Photodiode for light detection

[Telemachus]

Hi. I am willing to make an experiment with light. The idea is to have a light source (specifically a led light) and a photodiode to detect the light and make some spectroscopy. The photodiode has to detect the light in the same frequency that is emitted by the led.

I'm not an expert in electronics, so I don't really know where to start. The light led just has to be turned on at constant intensity all the time. But I would like to also have that intensity measured at the emission point to compare with the intensity detected by the diode at the other side of some material to be characterized. I had a lab of electronics lots of years ago, where we used protoboards and all that stuff, but I don't really know how to do this thing, how to build the circuits, and how to the detect the current induced by the diode and take it to a computer. My only contact with electricity was in that lab. I would also like to have the less noise to signal ratio as possible.

Any recommendations, ideas, or sources to get started?

Thanks.

 

[Charles Link]

Suggestion is to use a silicon photodiode=it works for visible light, and use this circuit with an op-amp. https://electronics.stackexchange.com/questions/36086/does-this-photodiode-circuit-work Pin 2 is the inverting(negative) input, pin 3 is the non-inverting input, and pin 6 is the output. You can vary the feedback resistor ($R_f=100 \, k \Omega$ usually works well) depending on how much gain you want, and a capacitor of about C=10 pF in parallel with the feedback resistor can decrease high frequency noise. The only other thing you need is +12 volts on one bias voltage pin (I think it's pin 7), and -12 volts on the other bias voltage pin (pin 4 I believe.)

 

[Telemachus]

Thank you verymuch. And how could I convert the electric signal from the photodiode in data in a computer to relate it to the intensity of light detected? I mean, the electronics, what kind of acquisition system do I need? I know it's basic stuff, but I'm an ignorant of it.

 

[Charles Link]

A simple voltmeter (e.g. a Radio Shack meter) to read the DC voltage between pin 6 (output) and pin 3(ground) is all you need. It's important to tie the ground (pin 3) to the grounds of both bias (+12,-12) power supplies. Your output voltage will be between -12 volts and +12 volts. The photodiode can be connected with either polarity (either direction), and that will determine whether your output signals are positive or negative.

 

[Charles Link]

Incidentally, the response of this circuit is quite linear, in that for a given wavelength, the output voltage is very nearly proportional to the incident power of the source.

 

[Telemachus]

Great, of course I would need to relate how the current induced in the photodiode is related to the power of incident light. Anyway, I would like the data to be registered automatically in a computer, for ease of use, to make different measurements and analyze the data posterior to that. I imagine there exists some kind of system that admits this instead of me just writing with a pencil the current I measure?

 

[Charles Link]

There are A/D cards, i.e. analog to digital converters that could be useful. I am not well-versed in the latest A/D cards that are currently used. $\\$ Meanwhile, the response of a silicon photodiode at $\lambda=1000 \, nm$ is normally about $r=.65 \, amp/watt=I_p/P_{in}$ , and the output voltage you get is $V_{out}=I_p R_f$. The response for wavelengths shorter than $1000$ nm normally goes approximately as $r(\lambda)=.65 (\lambda/1000 \, nm)$. (This is because the photodiode actually counts photons, with approximately one electron per photon). $\\$ One additional item: Even though the response is e.g. .65 amps/watt, this will usually be like .65 mA/mWatt. (Typical incident power levels are milliwatt or lower. In fact, it depends on detector area, but you normally don't want to exceed $\approx$ 1 mwatt incident on the detector.)

 

[Telemachus]

Thank you very much.

 

[Charles Link]

Please be sure and give us an update on your circuit once you get it working. :) :)

 

[Telemachus]

I think there is a problem with this. I need the photodiode to detect light only in a small band, around 780nm. From what I've read, it seems that silicon photodiodes have a very large wavelength spectrum, and I would need it to resemble a Dirac delta function as much as possible. The idea is to make some spectroscopy, so I need it to be sensitive to a specific wavelength.

 

[Charles Link]

I think there is a problem with this. I need the photodiode to detect light only in a small band, around 780nm. From what I've read, it seems that silicon photodiodes have a very large wavelength spectrum, and I would need it to resemble a Dirac delta function as much as possible. The idea is to make some spectroscopy, so I need it to be sensitive to a specific wavelength.

You can get a narrow band optical filter at a selected wavelength. They tend to be somewhat expensive=a 1" diameter precision filter might cost $100 or more.

 

[Charles Link]

OCLI (Optical Coating Laboratory Inc.) is somewhat well-known for their optical filters.

 

[Tom.G]

OCLI (Optical Coating Laboratory Inc.) is somewhat well-known for their optical filters.

A search for OCLI shows they were bought out in 1999. Their new owner, JDS Uniphase, split into two companies in 2015... etc.
Long story short, http://www.OCLI.com redirects to current owner VIAVI Link.

 

[Charles Link]

A search for OCLI shows they were bought out in 1999. Their new owner, JDS Uniphase, split into two companies in 2015... etc.
Long story short, http://www.OCLI.com redirects to current owner VIAVI Link.

@Tom.G I'm presently retired and not completely up-to-date. I would suggest this company as another possibility for optical filters: https://www.thorlabs.com/navigation...MImfud6-WF1gIVzEsNCh3QEgbLEAAYAiAAEgIyN_D_BwE

 

[Tom.G]

@Tom.G I'm presently retired and not completely up-to-date. I would suggest this company as another possibility for optical filters: https://www.thorlabs.com/navigation...MImfud6-WF1gIVzEsNCh3QEgbLEAAYAiAAEgIyN_D_BwE

That makes two of us. I had completely forgotten about ThorLabs. They are one of the majors and even list a stock 780nm filter with ±2nm bandwidth for a bit under your estimated price. Good call.

 

[Charles Link]

That makes two of us. I had completely forgotten about ThorLabs. They are one of the majors and even list a stock 780nm filter with ±2nm bandwidth for a bit under your estimated price. Good call.

One item of interest for the OP if he chooses to use an optical filter such as this to block stray light is that these are normally interference type filters (works by Fabry-Perot effect), so that the bandpass shifts slightly in wavelength as you go off-axis for the incident angle. I believe the pass wavelength becomes slightly shorter off-axis, so you need to choose the pass band wide enough if the application requires off-axis viewing.

 

[berkeman]

I think there is a problem with this. I need the photodiode to detect light only in a small band, around 780nm.

Then put your circuit in a dark box. If only the emitting LED and the receiving photodiode are in the dark box, then there is no interference from other light sources...

 

[Telemachus]

That makes two of us. I had completely forgotten about ThorLabs. They are one of the majors and even list a stock 780nm filter with ±2nm bandwidth for a bit under your estimated price. Good call.

That sounds pretty much like what I need!

 

[Telemachus]

Then put your circuit in a dark box. If only the emitting LED and the receiving photodiode are in the dark box, then there is no interference from other light sources...

Yes, I thought of this as a possibility, but what about thermal radiation? of course, if the detector is not sensitive to it, there would be no problem, but I don't know how many photons I would be counting from thermal radiation if let's say, the detector goes up from the visible to the far infrared.

 

[Charles Link]

Yes, I thought of this as a possibility, but what about thermal radiation? of course, if the detector is not sensitive to it, there would be no problem, but I don't know how many photons I would be counting from thermal radiation if let's say, the detector goes up from the visible to the far infrared.

Thermal radiation is not a problem with a silicon photodiode. Silicon cuts off at $\lambda=1000 \, nm$,(which is the very near infrared). At room temperature $T=295 \, K$, the blackbody curve for thermal radiation peaks at about 10,000 nm,(by Wien's law, $\lambda_{max} T=2.898 \, E+6 \, nm \, deg \, K$), and there is virtually nothing at $\lambda= 1000 \, nm$ or shorter wavelengths from thermal sources where silicon responds. $\\$ You might be able to save yourself the cost of an expensive optical filter if you don't have any stray light sources that need to be blocked.

 

[berkeman]

Yes, I thought of this as a possibility, but what about thermal radiation? of course, if the detector is not sensitive to it, there would be no problem, but I don't know how many photons I would be counting from thermal radiation if let's say, the detector goes up from the visible to the far infrared.

That's a good question for you to research as part of your project (and for general learning and information). What does the black body radiation curve look like for a room temperature test chamber? (Hint -- be sure to use Kelvin temperature units consistently in you check)

 

[Charles Link]

@Telemachus Here's a "link" you might find useful. https://astrogeology.usgs.gov/tools/thermal-radiance-calculator/

 

[Telemachus]

Thank you all, you've been really helpful. When I got things done I'll post a picture of it. Probably at this point it is better to skip the filter and get the photodiode isolated from other light sources. But getting a filter could be a good idea to try in the future, perhaps to enhance the signal to noise ratio. Could I achieve a resolution of ±2nm with a diffraction grating?

 

[Charles Link]

Thank you all, you've been really helpful. When I got things done I'll post a picture of it. Probably at this point it is better to skip the filter and get the photodiode isolated from other light sources. But getting a filter could be a good idea to try in the future, perhaps to enhance the signal to noise ratio. Could I achieve a resolution of ±2nm with a diffraction grating?

Quite easily. One thing you need to know about diffraction grating spectroscopy though is the optical layout. It requires focusing the source onto a narrow slit. The slit is placed at the focal point of a concave mirror that collimates the light from the slit=making parallel rays, and these are incident on the diffraction grating, which may be 2" across. The spectrum comes off of the diffraction grating=the different angles correspond to different wavelengths, and it essentially needs to be observed in the very far field in order to get a single wavelength and not a mixture of wavelengths. An optical trick is used to get the far field pattern at a short distance: A concave mirror is again used, this time to focus the far field pattern, and parallel rays at a given angle appear focused at a given position in the focal plane. A narrow slit is used at this focal plane to isolate a single wavelength. (The focused image of a monochromatic source that starts at the entrance slit is shaped like the entrance slit. It is essentially the image of the entrance slit, but it is the diffraction maximum, usually at m=1 from the grating. (You get several orders of diffraction maxima, $m \lambda=d sin(\theta_m)$, m=-5,-4, -3,-2, -1, 0,1, 2, 3, 4, 5, etc., including the m=0 (all wavelengths come together at this one), but most often, the m=1 maximum is used in the spectrometer). Thereby a diffraction grating spectrometer has two slits=one at the entrance and one at the exit, and they are normally chosen to be equal widths for best results. 2 nm resolution can be readily achieved with slits about 1mm wide. $\\$ An optical interference filter is really much easier to implement than using a commercially made (diffraction grating) spectrometer and/or constructing a (diffraction grating) spectrometer.

 

[Telemachus]

Great, thank you!

 

[Telemachus]

Another question, regarding the light source (which will be a IR LED), can I regulate the intensity of the emitted light by controlling the voltage? I mean, a typical circuit for the LED would be a 9V battery, with some resistor in series with the LED. If I used instead of 9V a higer voltage, that would increase the current through the LED, and in that way would I get a stronger light signal?

 

[Charles Link]

You can control the intensity of an LED with voltage, but you need to be careful not to burn it out. (They are generally inexpensive in any case.) The polarity is of course important, and you run the diode in the forward direction.

 

[Telemachus]

How much voltage and what resistance should I use in turn to get the maximum signal without burning the LED? does this depend on the wavelength of light emission in the LED or LED specifications? sorry for asking so many things, I haven't touched a circuit in years since my classes at the undergrad lab.

 

[Charles Link]

I bought several visible LED's at Radio Shack or a similar store a few years ago, and they told me to run them at about 3 volts. Suggest you look at the manufacturer's specifications though. The operating voltage range may vary from one LED to another.

 

[berkeman]

can I regulate the intensity of the emitted light by controlling the voltage?

The output intensity depends on the LED current. You use that series resistor in the circuit with the voltage source to set the LED forward current. You use the LED's Vf forward voltage drop as a function of current (specified in the datasheet) and the output intensity as a function of forward current (also in the datasheet) to determine what series voltage dropping resistor to use.