Choosing between photodiodes and phototransistor designs for a amplification circuit.

Hello everyone,

We need to construct a very low cost circuit that need to sense a scene about 4-7 meters away the maximum amount of information available through photonic sensors at various bands (visible light, infrared, ultraviolet)

All of these photon sensors are available as photodiodes or phototransistors... Given that we need to amplify each of these with an OpAmp in logarithmic amplification and that we wish to keep that part of the circuit under $3 for 8 channels, do you think we should opt for a design based on photodiodes or phototransistors? (The circuit needs to be physically small)

Because of our cost constraint, my intuition tells me we’re better off with phototransistors in order to use the lower-cost OpAmps, but perhaps they greatly reduce the signal-to-noise ratio and we’re better off with photodiodes? (We need to amplify the signal for a standard ADC attached to a voltage rail of 1.8V and obtain the best possible dynamic range and signal-to-noise ratio)

Any recommendation on a decent type of OpAmp and the amplification topology to use would be greatly appreciated!

Thank you!

Jean-Pierre

A photo-transistor, as you likely know, is more sensitive, but slower than a photo-diode. Regarding noise and dynamic range, the specification sheets for the devices should help you there. In general, I would expect that a photo-diode with a low noise amp would provide better noise performance than a photo-transistor, but don't know for sure.

Hi Crutschow, thanks for taking the time.

As speed is not an issue (only need to take a reading every 1 to 15 seconds) I hope to obtain low-cost amps that can correctly amplify logarithmically the output of a photodiode so that 1.8V = a hot stove element 1 meter away.

Would you recommend I inverse the photodiode or connect it straight?

Also, do you think low-noise opamps suitable for photodiodes are less than $1.50?

Thanks!

JP

you could choose one of them
https://www.ti.com/paramsearch/docs...og&familyId=1479&uiTemplateId=NODE_STRY_PGE_T

If you are measuring light then you must use photodiodes. Phototransistor output depends on the transistor current gain which is variable and temperature dependant.

Keith

sabouras said:

you could choose one of them
https://www.ti.com/paramsearch/docs...og&familyId=1479&uiTemplateId=NODE_STRY_PGE_T

Click to expand...

Thank you very much Sabouras for this recommendation. TI makes good stuff (especially after swallowing National Semi!) and is certainly a key resource to research.

So if I understand correctly, the type of OpAmp we need is a "Low Noise OpAmp"... Do you think these are optimal for the logarithmic amplification we need for our low-cost BOM?

Thanks again!

- - - Updated - - -

keith1200rs said:

If you are measuring light then you must use photodiodes. Phototransistor output depends on the transistor current gain which is variable and temperature dependant.
Keith

Click to expand...

Thank you Keith for your input. So as I suspected phototransistors would kill the SNR?

In other words, is a photodiode connected to a "Low Noise OpAmp" with a logarithmic OpAmp amplification topology the recommended "best practice" for our given circuit requirements??

Thanks again!

the low noise opamp it guarantee that produces very low noise by itself by removing power supply noise , frequency noise, thermal noise etc. So i believe for small signals are the optimum choice. It does care about the topology you like to use it.

I don't know why you are talking about logamps. For accurate measurement of optical levels you should be using a transimpedance amplifier configuration. Normally I would reverse bias the photodiodes but for large area photodiodes where reverse leakage can be a problem, zero bias can be the only way to go.

For amplifiers, as you are concerned about cost, look for the lowest cost quad CMOS opamp. Offsets aren't a big issue in transimpedance configuration because they aren't amplified. The poor noise with a cheap opamp shouldn't be an issue because you can heavily filter it with your slow measurement time.

I actually designed a chip with 16 transimpedance amplifiers and a multiplexer built in (I think that was the number - I would have to check) many years ago but I don't know if is still available. If this is a volume application (>10k units) then contact me by PM and I can send you details.

Keith

Photo diodes are really low power generators they have op current in the range of about FEW MICRO AMPeres so you need additional circuits for amplification

but photo transistors are independently used as they have higher Current op but they are temperature dependant

don't forget that they both sense only visible light and not IR or UV so kindly if you need to sense IR then Thermopile is most suitable for your need

and about LOG AMPS why you need it any way ?

why you posted in #1 to sense UV IR AND LIGHT what is that you exactly want to sense ??

answer some of these would be very helpful ...

Most photodiodes and phototransistors are sensitive to near-infra-red. They usually peak around 900nm and drop to next to nothing above 1050nm. UV sensitivity is usually poor but UV enhanced photodiodes will boost that. That would put the cost up though. If you want to sense longer infra-red it starts to get expensive because you move away from silicon devices.

Keith

Hi Keith1200rs, many thanks for that very useful post.

As photons get reduced by the square of the distance I felt that a logamp would be the way to go, but I'm no expert in opamp design and I will certainly research "transimpendance amplifiers" with reverse bias of photodiode... do you think that OpAmp topology would yield superior results and sensitivity at greater distance?

I'll certainly be researching this opamp configuration.

However, a friend recommended I switch to using PIRs instead of photodiodes for my application. These are quite cheap now and given the range that modern motion sensors get maybe I'd be far better of with a PIR... what do you think?

Also... if I switch to a PIR, what do you think is the appropriate OpAmp configuration to amplify that?

Thanks many times for sharing your experience!!

Jean-Pierre

- - - Updated - - -

jeffrey samuel said:

Photo diodes are really low power generators they have op current in the range of about FEW MICRO AMPeres so you need additional circuits for amplification

but photo transistors are independently used as they have higher Current op but they are temperature dependant

don't forget that they both sense only visible light and not IR or UV so kindly if you need to sense IR then Thermopile is most suitable for your need

and about LOG AMPS why you need it any way ?

why you posted in #1 to sense UV IR AND LIGHT what is that you exactly want to sense ??

answer some of these would be very helpful ...

Click to expand...

Hi Jeffrey, many thanks for taking the time.

The scene we're trying to measure is how to tell the difference between a hot stove element 4 to 7 meters away from that same hot stove element with a pan on fire on top. (It is for a low-cost fire prevention device.) While the device feature a camera for flame recognition, we're hoping to assist this with a well chosen amplification of very-low-cost photon sensors in IR, visible light and UV.

There are many IR & UV sensors in photodiodes / phototransistors.

However we're thinking of switching the IR photodiode for a PIR... what do you think? (I think thermopile are too expensive around $1.25)

Q: Would a PIR with a well chosen low-cost quad OpAmp offer the best possible performance / cost for our app focused on BOM cost below $3 for that part of the circuit?

Many thanks!!

Jean-Pierre

- - - Updated - - -

keith1200rs said:

Most photodiodes and phototransistors are sensitive to near-infra-red. They usually peak around 900nm and drop to next to nothing above 1050nm. UV sensitivity is usually poor but UV enhanced photodiodes will boost that. That would put the cost up though. If you want to sense longer infra-red it starts to get expensive because you move away from silicon devices.

Click to expand...

Hi Keith, related to that, do you think the fact that the mass availability of motion sensor made the PIR the best performance / cost sensor for our app?

Jean-Pierre

Given what you have now said, I think you need to be looking at PIR sensors - you need to be sensing heat, so not near infra-red. It is a while since I designed a PIR chip, but as I seem to recall they are simply a voltage output (relatively high impedance) and so you won't use the transimpedance configuration. The 1/f noise of the amplifier could be a problem so I would say you need to be looking for something with reasonable 1/f noise. JFET amplifiers are probably best because you need high impedance but you should be able to do it with CMOS ones (I certainly have in the past).

Keith

The scene we're trying to measure is how to tell the difference between a hot stove element 4 to 7 meters away from that same hot stove element with a pan on fire on top. (It is for a low-cost fire prevention device.) While the device feature a camera for flame recognition, we're hoping to assist this with a well chosen amplification of very-low-cost photon sensors in IR, visible light and UV.

Click to expand...

In this situation, only solar blind UV sensors will work as selective flame detectors. Far IR (as utilized in usual PIR "motion" sensors) will strongly react on hot stoves as such.
**broken link removed**

keith1200rs said:

Given what you have now said, I think you need to be looking at PIR sensors - you need to be sensing heat, so not near infra-red. It is a while since I designed a PIR chip, but as I seem to recall they are simply a voltage output (relatively high impedance) and so you won't use the transimpedance configuration. The 1/f noise of the amplifier could be a problem so I would say you need to be looking for something with reasonable 1/f noise. JFET amplifiers are probably best because you need high impedance but you should be able to do it with CMOS ones (I certainly have in the past).Keith

Click to expand...

Hi Keith, thanks very much for that recommnendation on amplifying PIR.

I guess at this point (thanks to the excelllent feedback received in this thread) that PIRs are the way to go for this app.

My thinking at this point is to obtain as many ref designs for motion sensors and try to understandy how they amplified and replicate the circuitry. Their wild market success has certainly driven the cost of BOM for their related parts and we can follow in the wake of these designs to reduce costs of BOM for our usage.

Would this be your recommended approach? Anything we can do to enhance the performance of the motion sensor toward better sensitivity to our fire-related app?

Thanks again guys!!

Jean-Pierre

- - - Updated - - -

FvM said:

In this situation, only solar blind UV sensors will work as selective flame detectors. Far IR (as utilized in usual PIR "motion" sensors) will strongly react on hot stoves as such.
**broken link removed**

Click to expand...

Hi FvM, thanks for this link... If I understand correctly, the only UV sensors that can be of use to pickup the UV signature of fires are 'solar blind UV sensors" such as the link you sent? (e.g. cheap UV photodiodes would not work?)

Usual UV-photodiode are more senstive to visible light than to UV, you'll need a filter to block visible light. UV photocells are often used for gas burner monitoring and not very expensive as far as I know. There are also solar blind photodiodes, AlGaN or some similar semiconductor material, but I don't know about commercial availability.

FvM said:

Usual UV-photodiode are more senstive to visible light than to UV, you'll need a filter to block visible light. UV photocells are often used for gas burner monitoring and not very expensive as far as I know. There are also solar blind photodiodes, AlGaN or some similar semiconductor material, but I don't know about commercial availability.

Click to expand...

Very well thank you. As we're so cost sensitive I'll research "UV photocells" to see if they can fit in the BOM.

I'm a lot more informed now!!

JPPoulin said:

Hi Keith, thanks very much for that recommnendation on amplifying PIR.

I guess at this point (thanks to the excelllent feedback received in this thread) that PIRs are the way to go for this app.

My thinking at this point is to obtain as many ref designs for motion sensors and try to understandy how they amplified and replicate the circuitry. Their wild market success has certainly driven the cost of BOM for their related parts and we can follow in the wake of these designs to reduce costs of BOM for our usage.

Would this be your recommended approach? Anything we can do to enhance the performance of the motion sensor toward better sensitivity to our fire-related app?

Thanks again guys!!

Jean-Pierre

- - - Updated - - -



Hi FvM, thanks for this link... If I understand correctly, the only UV sensors that can be of use to pickup the UV signature of fires are 'solar blind UV sensors" such as the link you sent? (e.g. cheap UV photodiodes would not work?)

Click to expand...

I would suggest you follow FvM's recommendations. You are really trying to detect flames and distinguish them from heat so infra-red may not be the best way to go. I think you need to do some experimentation. Have a look at this as well https://en.wikipedia.org/wiki/Flame_detector

I did some work in the past using "normal" photodiodes to measure UV by using two detectors and putting glass in front of one detector and comparing the signal between the two. This was to save the cost of UV filters on a low cost detector but relies on having cheap photodetctors that can sense the wavelength of UV you require, and having glass which attenuates at the right point. In my case, normal glass worked. It is a technique worth considering depending on what the cost of optical filtering is.

Keith.

I did some work in the past using "normal" photodiodes to measure UV by using two detectors and putting glass in front of one detector and comparing the signal between the two. This was to save the cost of UV filters on a low cost detector but relies on having cheap photodetctors that can sense the wavelength of UV you require, and having glass which attenuates at the right point. In my case, normal glass worked. It is a technique worth considering depending on what the cost of optical filtering is.

Click to expand...

Basically a good idea for a cheap UV sensor. But as any compensation scheme, it will suffer from type variations and other non-ideal design parameters. Reduction of residual ambient light sensitivity below 2 - 5 % seems questionable. The essential advantage of UV photo cells (and some special photo diodes) is that they don't sense visible light by nature, either due to work function of the cathode metal or band gap of the semiconductor material.

It was aimed at a very cheap solution. I seem to remember it worked well enough for what it was intended for. Decent UV sensors were far too expensive, as was a decent UV filter.

Keith.

keith1200rs said:

I would suggest you follow FvM's recommendations. You are really trying to detect flames and distinguish them from heat so infra-red may not be the best way to go. I think you need to do some experimentation. Have a look at this as well https://en.wikipedia.org/wiki/Flame_detector

I did some work in the past using "normal" photodiodes to measure UV by using two detectors and putting glass in front of one detector and comparing the signal between the two. This was to save the cost of UV filters on a low cost detector but relies on having cheap photodetctors that can sense the wavelength of UV you require, and having glass which attenuates at the right point. In my case, normal glass worked. It is a technique worth considering depending on what the cost of optical filtering is. Keith.

Click to expand...

Hi Keith, thanks for that very important information on constructing a reasonable UV sensor on the cheap by placing a cheap glass filter in front. (I trust that for this method to work that the UV photodiode has to be shielded from all light by the enclosure with just the glass interfacing with the scene?)

Unfortunately I could not find on Mouser & Digikey parts that fit FvM's keywords of "UV photocell". Are there any other names for this type of part?

(I'd like as soon as possible find the intersection point between UV sensors that are usable for our app and are as cheap as possible (e.g. below $0.75))

- - - Updated - - -

FvM said:

Basically a good idea for a cheap UV sensor. But as any compensation scheme, it will suffer from type variations and other non-ideal design parameters. Reduction of residual ambient light sensitivity below 2 - 5 % seems questionable. The essential advantage of UV photo cells (and some special photo diodes) is that they don't sense visible light by nature, either due to work function of the cathode metal or band gap of the semiconductor material.

Click to expand...

Hi FvM, thanks again for that continuing quality input!

From your experience, do you think that a design based on a cheap UV photocell with glass filter amplified by cheap low-noise opamp would work for 'reasonable low cost flame detection through UV' if I also sense the light level of the scene and compensate in the CPU's source code? (Our product also includes an inexpensive light sensor... and if we can amplify it with a similar curve as the UV photocell maybe we can 'remove' part of the visible light picked up by the cheap UV sensor and return (mostly) UV data?