Help needed in designing a circuit for photodiode.

[3DSS]

I’m working on a circuit to detect fire, using Photonic Detectors INC. photodiode type PDU-S104. I’m not very familiar with photodiodes and how to design a working circuit. My question is what type of amplifier should I use and what would an ideal circuit look like? If more information is needed pleas let me know.
This link is the data sheet for the PDU-S104.

http://www.photonicdetectors.com/pdf/pdus104.pdf

 

[IRstuff]

Much of this depends on whether you're actually trying to detect fire as opposed to detecting smoke.

Also, if you are really trying to detect fire, what is your discrimination criteria, i.e., how do you tell fire from not fire? TTFN

 

[3DSS]

I’m trying to detect fire and not smoke.
I’m trying to build a fire suppression system that will locate a fire using three of these sensors. These sensors are connected to comparators that operate motors for controlling the X and Y movement of a nozzle to guide it to the fire. Once this is done it will then trigger a timer and open a valve to distinguish the fire.
It is my understanding that these sensors can detect flame due to the wavelength they use, which is 200nm to 400nm. Other than that I don’t know enough about photodiodes to make any discrimination criteria.
I have already designed and tested most of the circuit, which includes the arming of the sensors, the motor control, turning the valve on, and the time for which the valve is on.
My intentions for the sensors is to use them in a way that gives me a varying voltage, which then would be put into a comparator. The output of the comparator would have three states A>B, A=B, and A<B. This information is then used to control the movement of the motors.

 

[IRstuff]

I'm afraid your sources are highly mistaken. Unless, you are lucky enough to have a smoke-free fire, the diodes won't see squat. The tranmissivity of smoke in the 200-400nm is pitifully low, that's why long-wave 8-12um infrared sensors are preferred for fire-fighting.

Moreover, because of the smoke, the position determination can be rather dubious, as the smoke billows and changes, your calculated fire position will fluctuate wildly. TTFN

 

[3DSS]

If I use the 8-12um infrared sensors what should I use and how would I wire it. I also understand about the difficulty of trying to focus on the fire, and plane to work on this after I get the sensors to detect a flame. Or would I be better off in trying to detect the heat from the fire, and if so how would this be done?
IRstuff I sure appreciate your insight and help on this.

 

[IRstuff]

Consider that a room that's say, 20'x20' and that a small fire would be maybe a sq. ft, means that you need to resolve 1/20th of the room. Given that many fires smolder before burning in earnest and the resolution requirement makes this challenging at best.

There also also the usual difficulties with false alarms and discrimination aganist expected things, such as people and smokers, as well as what do you do with the kitchen?

That's why standard fire control is done with straight-forward heat or smoke detection and broad area coverage with sprinklers. Consider that once your fire gets going and spreads, how would a single nozzle system deal with a spreading fire?

Standard smoke alarms are relatively simple conceptually, simply a light beam that's interrupted when smoke particles are present. BUT, they use a radioactive source to ionize the particles and there's some engineering to do to get the thresholds right. TTFN

 

[47gandalf]

Not to pic nits IRstuff, but 1 sq-ft is 1/400 of a 20'x20' room. Or much worse than your original estimate of 1/20.

 

[svberend]

In general, the photodiode can be used in one-of-two modes: photovoltaic or photoconductive. In photovoltaic mode, the photodiode creates a small voltage like a solar cell that you can measure with a DAC (or DMM) across a resistor. In the other mode, the diode is reverse biased and a small current flows (leaks?) in relation to the amount of light striking the diode. In this mode, you need a current-to-voltage converter that can be created with a simple op-amp to convert the current into a voltage that can be measured with a DAC or DMM. Each mode has trade-offs such as speed vs sensitivity. For more information, search Google for &quot;photodiode Photovoltaic photoconductive circuit&quot;.

Here is a good start with some theory:

http://www.siliconsensors.com/AN.html

 

[canyoncruz]

Not an EE, But been working in pfotonics R&D primarily optical and packaging design. Remember, there is no such yhing as negative light. so you will have to run the PD at a bias. you will have to determine the PD's quiescent point with the particular op amp such that you achieve the highest sensitivity / low noise from the detector. Next you will beed to find a high NA lens, sort of a &quot;photon Vacuume&quot; in order to gather light from a wide area. It should be placed such that it focal point is at a virtual point as to cover the the whole detector area at the furthest feild of view angle. After this, you will need a optical pass band filter. with our it your you PD might saturate will ambiant light, this filter should be at the wavelength of the flames, or thermal energy.

for the lenes i would use a polycarbonate one used for CD transport mechanisms. This can be purchased from Edmund Scientific along with the optical filter. For the PD, Talk to the guys at Fermionics in Simi Valley, CA. let me know if this helps!

 

[ScottyUK]

Flame detectors using emitted light are commonplace in burner management systems. They use a combination of wavelength and flicker pattern to establish whether the source is flame, or sunlight, or whatever.

IRStuff is correct - these detectors are good at seeing fires which have low smoke evolution such as saturated hydrocarbons (the ones ending in ...ane), and alcohols for example.

A couple of manufacturers to have a look at are: Forney Fireye; Detronics. No doubt there are others.

 

[Chuckles159]

I have a patent on a flame sensor using a photo pickup. I used a photo transistor but what worked for me was to AC couple the signal from the photodevice as the flickering is what distinguished the flame from ambient light. As well, I had to put a feedback sensor on the photodevice to keep the DC bias at a fixed voltage. Otherwise, ambient light variations kept moving the DC operaing point from one extreme to another. However, the fixed bias on the photo transistor had to have a high output impedance, about 2K ohms. Otherwise, the regulator would swamp out the AC signal from the flickering. Use a device with as much sensitivity in the IR range. UV sensitivity is meaningless.