Building an Affordable OD600 Sensor for Cell Concentration Measurement

[_maxim_]

Hi all,

I'd like to make an home-built OD600 sensor for measurement of cell concentration in suspension by Optical Density.

There are several spectrophotometers in the market, indeed quite expensive, so I am interested in realizing a cheap device with standard components.

Here's an example on how it should work:
https://tipbiosystems.com/blog/understanding-od600-and-measuring-cell-growth/

The problem is not to process the acquired data, I think it can be done easily with any cheap ESP32.
The crucial point for me is to create the optical reading part, a 600nm sensor capable of making accurate and reproducible readings in a glass flask.

Any idea?

 

[DaveE]

Hi all,

I'd like to make an home-built OD600 sensor for measurement of cell concentration in suspension by Optical Density.

There are several spectrophotometers in the market, indeed quite expensive, so I am interested in realizing a cheap device with standard components.

Here's an example on how it should work:
https://tipbiosystems.com/blog/understanding-od600-and-measuring-cell-growth/

The problem is not to process the acquired data, I think it can be done easily with any cheap ESP32.
The crucial point for me is to create the optical reading part, a 600nm sensor capable of making accurate and reproducible readings in a glass flask.

Any idea?

At 600nm you can just use a normal photodiode with an appropriate bandpass filter. The filters may be a bit expensive though, depending on how creative your sourcing is (eg. eBay).

Alternatively, you could just have the only illumination be from a https://www.digikey.com/en/products/detail/marktech-optoelectronics/MTE6000N/5423795 (or laser for narrower line width) then leave out the filters.

Both may require some simple imaging lenses to define the sample volume.

 

[_maxim_]

Thank you, DaveE

it looks that the https://www.digikey.it/it/products/detail/marktech-optoelectronics/MTE6000N/5423795?s=N4IgTCBcDaILIBUCiA2ADBgcgAhAXQF8g + a photodiode https://www.digikey.it/it/products/detail/excelitas-technologies/VTP9812FH/5885875 is the way to go...
I will look for Application notes... Any idea here?

I guess that the LED ray should also be protected by any external interferences as the glass flask could be exposed to the environment light...

 

[DaveE]

Thank you, DaveE

it looks that the https://www.digikey.it/it/products/detail/marktech-optoelectronics/MTE6000N/5423795?s=N4IgTCBcDaILIBUCiA2ADBgcgAhAXQF8g + a photodiode https://www.digikey.it/it/products/detail/excelitas-technologies/VTP9812FH/5885875 is the way to go...
I will look for Application notes... Any idea here?

I guess that the LED ray should also be protected by any external interferences as the glass flask could be exposed to the environment light...

Yes, ideally the only photons that hit the detector are from the LED and have passed through the sample area. That could be as simple as some beam tubes and a dark enclosure if you don't need high performance.

 

[DaveE]

Any idea here?

Sorry, I need a real question. The rest of the design depends entirely on your requirements, which I don't know.

 

[_maxim_]

Yes, ideally the only photons that hit the detector are from the LED and have passed through the sample area. That could be as simple as some beam tubes and a dark enclosure if you don't need high performance.

I need only good accuracy and reproducibility

 

[_maxim_]

Sorry, I need a real question. The rest of the design depends entirely on your requirements, which I don't know.

You are absolutely right!
I need just a simple schematic diagram as I have not much experience here and the device has to be controlled by an ESP32 for getting and storing data.

 

[Baluncore]

To get reliable results may be more of a challenge.

You will probably have to split the 600 nm source into two paths. One, the sample measurement cell, the second, a reference path. You would arrange to have an empty sample cell or a compensation optical attenuator in the reference path. You would use the reference path to regulate the emission of the 600 nm source.

To eliminate the background reading of the detector, you would need to chop the source on and off, then synchronously sample the amplitude of the detected square wave.

 

[DaveE]

To get reliable results may be more of a challenge.

Yes. The reason those lab instruments are expensive is that they are good. You need to think about what kind of performance you really need. How will you actually make the measurements and what effect common errors will have. It can be simple or complex, the choice is yours to figure out.

Those phrases "accurate" and "reliable" need both definitions and numbers attached if you want to do this like the pros. OTOH, if you just want to play in the lab, you can just build it and see what you get.

 

[berkeman]

ESP32

I'm not that familiar with that uP module, but so far I'm not seeing much in the way of support for Data Aquisition (DAQ) add-on modules or other I/O modules. Can you link to some of those resources that you are planning on attaching to your ESP32 uP?

https://www.espressif.com/en/products/socs/esp32

 

[DaveE]

a photodiode https://www.digikey.it/it/products/detail/excelitas-technologies/VTP9812FH/5885875 is the way to go...

You might do better with a large area diode like https://www.digikey.com/en/products/detail/opto-diode-corp/ODD-12W/2347746. It is about 10 times more responsive than the other.

 

[DaveE]

I need just a simple schematic diagram

OK, try this one. The simplest possible version, I think.

I usually hesitate to do this because there's a lot to know about building circuits that isn't in the schematic. If you don't mostly understand this, you'll need to get the help of someone familiar with circuits, I think. The last amp stage may not be necessary, or may need the gain adjusted; I just guessed at that (honestly it's the only way to do this since we don't know how much light gets through your system).

Next (first?) you need to deal with @berkeman's question. What will you do with the voltage out from this circuit?

 

[_maxim_]

You might do better with a large area diode like https://www.digikey.com/en/products/detail/opto-diode-corp/ODD-12W/2347746. It is about 10 times more responsive than the other.

Yes, sure... But so far as I see it is not shielded from environment light sources as the other one and, unfortunately, any measurement in a glass flask could be easily exposed at those type of interferences.
Am I wrong?

 

[Baluncore]

The problem with a LED source is that, for a constant current, the light output reduces with temperature. If you build only one channel, you are building a thermometer.

With one 600 nm LED as a source, two identical detectors are needed. One detector is used to regulate the LED light output, at a set point on the reference detector. The other detector gives the transmission through the sample. That arrangement will give reliable results.

To get precision results, you will need to greatly attenuate any stray background light levels. If that cannot be done, you will need to chop the LED light, to get an estimate of the background light level.

 

[DaveE]

Yes, sure... But so far as I see it is not shielded from environment light sources as the other one and, unfortunately, any measurement in a glass flask could be easily exposed at those type of interferences.
Am I wrong?

No, you are right, if you are allowing other light sources to get in there. Don't do that. IDK, put a cardboard box over it? The easiest way to improve your signal to noise ratio is to not add extra noise. It's almost impossible to get rid of it later.

You can provide shielding and/or filters, you don't have to buy it in the detector package.

 

[DaveE]

I need just a simple schematic diagram as I have not much experience here

With one 600 nm LED as a source, two identical detectors are needed. One detector is used to regulate the LED light output, at a set point on the reference detector. The other detector gives the transmission through the sample. That arrangement will give reliable results.

Yes, that's a better instrument. Plus lots of other improvements. That's why the lab instruments are expensive. This isn't about doing it better than Becton-Dickinson. He's not asking for that... yet. Clearly this is a learning experience about metrology.

OTOH, you get an A+ for the lock-in amplifier idea. That's the gold standard for this stuff.

 

[Baluncore]

OTOH, you get an A+ for the lock-in amplifier idea.

40 years ago, I was designing and building electronic instrumentation in a research chemistry laboratory. They paid me, but no one ever gave me an A+ then.

 

[DaveE]

40 years ago, I was designing and building electronic instrumentation in a research chemistry laboratory. They paid me, but no one ever gave me an A+ then.

Yea, the problem with real EE jobs is that often the people you work for have no idea how you do what you do or how awesome it is. No A+'s just schedules and budgets and how many service calls they get a year later.

Honestly, that's probably why I retired early. No A+'s, just a nice salary raise. Managers don't really understand how much technical types want approval and real understanding of your accomplishments. OTOH, for the EEs that worked for me, I was, like, I don't have time to understand what you did, I know you can do it or I wouldn't have turned you loose. I'll know it's good later when no one complains about it.

 

[_maxim_]

The problem with a LED source is that, for a constant current, the light output reduces with temperature. If you build only one channel, you are building a thermometer.

With one 600 nm LED as a source, two identical detectors are needed. One detector is used to regulate the LED light output, at a set point on the reference detector. The other detector gives the transmission through the sample. That arrangement will give reliable results.

To get precision results, you will need to greatly attenuate any stray background light levels. If that cannot be done, you will need to chop the LED light, to get an estimate of the background light level.

Thanks a lot for your contribuite, Baluncore!
Perhaps I could solve simply by dyeing the glass flask black, or by moving the photodiode inside a channel where it is surely protected from ambient light.
What do you think?
Sure, the idea of using two channels fascinates me, but I don't understand how it could be useful if the ambient brightness changes during the measurement... That is, during cell growth.

 

[_maxim_]

Clearly this is a learning experience about metrology.

Yes, DaveE: it's a learning phase, with the only requirement being to be able to make it with easily available and not excessively expensive components.
Your scheme with constant current source on the LED and precision op-amp chain on the receiver is very practical and interesting...
I would say, a very good starting point. Thank you so much for having shared it.
I just have to solve the voltage demand higher than the 5Vdc provided by the ESP32 SoC...

 

[_maxim_]

40 years ago, I was designing and building electronic instrumentation in a research chemistry laboratory. They paid me, but no one ever gave me an A+ then.

I'm sorry about that, and I know the environment well.

At least you got paid!

I would like to do this project primarily for myself, as DaveE well observed.
Than, we'll see.

 

[_maxim_]

Yea, the problem with real EE jobs is that often the people you work for have no idea how you do what you do or how awesome it is. No A+'s just schedules and budgets and how many service calls they get a year later.

Honestly, that's probably why I retired early. No A+'s, just a nice salary raise. Managers don't really understand how much technical types want approval and real understanding of your accomplishments. OTOH, for the EEs that worked for me, I was, like, I don't have time to understand what you did, I know you can do it or I wouldn't have turned you loose. I'll know it's good later when no one complains about it.

At least you managed to retire, even earlier.
I don't even know if I'll ever go there, there are still 15 years left and everything is so smoky it's scary...
That's why I try to keep my mind busy with projects that can give me at least some stimuli, as the rest of the work is calibration, analysis and data processing..

Thanks to your contributions, my passion for this idea is reignited again!

 

[DaveE]

I just have to solve the voltage demand higher than the 5Vdc provided by the ESP32 SoC...

The LED requires about 3V, as I recall. So you can power it from 5V, just not with that IC. It could be as simple as just a resistor. Or you could design a low dropout current regulator in a few different ways. But that's a bit more complex.

 

[_maxim_]

The LED requires about 3V, as I recall. So you can power it from 5V, just not with that IC. It could be as simple as just a resistor. Or you could design a low dropout current regulator in a few different ways. But that's a bit more complex.

What about using a common OPT101 on-chip transimpedance amplifier photodiode?

Probably a 600nm bandpassfilter like this could help to select the wanted radiation

 

[DaveE]

Maybe something like this for the LED.

edit: Oops. There is no U2. U2A should be U1B, part of the LT1635 IC.

 

[DaveE]

Probably a 600nm bandpassfilter like this could help to select the wanted radiation

Yes, but you'll lose a lot of the LED emission spectrum. Also, it doesn't stop unwanted radiation within the passband. Something wider is probably better to maintain a high signal, assuming the absorption spectrum of your cells is also broad enough. Get rid of the unwanted radiation within the passband too. Like the cardboard box idea.

The other good alternative is fancier electronics that modulate the LED and only detect signals at that frequency. A narrowband AC voltmeter as @Baluncore suggested. This requires a lot more EE effort though. I still prefer the cardboard box idea.

 

[Baluncore]

I still prefer the cardboard box idea.

That is a good start. Keep it simple. Paint the inside of the box matt black.

Here is a concept schematic for a two channel instrument that would probably do what you need.
The output voltage will be between 0 and Vref, say about 2 volts.
The output amplifier can be adapted to the signal destination.
Maybe with the careful choice of CMOS R2R op-amp, it will not need a negative supply.

 

[_maxim_]

Thank you very much for the skills you make available to others.
I will follow your advice developing the circuit around the cardboard as soon as I have studied its operation.

You guys are really great, Baluncore and DaveE, thanks again.