide to Bridge Measurements (Rev. A)
You’d have to find other sources of course since yours is capacitave.
Thank you, Kaleb. I really appreciate the detailed guidance and the resources you shared.
I agree that the sensor geometry, shielding, and calibration will probably be the most challenging parts of the project. I’ll start by studying the FDC1004 evaluation boards and TI reference designs instead of relying only on the datasheet.
My first goal will be to build a simple prototype of the bean holder and electrodes, then test repeatability with multiple coffee beans and compare the readings with a trusted moisture meter. After that, I can determine the best calibration method before designing the final PCB.
Thanks again for pointing me toward Maker.io and the bridge measurement documentation. This gives me a much clearer direction for the next stage of Sikwal BeanSense.