AS7265x sensing for Localized Surface Plasmon Resonance (LSPR)

[Dong3867]

Hi everyone,

I am currently using the SparkFun AS7265x Spectral Sensor to develop a portable biosensor based on Localized Surface Plasmon Resonance (LSPR). My target LSPR peak typically falls between 500 nm and 700 nm.

Background & My Signal Processing Strategy:

• Optical Path: Full-spectrum Light Source ➔ LSPR sample chip ➔ Convex Lens (for light collection) ➔ Ground plastic diffuser ➔ AS7265x sensor.

• Signal Processing: To overcome the resolution limit of the 18 channels, my signal processing strategy involves using high-resolution spectrometer data as a reference template.

I am currently facing two major obstacles and would deeply appreciate your advice:

1. Abnormally low signal intensity on the 680 nm channel

Even though I am using a diffuser to scatter the collected light, my raw data consistently shows that the 680 nm channel is significantly lower than the surrounding wavelengths. This creates a severe discontinuity in the spectrum. This missing/low data directly causes massive errors during my Template Matching process, making it more impossible to accurately calculate the peak shift. I understand that the three chips (AS72651, 2, 3) are physically separated in a triangular layout on the PCB. Could this mean the chip handling the 680 nm channel is falling into an optical “shadow” or blind spot?

2. Optical design recommendations for the tri-chip layout

Currently, convex lens creates a circular spot that roughly covers the entire sensor area. However, the energy distribution across the different chips still seems highly uneven (possibly worsened by chromatic aberration from the lens). Given the physically separated nature of the AS7265x chips, does anyone have recommendations for an optimal optical architecture? Are there specific types of light pipes, collimators, or ideal distances for the diffuser that can ensure a single beam of light is distributed perfectly evenly across all three ICs?

Thank you in advance for your insights and support!

[jremington]

@Dong3867

Full-spectrum Light Source

What does that mean? The diagram shows an LED source.

Please post the light source emission spectrum.

[Dong3867]

The led covers roughly 420~720 nm spectrum range.

And this LED has a 120-degree emission angle. (That’s why I applied a convex lens)

[jremington]

@Dong3867

my raw data consistently shows that the 680 nm channel is significantly lower than the surrounding wavelengths

That is expected. A glance at the light source emission spectrum you just posted suggests that the intensity in the 680 nm channel is around 60% of the intensity at shorter wavelengths.

The usual solution is to normalize the data to those obtained from a suitable blank sample.

My target LSPR peak typically falls between 500 nm and 700 nm.

Then that light source is not a good choice, based on the very limited amount of information you have provided.

[Dong3867]

Sorry for not making my premise clear earlier! To clarify: I measured the pure LED with an spectrometer (Ocean Optics USB2000+ ), and its spectrum looks perfectly normal. However, capturing that exact same light through my AS7265x setup severely distorts the spectral shape.

This discrepancy is why I suspect the flaw lies within my optical path, rather than the light source itself. I’ve also tried adjusting the LED intensity and increasing the sensor’s exposure time. While this increases the absolute overall intensity, the distorted spectral shape remains exactly the same.

I have also attached a photo of my current optical setup for your reference.

[jremington]

Yes, this statement:

the 680 nm channel is significantly lower than the surrounding wavelengths.

is very misleading. Based on the above spectra, I would say that the readings from most of the AS7265x channels are unlikely to be very useful.

To test your hypothesis that “the energy distribution across the different chips still seems highly uneven”, experiment with slightly different positions/orientations of the sensor module.

The detector may not be useful for this project, period. Obviously, the Ocean Optics spec is a better choice.