First Test of Calibration Plate

A project log for Compact, $25 spectrometer

AMS's new AS7265X 3-chip set promises a compact, 18-channel, 20 nm FWMH spectrometer for less than $25

Kris WinerKris Winer 05/14/2018 at 03:192 Comments

May 13, 2018

I received the calibration plate pcbs on Friday and put one of them together. I could see visually the light from the 400 nm, 470 nm, 555 nm, 615 nm and 700 nm leds; the 740 nm, 850 nm, and 940 nm are invisible to the human eye.  I had 333 and 100 ohm 0603 resistors sitting around so I chose to use 330 ohm resistors for current limiting for all eight of the leds. The 400 nm, 470 nm, and 615 nm leds were quite bright (to my eye) while the 555 nm led was not; next time I will use a 100 ohm resistor for it. I am using ~4 V from a LiPo battery as the power source and jumper pins to choose which of the leds to enable like this:

Here the 400 nm led would be on if the battery was installed.

Unfortunately, after I solderd the headers on to the board the 615 nm led became quite faint; obviously somehow the soldering affected it but I don't know how.

Also, my preliminary tests with the DemoBoard spectrometer lead me to believe that both the 740 nm and 850 nm leds are simply not working. Again, I don't know why. So surprisingly, making a calibration plate with eight narrow-band led sources turned out to be harder than it would seem. I have two more pcbs so I will try again.

Here is another "problem" with the leds that do work. They seem to output different amounts of light. Now, I don't require that the radiant fluence be identical for the kind of tests I have in mind, just that the light is bright enough that the spectrometer can get a strong enough signal to register the light. Here is what four of the visible leds look like:

400 nm

555 nm

615 nm

715 nm

The 615 nm is particularly disappointing since this would have been a great test of the AS72651 sensor, but it is so faint here that it just doesn't register at all.

I created a sophisticated measuring apparatus using a Mark 1 toilet paper roll at the bottom of which I placed the calibration plate and placed the DemoBoard on the top approximately 130 mm away. The toilet paper roll provides some isolation from external light sources and a fixed distance between sensors and sources but no way to manage alignment. The latter attribute is fine, since one of the questions I want to answer is what kind of results can be expected from just using the 3-chip solution as a single (hand-held) spectrometer.

DemoBoard sits on top with sensors looking down.

This is a question raised in the comments. The gist of the concern is that the 18-channel spectrometer is really composed of three spatially-separated six channel spectrometers and that obtaining sensible results might require special care in alignment of the sensors or a very large sample size or even additional optical elements. My hope was to simply show that, at least at the separation and diameter of a toilet paper role, I could record eight signals from a small area source without any special alignment or other kind of fussing.

While the flaws of the calibration plate make this first test non-definitive, I would say there are a few things to learn still. Here is the spectra I recorded from eight single samples with one each of the eight leds enabled using the default AS7265X DemoBoard conditions (165 ms integration time, gain setting = 2):

I see four peaks, one at 435 nm for the 400 nm source led, one at 460 nm for the 470 nm source led, one at 705 nm for the 700 nm source led, and one at 940 nm for the 940 nm led. The intensities (counts) are rather low. Disappointingly, I see nothing from the 555 nm nor 615 nm leds even though I can see their light. And it looks like the 740 nm and 850 nm leds are simply not functioning.

The good news is that of the four "peaks" I do measure, half come from the AS72652 and half come from the AS72653. So lack of alignment doesn't prevent light to be registered on these two spatially-separated sensors from a small source. It almost seems like the AS72651 is not working, or that the 740 and 850 nm leds are not working or both.

I want to repeat this test with at least the 555 nm led limited by a 100 ohm resistor and I hope that another try fixes the 615 nm led, which was glowing brightly before I soldered on the headers. I also want to repeat it using my own sensor board once I am sure I have the design right (I expect to hear back from AMS this week with answers to my several questions about the reference design). So, much more testing to be done just to answer the basic question "will the spectrometer even work as intended"!


Kris Winer wrote 05/14/2018 at 03:40 point

Thanks Ted. These might be useful when I get to the reflection spectroscopy part of the project (about a microsecond after I verify the reference design is working!). On the current design I have a bright 5700 K 90 CRI source led so having either color patches to reflect well defined colors or color filters, as you suggest, would be very useful to test the spectrometer. Thanks for the link!

  Are you sure? yes | no

Ted Yapo wrote 05/14/2018 at 03:29 point

Another possibility for go-no-go tests and maybe rough calibrations are theatrical gel filters.  You can get a swatchbook of maybe a hundred different color filters for around $5 from amazon.  The typical filter curves can be found on-line. I've used them for various photographic and display filtering projects over the years.

If you have a local shop that sells theatrical lighting, you can also score these sample books for free.

  Are you sure? yes | no