Solved: Suggestions detecting bioluminescence

varnon

Hello again,


This is a weird project, but as you guys may learn, I'm often involved in this kind of thing.


I am working on a project and want to detect the light level emitted from bioluminescent bacteria. I found a few photoresistors that work "well" at the light level emitted by the bacteria. The light is very dim, more akin to a glow-in-the dark plastic than anything else. Originally, the idea was to use an op-amp to amplify the voltage coming from the photoresistor, then use this to activate a transistor, then a relay. I do alright with coding, but I'm not as good with circuits. I've never been able to get the op-amp to work, and I've spent weeks on it.


Then it hit me, a sigma delta circuit! I completely forgot about this possibility. I did some tinkering and set one up last night. I didn't have too many problems with the general test. However, it looks like it may be difficult to set up a proper circuit for the bacteria. I've gone through a bunch of the old threads and read about the issues with the circuit, the close spacing needed, not being able to set up a decent circuit on a breadboard etc. It sounds like a mess. An IC might be easier. Additionally, using ADC objects (both the ADC object included in the default library and another version coded in spin) I have no idea what the units are. Am I looking at millivolts? Who knows. (I'm pretty sure the measurements aren't accurate anyway, given that I set the circuit up on a breadboard.) I really think I would much rather use an integrated circuit.


I've looked at a few ICs, and I would like some suggestions if anyone has any. I need to measure a very small voltage. I don't know what the voltage range is when powering the photoresistor through the 3.3v pin on the propeller right now (I can check). But in the original 12v circuit only .03 volts come through the photoresistor during bacteria-light (.003 in complete darkness). Its a 10x difference, but the voltage levels are still very low.


I've seen a few of the integrated circuits on parallax's website. I'm a little unsure of how to interpret all the specifications on the data sheets.


http://www.parallax.com/Store/Components/AllComponents/tabid/762/CategoryID/26/List/0/SortField/0/catpageindex/5/Level/a/ProductID/573/Default.aspx
This one lists a few specs:
Input Voltage Range (V) 2.7 to 5.5
All inputs and outputs w.r.t. VSS -0.6V to VDD +0.6V
Voltage range 0.25 — VDD
I'm not really sure which of these describes the range of voltages that can be sensed. (I think "voltage range" is correct.)
I'm also not sure how precise this one is. Remember, tenths of a volt is probably not specific enough for me.


http://www.parallax.com/Store/Components/AllComponents/tabid/762/CategoryID/26/List/0/SortField/0/catpageindex/2/Level/a/ProductID/232/Default.aspx
This one states "...providing voltage measurements with 1.22-millivolt resolution..." which sounds great but I'm not really sure what the range is.


Does anyone have any suggestions? I'd really prefer something with a reliable object already in the data base. I'll also except "check out this awesome digital lumen detector" as an answer.
Thoughts are much appreciated.
Thanks

Mike G

Amplify the photoresistor output.

Duane C. Johnson

What color of light does your bacteria emit?

I would think the light is quite small.

Photo resistive cells, possibly cadmium sulfide?, are notoriously non linear.

You might want to use a larger sensor area such as you get with a silicon PV, Photo Voltaic cell.

With PV cells you want to convert the PV current to a voltage and then read the voltage.

Here is a circuit I have used:

Note! The op-amp maintains zero voltage across the PV cell. The feedback resistor opposes the current generated by the PV cell. The resultant voltage is the proportional to the photo current.

This basic circuit can be both very sensitive and very fast.

Duane J

Duane Degn

Normally with a photoresistor you make a voltage divider with photoresistor as half of the divider and an appropriate resistor as the other divider.

Robotics with the BOE-Bot version 2.2 (IIRC) has a circuit using a photoresistor. It also has a sidebar explaining a voltage divider.

The voltage divider sidebar is gone in version 3.0 and they use a phototransistor instead of a photoresistor.

The voltage range from the voltage divider will depend a lot on the resistance of the resistor you pair the photoresistor with.

jmg

varnon wrote: »

I am working on a project and want to detect the light level emitted from bioluminescent bacteria.

Have you looked at Light-to-Frequency Converters ? Keeps the Analog domain stuff right inside the sensor.

Add an LED at around the bacteria levels, SqWave driven to very low levels (ie comparable optical levels) , and you should have a calibration square wave you can use across sensors, and across measurements.

Tracy Allen

Are you sure your original sensor was a photoresistor? If you were measuring a voltage change from 0.003 to 0.03 V directly (multimeter?) then it was more likely a photodiode of some sort.

The light levels you are dealing with may be well below one lux, which is the level you'd see with a full moon overhead. I'd hazard a guess this project needs sensitivity down at the 0.01 or 0.001 lux level. There are a couple of chips for ambient light that do approach those levels.

Analog: Microsemi LX1973a, uses 4th root analog compression with dark current equivalent 0.005 lux at 25 °C. MSOP8 I've used that one myself in connection with activity of nocturnal animals.

Digital: Intersil ISL29020, i2c interface, lsb equivalent 0.002 lux, ODFN6.

If quantitative measurement is not so important, to flip a relay, then a straight voltage amplifier or straight into a 16 or better ADC should work to detect that 0.003 to 0.03V change. The open circuit voltage of a photodiode is proportional (roughly) to the logarithm of the incident light level. It also depends strongly on temperature and on leakage. Most circuits you see monitor the photocurrent, which is linear with light level, rather than the voltage.

Duane C. Johnson

Hi Allen;

Tracy Allen wrote: »

The open circuit voltage of a photodiode is proportional (roughly) to the logarithm of the incident light level. It also depends strongly on temperature and on leakage. Most circuits you see monitor the photocurrent, which is linear with light level, rather than the voltage.

The cool thing about using the current to voltage converter circuit I posted is the voltage across the PV cell is essentially 0V. So the leakage current is essentially 0 also. What is left over is the highly linear photocurrent.

Another advantage when the voltage is 0V is one is not slowed down by the relatively high capacitance of PV cells.
(Ya, speed is of no concern in this bacteria application, just saying.)
I have used this circuit to detect an extremely weak 6.4KHz light signal amongst very bright sunlight at several miles.

Duane J

varnon

Hey everybody,
Thanks for the responses. Sorry I didn't address them sooner.

The bacteria emit light somewhere between 450 and 500 nm. The glow in the dark object I'm using for testing is probably a little more towards the 500 end.

The photoresistors I have peak at 515 nm. It seems to be the closest match.
http://www.alliedelec.com/search/productdetail.aspx?SKU=70136801

I'll look into the suggestions you guys provided. I'm especially interested in the products Tracy Allen listed. I can't respond much more on the matter until I have had a chance to consider the suggestions in more detail.

Thanks,
Chris

Duane C. Johnson

Hi varnon;

Let me state the problem:
1. The light generated is relatively small.
2. Its a science project so measurement accuracy and repeatability is a must.
3. The light is in the Blue to Green range.

Since the light power is so low one needs to have a sensor the captures as much as possible.
Small sized sensors don't capture very much light.
Large sensors capture more.

My suggestion was to use large photovoltaic cells running in current mode using the basic current to voltage circuit I posted..
Silicon sensors have a peak sensitivity in the 950nm range. However, they work well up into 400nm or so. See:
Spectral Response
I have a bunch of experience measuring very low light levels and the silicon cells work well.

Silicon cells in current mode have a quite linear response to light intensity and very repeatable. Light dependent resistors are not.

Lets say you put the bacteria into a cuvette or a Petri dish. The cell can cover one whole side.

After the amplifier you can input to the Prop with the sigma delta circuit which works well.

Duane J

Lawson

Duane C. Johnson wrote: »

What color of light does your bacteria emit?
I would think the light is quite small.
Photo resistive cells, possibly cadmium sulfide?, are notoriously non linear.
You might want to use a larger sensor area such as you get with a silicon PV, Photo Voltaic cell.
With PV cells you want to convert the PV current to a voltage and then read the voltage.
Here is a circuit I have used:

Note! The op-amp maintains zero voltage across the PV cell. The feedback resistor opposes the current generated by the PV cell. The resultant voltage is the proportional to the photo current.
This basic circuit can be both very sensitive and very fast.

Duane J

I'll second this. I've personally used this circuit with a 90Mohm feedback resistor, (3 30Mohm resistors in series really) a BPW34 photo-diode,and a low-noise fet/j-fet input op-amp fed into a 24-bit sigma-delta ADC to measure single digit picowatts. For instance a TLV272 op-amp should work well. If the feedback resistor drops below ~1Mohm I'd want a different op-amp with lower input voltage noise. A good DMM can also be used instead of the 24-bit ADC, if it reads to 100uV it will detect a minimum of ~2 picoWatts hitting the sensor. (aka a bloody dark room!)

My other favorite photo sensor is a BPW34 operated as a photo-voltaic driving a gain of five non-inverting op-amp. (using a FET input op-amp like above) I get a couple of killohertz of bandwidth and MASSIVE dynamic range due to the logarithmic response. Accuracy is crappy, but the dynamic range makes it awesome for alignment of optics.

Lawson

Phil Pilgrim (PhiPi)

Some interesting ideas in this thread over at Sparkfun:

http://forum.sparkfun.com/viewtopic.php?f=14&t=27509&view=previous

I second the use of a lens to gather more light to focus on your sensor. It will eliminate a lot of hypersensitive electronics. With a big enough lens, you may even get by with a TAOS light-to-frequency or light-to-digital sensor, which gets rid of the analog electronics altogether.

-Phil

varnon

Just an update in case anyone else comes across a similar project.


I ended up using a TSL2561.
https://www.adafruit.com/products/439


I ended up passing this part of the project to a colleague who is more skilled in this area, and who had more time for the project.
We talked about the various suggestions in the thread and had a list of things to try. He had a TSL2561 so we started with that.
And actually it worked pretty well. There still are some issues, but these are more related to the size and concentration of the colony, and other non-sensor issues. The light is extremely dim, generally moonlight is brighter. With the current set up we can get about a 0-4 measurement from the light, room lights read at about 1000. I can't remember offhand if these are actual units of light or just raw data. They don't produce any measurable infrared. I've also made a lot of progress in other programming skills since this thread, so the other suggestions might be something I explore later. For now the goal is to adjust the size and concentration of the colonies, methods to maintain them, and methods to entice them to glow to maximize the measurement we get out of this sensor.


Also, I learned the species we are using is not a bacteria at all, it is a protist (Pyrocystis fusiformis). I'm developing an apparatus for another student and somehow she told me a completely different species. A lot of the properties of the light are different than what I expected. But it all worked out.


Now that I have some experience with integrated circuits and I2C protocol, I really can't making circuits with photoresistors/transistors and op-amps anymore. The programming aspect is just so much easier than the circuitry aspect. I spent forever messing with op-amps and could never get anything to work. I can get a protocol for an IC sensor working in about a day. And if I can't or just don't want to, there is probably an object for it already. Got to love working with the propeller.

Thanks for all the suggestions.