Alternative chemical sensors, such as webcam spectroscope

rsevs3

BullwinkleII wrote:

rsevs3 wrote:

You are correct katonafull. I was just suggesting that i think it would be a good idea to work on detection first and then worry about concentration. Just knowing that the chemical is there or not will be a big enough challenge by itself i think.

And perhaps all that's required for an aquaponics monitor.

Concentrations are not particularly important for a chemical that should read zero. I realize there will always be some ammonia in a system, but if you create a device that outputs some numbers in a healthy system as tested by a chemical kit, reset all the numbers to zero, then any spike of any kind is a problem.

The chemical tests work fine with only 6-8 different points of reference, but for me, the important one for nitrites and ammonia is simply that there is some.

Once you have some values, it might be possible to just test against know concentrations as discovered by a chemical test kit, check it with the spectroscope, then just call whatever value the spectroscope says the same level that the test kit shows. Then add some more ammonia or whatever, test again, etc etc.

That way you have a result without having to know what the numbers really mean, which as I understand it should mean you can do it with any light source as long as it casts a spectrum at the relevant points of interest.

Perhaps I'm missing something. I often find that that's the case

This is fairly similar to how i am looking at it at the moment. With the plan that if you can reliably get it to detect the chemical, then look at trying to calculate the concentration. I am interested in getting concentration to be able to log it. As well as the homemade flow meter i have in the pipe work (pun intended

) I think that you should be able to make up some known values by doing alot of tests and comparing with the test kit and averaging them out. Should get close enough.

I think in the article it mentions using the light as a heat sink?

katonafull

I discovered an interesting...

So i had a spectroscope (version 1), and I build that fallow the instruction of this:
http://www.scribd.com/doc/86801658/DIY- ... ectroscope
and a i made a spectra from an Osram Fluora....

tha i destroyed that...because of the wrong webcam.

Than i build a new spectroscope, allow the instructions of PLOTS (version 2 )..... and i made again a spectra from the Osram Fluora....

but the specrtas are different

and i think the spectra (that i made with the version 1 spectroscope) is closer to the original spectra that published by Osram company.

So i will build a spectroscope version 3

katonafull

SuperVeg wrote:

By the way you don't want to change the voltage on the LED. It is better to vary the current through the LED. If you had a large current source and slowly increased the voltage, when you reached the forward voltage Vf the LED would be basically a short circuit and it would be destroyed. You need a variable current source.

I forgot to answer...

R=U/I

I think...
It means the current through (I) the LED is determined by the resistance (R) ad the voltage(U) of the LED.

So you cant change the current without change the voltage, because R is a constant.

If you want to change the light intensity of the LED, you have to change the voltage... consequently the current will change too.

And if you increase the voltage too much...it results too much current through the LED, and it will causes the destruction of the led.

Ronmaggi

Once a method of detection has been found, perhaps a method of finding the concentration could be found by diluting with distilled water until it is below detectable levels. I.e. I detect ammonia, and it took so much distilled water until it was below detectable limits there for it is so many parts per million. Just a thought.

SuperVeg

katonafull wrote:

SuperVeg wrote:

By the way you don't want to change the voltage on the LED. It is better to vary the current through the LED. If you had a large current source and slowly increased the voltage, when you reached the forward voltage Vf the LED would be basically a short circuit and it would be destroyed. You need a variable current source.

I forgot to answer...

R=U/I

I think...
It means the current through (I) the LED is determined by the resistance (R) ad the voltage(U) of the LED.

So you cant change the current without change the voltage, because R is a constant.

If you want to change the light intensity of the LED, you have to change the voltage... consequently the current will change too.

And if you increase the voltage too much...it results too much current through the LED, and it will causes the destruction of the led.

Katonafull I explained it that way just so it was easier to understand.
Yes of course you are correct (in general) you cannot change current without affecting the voltage. However driving LEDs is a specific problem. A LED is just a diode, which means that it has a specific voltage rating. Power diodes 0.7v Schottkey diodes about 0.3v, LEDs vary more but maybe around 2.0-2.5v, depending on the device.

So if you connect a LED (say 2.5v) across a 12v battery then (before it blows up) there will still only be 2.5v across the LED. The other 9.5v will be dropped across the wires connecting the LED to the battery, which means the current might have to go up to many many amps (which kills the LED). Putting a resistor in the circuit allows the 9.5v to appear across the resistor and so the current is much lower.

If you control a LED with a "voltage controlled" supply without a current limiting resistor then it is very easy to exceed the Vf (forward voltage) of the LED and damage it.

HOWEVER if you control a LED with a "current contolled" supply then it doesn't matter !!
The current controlled supply can be set at say 100mA (or whatever the LED requires) and the voltage will automatically be correct.

This is why all specialised LED driver ICs (chips) measure the current through the LED and adjust the output accordingly. Even LEDs of the same type have slightly different Vf (forward voltage) and so using the voltage to control them is unreliable and risky.

Of course if you do not need to change the brightness of the LED then all you need is a current limiting resistor.

If you want me to give you some examples of how to control the brightness of LEDs please let me know. Linear Technology have a very large range of LED driver ICs that are relatively easy to implement. Adding brightness control using a microprocessor is fairly straightforward.

Ronmaggi

Most micro controllers have a pulse width modulated output so you can control the current without affecting the voltage. If you go onto instructables dot com you will see multitudes of circuits to do just what you are trying to do with the LEDs.

rsevs3

^^^ What they said.

I have been trying to think of a way to be able to make this an online sensor rather than a sample sensor. In my mind this is easy to do, but the practical side is a little less so.

My thoughts are something like this. Create a side stream which will be your sample point. UV filter to kill any algae/biofilter etc to try and minimise cleaning frequency. Then into the sensor which i think will need to be a glass cube type tube with a light source and the camera. All enclosed obviously. The hard part is actually sourcing the parts to make it. Then the really hard part will be to tweak the software to work as required...

The light source i am not too sure about. I dont think LED is the best solution. Full spectrum lighting like used in a proper photo-spectrometer is stupidly expensive. Maybe a metal halide globe? :dontknow:

Ronmaggi

As long as the light source puts out light in the wavelengths that are absorbed by ammonia, nitrite and nitrate, detection should be simple enough. Perhaps use a peristaltic pump to pump the sample into the test chamber. Then another peristaltic pump to pump in distilled water to dilute the sample to figure out the concentration of the chemicals. The sample could go back into circulation as no chemicals would be added. The test chamber could be any number of materials, though we would want to ensure that that material doesn't absorb the same wavelengths as the chemicals we are trying to detect.

BullwinkleII

What's the requirement for the sample. does it need to be in a gasious state? Water? If water, does it need to be the same ... .... "depth" or does it not matter. Can you just have a situation where a sheet of water is passed by the webcam? A hose puts out a fairly regular tube of water, and if the cam was in a given position, even the irregularities seen in water coming from a hose become extremely regular at a given point, say 1cm from the nozzle. If ongoing sampling produced an average, those irregularities would be even less relevant. Even if the data stream had irregularities, it's unlikely that they would just happen to exactly correspond to the exact wavelength we would expect to see from, say, ammonia. They are more likely to be noise rather than a nice discrete digital value that would confuse the readings. That being the case, with averaged multiple samples, I'm guessing it wont matter.

All that relies on water being testable rather than the sample needing to be in gaseous form.

If it does need to be as a gas, a $24 pond fogger and a heating element might do the trick. I dont think evaporation would work, because you would always get distilled water, but a pond fogger might put whatever is in the water into the air.

I have a fever of 39.86 so take all that with a a something of salt or whatever the saying is

rsevs3

A water sample will be fine. But it needs to be a known thickness of sample. I think that for any kind of consistency between setups, you need to flow through a transparent piece of glass.

Ronmaggi

A flow through design would be convenient. But I believe that drawing off batches regularly might be easier to control. Does anyone have any feedback regarding the use of distilled water to help find the concentrations of the chemicals? Peristaltic pumps are not too difficult to construct, and the have fantastic control over the volume of water pumped. All could be automated on an Arduino, even the brightness controls for the rgb LEDs, though the paper on the Schleene-Kelley cellphone spectrophotometer suggested plain white LEDs. There are LEDs with higher color rendering indexes. We sell LED replacement bulbs for nav lights, and the ones made just for white lights look funny when you put them in a tricolor light, the green comes out blue and the red is almost white, but the ones made for tricolor come out perfect. The bulbs look identical, but the high flux LEDs used are different.

sam_uk

Hi all

I am also interested in cheap sensors for measuring nitrates in small aquaponics.

Is this USB arduino spectrometer of any interest?
http://myspectral.com/

Not sure if it could do the job?

katonafull

sam_uk wrote:

Hi all

I am also interested in cheap sensors for measuring nitrates in small aquaponics.

Is this USB arduino spectrometer of any interest?
http://myspectral.com/

Not sure if it could do the job?

See the datasheet:

Measured wavelength: 400 nm — 800 nm

It isn't work in UV range.
And in the IR range isnt the best in that spectruino...
But the IDEA isn't bad.

I will have time from july, so than i will continue the development of the webcamspectroscope.

SuperVeg

rsevs3 wrote:

I know this has nothing to do with anything but i figured the people who would find this interesting keep an eye on this thread.

http://www.wired.co.uk/news/archive/2012-03/09/230-percent-efficient-leds

Thats awesome ! (If it ever becomes commercially practical)

Then you can put some heat producing conventional power LEDs with those heat sucking ones and there is no limit to the brightness you can get in one unit..
Leds are cool

-yes, I might be a bit of a LED nerd

rsevs3

Just a bit of an update for anyone who cares about this.

I have been doing research to see if this is something that is at all possible to achieve at home easily. I think the simple answer is yes. I have decided that for my own efforts and research that i will build my photo-spectrometer on the Raspberry Pi base. This offers two major advantages straight away, firstly being the cost. At only $35USD is is easily accessible. You only need to add a webcam, power supply and a case. Secondly is that it is basically a full mini computer (read: can run linux) and as such, will not need any technical knowledge to be able to build the system. Plus being not much bigger than a credit card makes it super portable. The hardest part is getting your hands on one at this stage.

http://www.raspberrypi.org/ - The official website and forum.

http://www.alliedelec.com/RaspberryPi/ - Specs and cost.

I have been looking at extracting the data out of the spectrum image that i have already been able to take. (Note, the attached spectrum is not one of my own) I have attached an excel spreadsheet with the data and the comparison to the original image. I have not been able to test the accuracy of intensity graph and i dont think that i will until i can build the system and try to calibrate it. I can weight each RGB channel though so it will be a case of trial and error. I know that warren over at the public laboratory is very keen to see a photo-spectrometer on the RP too so i expect we can collaborate and see this happen sooner.

http://publiclaboratory.org/dashboard

So far i am still not too sure how i will make the calibration fluids. Ammonia i can get but i wont know the concentration to any accuracy. That is a while off yet.

Backyard Aquaponics
A place to discuss aquaponics