# Accessible procedure for calibrating conductivity measurements?

by donblair |

### What I want to do

Find a way that folks can calibrate a conductivity sensor using readily-available items.

For this first pass, I'm just trying to see if I can get a range of water samples that more or less span the range of conductivity values measured in the Stony Brook river, so that the conductivity circuit we're using is tuned to approximately the right range of values. Will fill in more detail later, dumping some notes here for reference:

### My attempt and results

1 tspn salt ~ 5.7 grams of salt

1 cup of water ~ 240 mL of water

For NaCl, 1 uS/cm at 25 C ~ .5 mg/L = 1 ppm

Recent USGS Stony Brook conductivity sensor measurements span around 0 to 2000 uS/cm:

Let's see if we can come up with a bunch of water samples that span the range from 0 to 2000, without needing special equipment ...

### Equipment:

• Distilled water from CVS is ideal -- likely around 0.5 uS/cm
• Tap water is likely around 500 uS/cm to 800 uS/cm, will increase all values by that amount -- but might be okay if we're really just trying to generate some ballpark solutions ...

### Procedure

• Place 6 cups of water in a large pot
• Add 1 tspn of salt to pot

This yields: 5.7 / (6.0*0.24) = 4 g / L --> 8000 uS/cm

Now reduce concentration by 50%, in stages, keeping each 3 cup sample that results as a reference solution:

• pour off half into new container labeled "8000 uS/cm", then, in original container, mix in 3 cups of water --> 2 g / L --> 4000 uS/cm

• pour off half into new container (--> "4000 uS/cm"), mix in 3 cups of water to original container-> 1 g / L --> 2000 uS/cm

• pour off half (--> 2000 uS/cm), mix in 3 cups of water --> .5 g / L --> 1000 uS/cm

• pour off half (--> 1000 uS/cm), mix in 3 cups of water --> .25 g / L --> 500 uS/cm

• pour off half (--> 500 uS/cm), mix in 3 cups of water --> .13 g / L --> 250 uS/cm

• pour off half (--> 250 uS/cm), mix in 3 cups of water --> .06 g / L --> 125 uS/cm

### Questions and next steps

Did I goof up around anything really basic here?

If this works out, more or less, the next step will be to attempt to calibrate the 555 conductivity meter against these solutions ...

Hi Don, looks great procedure, and while not knowing if it actually works - it does give a procedure for establishing a range for a specific set of sensors.

My comment on your procedure is it might be easier/more accurate adding water then figuring out how to drain it, but thats perhaps a refinement for who ever gets to use.

seems like a good approach. maybe specify the type of salt: fine sea salt, table salt (with iodine), kosher salt (no iodine), or Himalayan pink salt maybe (maybe not). Not sure if it really makes a difference though.

I think it might be easier to work in increments of 2 cups, since a typical pyrex measuring cup is 2 cups (but maybe that's just me). Every time one has to measure a cup I think that's added error. Also, it seems like your current procedure would require more than one gal of water. I would try and keep it under 1 gal. total so people only need to buy one gal of distilled water. You'll also need to account for some clean water to dunk the probe in between measuring standards as well.

Proceduraly, you'd probably want to start by dividing up the clean water into containers. Then you add the salt and start splitting. That would reduce cross contamination somewhat.

if one had a digital scale, weighing the water rather than measuring in cups is a lot easier and more precise. Likewise, if one had a small scale, weighing the salt would obviously be better too. Another cheap tool for measuring out small quantities is a plastic syringe (without needle). Like another approach would be to make the high concentrate then spike each cup with the syringe (a.k.a. the poor man's pipette)

You'll probably need to correct for differences in sample temperature.

i.e. temperature will affect the conductivity of the salt solution.

So, to properly calibrate, you'll need not just your range of calibration solutions that you're planning to prepare, but (ideally) you'll datalog their measured values continuously over the entire temperature range you might ever encounter. Then you'll most likely need to do a polynomial regression to formulate a correction formula so that your measurement can be standardized to a particular temperature (25C being a common reference temperature). It's doable, but it can be rather time consuming to setup and properly execute and collect all the measurements. To be extra safe, you should also measure the temperature of your 555 conductivity measuring instrument, and perhaps your battery voltage as well, and include that in the logged datasets you develop. Frankly, anything which might affect your 555 instrument's measurements should be measured and logged during the calibration phase, as well as later. Luckily, Microsoft Excel (or other software) can do the regression analysis for you once you have the data.

Sorry to be the messenger, but it's better for you to know early on what you might be facing rather than you get going and later wonder why your field measurements are strangely skewed, at which point it would be too late (after the fact) to measure your sample temperatures if you hadn't already done so. I hope the above helps. Maybe someone else here knows of a different/better way to do a proper calibration?

A short-cut alternative might be to bring your sample to your lab (if you are doing one-off measurements and not continuous measurements) and then bring both your sample and your 555 instrument to a standard, fixed temperature (that's why 25C is a convenient reference temperature) before measuring the conductivity.

Good luck, and please do let us know how it goes!

You must be logged in to comment.