time-series collecting spectra

[dpmchugh]

Hi all,

I am finding that python-seabreeze is a wonderful tool for connecting my Flame spectrometer to my Raspberry Pi 4 (thanks ap--)! This might be a basic question, but is there a way to collect spectra continuously at a set time interval? Ideally, I'd like to take a spectral measurement every five minutes and store the date/time/spectral msmt in a .csv or .txt format. I'm sure it is pretty straight-forward process to do so with a couple lines of code, but I am incapable of doing so with my limited experience. Any help would be greatly appreciated!!

Thanks - DPM

[ap--]

Hi @dpmchugh

Do you mean collect spectra with 5 minute integration time? Or collect a short integration time spectrum every 5 minutes?

the first is possible if your spectrometer supports it. check spec.integration_time_micros_limits to get the supported range in microseconds.

The second depends a lot on your time accuracy requirements.
A simple time.sleep(300 - integration_time) between measurements can do the trick if you dont mind that you accumulate relative time interval error.
Otherwise write a small function that reads the spectrum and use Python's threading.Timer to call it every 5 minutes.

You could also write a small script that opens the spectrometer and appends the measurement to your csv, and call that script via a cron job on the pi. Then it'll also auto start on boot, which might be good if you run the pi headless.

Let me know if you have more questions or if anything is unclear,
Cheers,
Andreas 😃

[dpmchugh]

Thanks Andreas! I am indeed looking to repeatedly collect a short integration time spectrum for a given frequency (eg. every minute). Would you happen to have an example, or know of a resource where someone has done something similar? Thanks again :)

[ap--]

For completeness, here's a minimal example of how to store spectra in a csv file every minute:

import csv
import time
from itertools import chain, count, islice

from seabreeze.spectrometers import Spectrometer


def store_to_csv(csv_prefix, integration_time_us, interval_s, max_samples=None):
    """write spectra to a csv"""

    assert interval_s > integration_time_us / 1_000_000
    spec = Spectrometer.from_first_available()
    spec.integration_time_micros(integration_time_us)

    fn = f"{csv_prefix}_it{integration_time_us}us_{time.strftime('%Y%m%d-%H%M%S')}.csv"
    with open(fn, mode='x') as f:
        writer = csv.writer(f)
        # header
        writer.writerow(chain(['time'], (f"{w:0.2f}nm" for w in spec.wavelengths())))

        t0 = time.monotonic()
        for _ in islice(count(), max_samples):
            writer.writerow(chain([time.time()], spec.intensities()))
            dt = interval_s - ((time.monotonic() - t0) % interval_s)
            time.sleep(dt)


if __name__ == "__main__":
    store_to_csv("spectra", integration_time_us=10000, interval_s=60, max_samples=10)

Cheers,
Andreas 😃

[mhagdorn]

Hi DPM, we have a multispectral system based on raspberry pis and ocean optics instruments that does this sort of thing. Have a look at https://blogs.ed.ac.uk/teampiccolo/ all the code is on github https://github.com/TeamPiccolo Regards magnus

…

On Thu, 2021-06-17 at 13:43 -0700, dpmchugh wrote: Hi all, I am finding that python-seabreeze is a wonderful tool for connecting my Flame spectrometer to my Raspberry Pi 4 (thanks ap--)! This might be a basic question, but is there a way to collect spectra continuously at a set time interval? Ideally, I'd like to take a spectral measurement every five minutes and store the date/time/spectral msmt in a .csv or .txt format. I'm sure it is pretty straight-forward process to do so with a couple lines of code, but I am incapable of doing so with my limited experience. Any help would be greatly appreciated!! Thanks - DPM

[ap--]

Very cool project @mhagdorn!
Are there step by step instructions on how to setup such a system for a raspberry pi that could be linked to from the python-seabreeze docs?

[mhagdorn]

I suspect the biggest problem is that the project relies on extra hardware to control the shutters and the shutters and foreoptics themself. We are planning to write a paper in the autumn to describes the system in detail. We'd love a link to our project (probably https://blogs.ed.ac.uk/teampiccolo/) somewhere and are happy to discuss any details.