Hacking Hanna

[RubyU235]

Has anyone seen a tear down video of the Hanna checkers?

I am curious about the electronics inside and the implications for automation if you could interface with them.

 

[Dan_P]

Has anyone seen a tear down video of the Hanna checkers?

I am curious about the electronics inside and the implications for automation if you could interface with them.

@taricha has cut one up. I think there was also someone else that @taricha might remeber.

 

[PeterEde]

If it didn't provide a number I'd never buy another.

 

[taricha]

@taricha has cut one up. I think there was also someone else that @taricha might remeber.

Here's the other thread. Below are my notes about taking one apart.

Here's some notes from when I took apart my Low Range Iron checker.

On mine the only part that actually had to be removed is opposite from the hinge side, where the white top piece is glued to the plastic body cover. Once the plastic tab is separated, the plastic body ought to come off the electronics.

(I figured out after cutting, that most of the cutting wasn't necessary)

This is a closeup of the Photocell and the attached filter.


This is a closeup of the LED.

The LED is a 571-2nm yellow-green with a very narrow width. FWHM is only about ~11nm wide. The wavelength is within manufacturing tolerances of 575nm which is what it’s labeled as. I checked to see if the cyan filter was being used to shift the LED peak, but it does not shift the peak by even a nm. The filter allows all LED light through. It simply blocks anything longer wavelength than the LED itself, so any fluorescence would be blocked.
Later I realized many light sensor devices have this similar cyan filter, so it's not specific to the LED, but simply to cut off the photocell from picking up extraneous IR, or other unrelated long wavelength light. If I found the correct specs for a similar photocell, these things have a sensitivity that's throughout entire visible, and very high in IR. So I think the filter is just to tame that behavior.

(Optical Properties of the LED and filter over the photocell).

I tried and mostly succeeded to run the checker through a calibration process while measuring the current and voltage. What I found was that the calibration routine is surprisingly complex.

attempting to run the disassembled checker through a measurement cycle, and though it would not finish because the room light caused a “Low Light” or "Hi Light” error, it would do the “C1”-blank checking phase. Below are the Voltage and Current performance of the
LED during the “C1” check phase. It starts, then in the long calibration It will vary the light brightness every couple of seconds for about 30 seconds. Max voltage measured was ~1.9V, minimum lit voltage ~1.8. Max current was 18mA with min lit current 10mA.
(when voltage went off the chart, is where my hands slipped)

So for calibration, It shoots a series of measurements, getting the photocell response at a bunch of different LED brightness through the blank "C1" cuvette.

I'll add one thing I found out later. The photodetector part is easy to get info from, it's simply a light-dependent voltage output. I connected leads and got a voltage from it without needing a power source. The power is just to run the LED / computer side.
The photodetector as I understand it is pretty cheap though, and if you just want a light detector, getting those would be better than taking apart a checker to get access to a very common, cheap part.

 

[Wolfw28]

Has anyone seen a tear down video of the Hanna checkers?

I am curious about the electronics inside and the implications for automation if you could interface with them.

I’m working on something similar using a rgb light sensor you can buy

TCS34725 RGB Light Color Sensor Colour Recognition Module RGB Color Sensor if anyone wants to join in I’d greatly appreciate the help.​

Doing a nitrates calcium kh and mg test. Im not a programmer just a dreamer and thinker with all kinds of ideas who needs help executing them.

 

[RubyU235]

Here's the other thread. Below are my notes about taking one apart.


I'll add one thing I found out later. The photodetector part is easy to get info from, it's simply a light-dependent voltage output. I connected leads and got a voltage from it without needing a power source. The power is just to run the LED / computer side.
The photodetector as I understand it is pretty cheap though, and if you just want a light detector, getting those would be better than taking apart a checker to get access to a very common, cheap part.

Very cool. Could you foresee a way to access the data it displays? Like get it to transmit to a device?

 

[taricha]

access the data it displays? Like get it to transmit to a device?

I don't think so.
I'm guessing that the regression equation is stored inside the electronic part that does the display so unless there's a set of connections that allows you to get the actual numbers off the display then I don't see how it could work.

 

[dmsc2fs]

I think this might be a fools errand. I have read numerous times about Mastertronics out of service due to calibration or other issues preventing them from working. Never bad to tinker and try and make things work and it would be cool if you solved this puzzle.

 

[Sral]

Interesting Idea.

I believe the concept is very simple, judging from this video.
First the LED and sensor get a baseline, so at which voltage does the LED start to shine, what is the sensor's dark voltage and how much light gets through the blank cuvette and water. Maybe it also tests the LEDs I-V-curve or sensors signal to cancel aging.

Then you add the test fluid and then the IC basically tests how much light gets absorbed. Since there is only one color, more color means less light at the sensor. This can happen two ways: either shine an amount of light that you tested on the blank cuvette before and see how much signal comes through, or shine increasing amount of light on it until you reach a set threshold at the sensor. First method looks at the remaining sensor voltage, second method looks at the needed LED voltage/current.

The constant light method has the upside of using sensor linearity, while having the downside that the highest concentration values are hard to measure, since the remaining signal might be close to the dark reference.
The constant sensor signal has the upside of being very easy to detect irrespective of noise and background, while it has the downside that you would need to know the LED's output very well. Not sure how well LEDs output behaves at age and with rising voltage and current.

It basically does the same as a person would do with a color chart and eyes.

The output is either by a voltage that's fed into a ADC display, or the IC sends the data directly to a display. Only the first would be easily extractable I think.. Another method would be to look whether the IC has some kind of UART ports in the board that one could hijack and hope that the IC sends debugging messages throughout the process.

 

[Sral]

Only difference between the different devices I could think of is that the LED color might need to be adjusted to the specific color that the test liquid is going to produce.

 

[taricha]

@Sral Building a colorimeter is simple (in principle). You need only know the ratio of the light that makes it through the blank to the light that makes it through the sample. Then you calculate absorbance plug it into your regression equation to get ppm of the thing you want from the absorbance.
@RubyU235 is apparently looking to automate some tank management process based on the results of hanna tests of various concentrations. That sounds hard.

Frankly it'd be easier to use an open source colorimeter - generate your own regression equations for whatever test kits you want to use, and that result could be less difficult to transfer to whatever tank maintenance system you are trying to use.
Even then I don't see the upside - since all the test kits have to be executed manually, I'm not sure how automating the transfer of result helps any?

anyway, before anyone spends time trying to build a hanna checker as an open source thing, it's already been done by io rodeo.
old one here: been out of stock for years :-(
http://public.iorodeo.com/docs/colorimeter/

they are apparently close to releasing a new one.
https://blog.iorodeo.com/open-colorimeter/

 

[Sral]

@taricha That's basically what I was trying to say, it's simple in principle and only requires very little components.

I do agree that an open source or DIY colorimeter would likely be much easier than hacking a Hanna one.
If you go that way you can even automate it completely by using stepper motor peristaltic pumps and a self-emptying test chamber by using a greedy cup siphon, but that would require some development work ^^

 

[RubyU235]

Great discussion all. Yes I agree that automating manual tests would be a bit silly. However, it would be pretty cool to just be able to put a vial on a colorimeter and have it read it for me, no squinting, trying to guess the color depending on the lighting, etc… besides, I’m the further nerd away from being able to assemble, code, and test this.

like, I tinker with RPI’s and arduinos for fun projects and 3D printing but it ends at GitHub because I don’t know anything about that lol