Luxmeter modules comparison for DIY

[bnxvs]

BH1750 (GY-30)
Max44009 (GY-49)
OPT3001 (CJMCU-3001)

Scale Y in EV numbers.
EV (100) = sqrt(lux / 2.5)

Max44009 (GY-49) showed the most unstable results, but it is more sensitive in low light than the BH1750.
The OPT3001 was the most sensitive.

 

[bernard_L]

Don't understand much in that plot. Are you expecting comments?

Scale Y in EV numbers. EV (100) = sqrt(lux / 2.5)​

EV is a log scale. So sqrt function does not belong here. Plus, EV is a combination of shutter speed and f-stop, with EV0 for f:1 and 1sec. What measures light is LV. We have LV=EV only for ISO100 film.

Why should the various sensors' sensitivity be measured in EV or LV?? I suppose that, to compare them, they are all subjected to the same illumination, and that sensitivity is judged from the electrical signal (current or voltage).

What is represented by the horizontal axis? What happens between the "high" and "low" steps?

 

[bnxvs]

Why should the various sensors' sensitivity be measured in EV or LV??

Just for convenience. In the "lux", small peaks of volatility are not visible due to too much data difference. And also in order to make it convenient to compare the readings with the results of measurements with a Sekonic L-308 (the closest results - OPT3001, at level 0.1-0.3 ev).
Of course, all sensors were tested simultaneously, under the same lighting and power conditions. "X" axis shows time. Data peaks means turning on and off an additional lighting.

 

[ic-racer]

Is this to test their suitability for flash photography exposure measurement? What are the units on the x-axis? Milliseconds?

 

[bnxvs]

Is this to test their suitability for flash photography exposure measurement? What are the units on the x-axis? Milliseconds?

I don’t think it will work like that. Unfortunately, the response time of such sensors is too long to respond to the speedlight impulse.
I use these sensors to create densitometers, exposure meters for a darkroom, etc., i.e. in places where excessive performance is not required, but the accuracy of readings in low light is important.

 

[radiant]

@bnxvs I've used TSL2591 with good success on darkroom light meter and on a densitometer. It has enough accuracy for both, but not in excessive amounts.

 

[bnxvs]

@bnxvs I've used TSL2591 with good success on darkroom light meter and on a densitometer. It has enough accuracy for both, but not in excessive amounts.

Here is a link to an article on testing TSL2591 versus BH1750FVI and MAX44009. Unfortunately in Russian, but I think google will translate it quite tolerably.
https://ab-log.ru/smart-house/ethernet/megad-tsl2591

After reading this article, I bought MAX44009 ... But now I decided to try OPT3001, because MAX seemed to me not too stable in low-light mode. And since I needed to fairly accurately calibrate a homemade sensitometer, accuracy was a priority.

 

[bernard_L]

I took a look at the TSL2591 on the Adafruit website. Lots of information for hooking up to a microcontroller, programing, but zero information on technical specs, not even the name of the actual sensor sitting in the middle of the small PCB. Then I read this sentence:

The best part of this sensor is that it contains both infrared and full spectrum diodes! That means you can separately measure infrared, full-spectrum or human-visible light.​

Which reminded me of the datasheet for a sensor that I came across for the same application. Actually, there are two diodes: one sensitive to visual and IR, another one sensitive to IR only (filtered). The "visual" channel of interest for darkroom meter (blue/green would be better) is actually obtained by subtracting the signal of the IR sensor from that of the Vis+IR sensor. In my book this is a definite No-No (accuracy, stability). Speaking after 30+ years designing, building, testing and using instrumentation for astronomy. Subtraction of signal is allowed only from the same sensor, same instrument configuration.

This might be usable. Operate in zero-bias mode, with an op-amp to convert current to voltage; and need to read the analog vltage using one of the analog inputs of the microcontroller.
https://fr.farnell.com/osram-opto-semiconductors/sfh-2270r/photodiode-560nm-cms/dp/2981701

This one is no good: some unwanted sensitivity in the IR.
https://fr.farnell.com/vishay/temd6200fx01/capteur-de-lumiere-ambiante-0805/dp/2504146

 

[radiant]

@bernard_L it is not adafruit product, here is the product itself https://ams.com/tsl25911

"Highest sensitivity to 188µLux
Patented dual-diode architecture
600M:1 dynamic range"

Datasheet here https://ams.com/documents/20143/36005/TSL2591_DS000338_6-00.pdf/090eb50d-bb18-5b45-4938-9b3672f86b80

 

[Fraunhofer]

I think the blips you see at the beginning of each pulse are entirely driven by the transient response to fast changing signals. So I am not sure this test is doing what you want.

I used a TSL235R from Taos to build a UV sensitometer. This is a single diode light to frequency convert with logarithmic output. Per data sheet covers 6 orders of magnitude of irradiance. Peak sensitivity is around 750nm. Linearity is really quite good as tested against a Stouffer wedge.

For a sensitometer you always can overcome low light levels with a stronger light source.

 

[bernard_L]

@bernard_L it is not adafruit product, here is the product itself https://ams.com/tsl25911

@vedostuu, thank you for providing the link to the component datasheet; actually there is an Adafruit minishield with the exact same reference
https://www.adafruit.com/product/1980
Anyway, look at Fig.11 on the datasheet. Assume you want to use the IR sensor to cancel the IR response of the broadband sensor; I do rough eyeball estimates. At 900nm, the response of the IR channel is approx. 70% of the broadband one; so we need to subtract from the broadband signal 1/0.7, approx 1.4 of the IR signal; now look at 700nm; 1.4x the IR response is 1.4x0.35 approx. 0.5, while the response of the broadband at 700m is 0.9; which leaves 0.45 uncorrected response at 700m (boundary between red and IR), almost as large as the response at 500m, and much larger than the response at 450m (blue channel for multigrade papers).

Now, if you intend to measure only b/w films, that is probably not a problem, since the transmission of these films is almost independent of wavelength, as long as you do not use a staining developer.

Good luck with your project.

 

[radiant]

Assume you want to use the IR sensor to cancel the IR response of the broadband sensor; I do rough eyeball estimates. At 900nm, the response of the IR channel is approx. 70% of the broadband one; so we need to subtract from the broadband signal 1/0.7, approx 1.4 of the IR signal; now look at 700nm; 1.4x the IR response is 1.4x0.35 approx. 0.5, while the response of the broadband at 700m is 0.9; which leaves 0.45 uncorrected response at 700m (boundary between red and IR), almost as large as the response at 500m, and much larger than the response at 450m (blue channel for multigrade papers).

That has been discussed here too: https://github.com/adafruit/Adafruit_TSL2591_Library/issues/14 and https://ams.com/documents/20143/360...2-00.pdf/ce3360f8-fb85-bc22-0bad-b60d3b31efc8 .. Of course one should know what range of the spectrum one is measuring to adjust the corrrection factors for that specific use.

 

[bnxvs]

For a sensitometer you always can overcome low light levels with a stronger light source.

A sensitometer isn't the same as a densitometer, at all. Rather, it is the opposite ))) This is a device for exposing test film samples. Calibration of the sensitometer, in my understanding, will consist in adjusting the exact level of illumination (in the range of 2 - 2,5 lux-s before 1-th field of optical guide) to obtain the appropriate values for each field.
Of course, to a certain level, the illumination can be adjusted by the duration of the light pulse, but in any case it will be in the range up to 1/20 - 1/25 of a second, to exclude the influence of the Schwarzschild effect (reciprocity failure). Therefore, it's possible to increase the illumination only within very limited limits.

 

[Fraunhofer]

A sensitometer isn't the same as a densitometer, at all. Rather, it is the opposite ))) This is a device for exposing test film samples. Calibration of the sensitometer, in my understanding, will consist in adjusting the exact level of illumination (in the range of 2 - 2,5 lux-s before 1-th field of optical guide) to obtain the appropriate values for each field.
Of course, to a certain level, the illumination can be adjusted by the duration of the light pulse, but in any case it will be in the range up to 1/20 - 1/25 of a second, to exclude the influence of the Schwarzschild effect (reciprocity failure). Therefore, it's possible to increase the illumination only within very limited limits.

You are absolutely right, I misspoke. I meant a densitometer, thanks for catching my mistake

 

[bnxvs]

You are absolutely right, I misspoke. I meant a densitometer, thanks for catching my mistake

No problems. These names are often confused ... I think that the "sensitometer" is not quite the correct name (this device does not measure anything), but it happened historically ...)))

 

[ic-racer]

Yes, the sensitometer measures things analogous to a "Mass Balance." So, by itself (without the calibrated step wedge reference) it can't measure anything.

 

[bnxvs]

I apologize, but my English is not very good - I didn't understand exactly what you had in mind?
But sensitometer doesn't measure anything, at all. This is just a device for accurate and reproducible exposure of a film sample. A very different design of such devices are known, and are not always based on an optical wedge (scale). For example, constructions based on the movement of a light source along a sample with a non-linear velocity are known. As well as slotted disk sensitometers.
Moreover, since the age of affordable electronics is now, it probably makes sense to talk about designs of sensitometers based on LEDs controlled by microcontrollers, incl. PWM signals. I think it is much more economical, simpler and more accurate than mechanical or analog devices.

 

[swdick]

I tried buidling a densitometer to measure paper densities (so for refelective values) using a 2591. It seemed to work except that I couldn't measure density values greater than about 1.5 while a "real" commeercial meter gave values up to 2.2. It might have to do with reflections of the paper (miroring the LED's) in stead of the the diffuse reflection of the paper
The lightsource (2 LED-s) projected under an angle of 45 degrees to avoid this problem.
Has anyone any experience or suggestions?

 

[bnxvs]

http://buildingadensitometer.blogspot.com/