Thinking about no longer calibrating probes

[Tangdora]

As the title states, with the exception of the Neptune temp probe whic lets you set it to a value of your choice, I’m thinking about no longer calibrating my probes. Thought process here would be to stop constantly going through periodic calibrations that shifts the probes offsets to bring it back into a proper reading.

Instead would just reset everything back to factory calibration and simply notate the variance against a calibration solution when a periodic check is performed. This way the setting can remain fix. And you don’t have to worry about a calibration making things better or potentially worse.

So for example if an apex ph probe with the factory calibration is reading 6.85 in a 7.00 solution after testing I would just rename my probe from PH say to PH-M0-15 ie (Minus 0.15). The naming convention would indicate that the probe reads 0.15 less than actual and any additional coding against the PH probe would need to be offset to factor this in difference.

Doing the math on the fly is easy enough only down side would be after each check if the variance changes all outlets using would need to be update. But a quick search can find all outlet pointed to a specific probe. Sorta already do this with orp which Neptune recommends not to calibrate.

 

[BeanAnimal]

I don’t get it? Why adjust various set points and do mental math when looking at charts -vs- just calibrating the probe?

You still have to use the calibration solution to record the offset to begin with.

 

[Randy Holmes-Farley Verified Community Expert]

That fixed correction is not an accurate way to deal with a pH recalibration. It may be close enough for values near 7, especially since many folks do not measure pH at all, but calibration is not a fixed correction.

 

[BeanAnimal]

That fixed correction is not an accurate way to deal with a pH recalibration. It may be close enough for values near 7, especially since many folks do not measure pH at all, but calibration is not a fixed correction.

Good point as well for any probe that has a dual point calibration.

 

[Tangdora]

That fixed correction is not an accurate way to deal with a pH recalibration. It may be close enough for values near 7, especially since many folks do not measure pH at all, but calibration is not a fixed correction.

Would not be calibrating anything though. the probe would being using the factory preset calibrations. All I would be doing is adding a number to the final reported value. So back to my example. If place in the calibration solution with the factory preset calibration and was reporting a value lower by 0.15 is it wrong to say that the probe then placed in the tank now reading 8.00 is not really a value of 8.15 after factoring in how off the preset reading is.

 

[BeanAnimal]

Would not be calibrating anything though. the probe would being using the factory preset calibrations. All I would be doing is adding a number to the final reported value.

Ignoring error (see Randy’s comment, and mine below) using a single offset value, you are still going through the process of dunking probes in calibration solutions and then adding the manual work of adjusting any programming tied to the values and the mental math every time you look at a reading or a chart? I am at a loss as to the benefit even if it were accurate.

So back to my example. If place in the calibration solution with the factory preset calibration and was reporting a value lower by 0.15 is it wrong to say that the probe then placed in the tank now reading 8.00 is not really a value of 8.15 after factoring in how off the preset reading is.

pH uses two points, one to get the offset and the second to get the slope of the line running through that offset. So you can’t simply apply your 0.15 adjustment to every reading.

Conductive is similar, but the two readings form a multiplier, so your measured “offset” at the calibration solution value is also not fixed and only valid at that value.

 

[Tangdora]

Ignoring error (see Randy’s comment, and mine below) using a single offset value, you are still going through the process of dunking probes in calibration solutions and then adding the manual work of adjusting any programming tied to the values and the mental math every time you look at a reading or a chart? I am at a loss as to the benefit even if it were accurate.



pH uses two points, one to get the offset and the second to get the slope of the line running through that offset. So you can’t simply apply your 0.15 adjustment to every reading.

Conductive is similar, but the two readings form a multiplier, so your measured “offset” at the calibration solution value is also not fixed and only valid at that value.

So I fully understand where you guys are coming from, and I agree. I’ll stay on the topic of pH for now. I’m home now and have the actual numbers from a test I did the other day, which is what led to this question.

Last Sunday was my six-month calibration checkup. I have a Hanna HI 98121, which I use from time to time to validate my Apex readings. Before any potential recalibration, the tank pH reading was 8.01. When I placed the meter in the 7.01 solution, it read 7.00. In the 10.01 solution, it read 10.00. Based on that, I deemed the current calibration of the meter to still be good and assumed the 8.01 tank reading was accurate.

Just for fun, I decided to clear the calibration and revert the meter back to its factory settings. Afterward, I got a reading of 6.72 in the 7.01 solution and 9.64 in the 10.01 solution. This is in line with what you were saying, that the difference is not fixed across the range. However, the difference was not substantial. The deviation at the lower end was 0.28, and at the higher end it was 0.36. Averaging those two values gives 0.32. The tank reading with the factory calibration was 7.70. Adding 0.32 gives 8.02, which is very close to the calibrated reading of 8.01.

The difference between the high and low deviations is only 0.08. There have been times when I’ve run what appeared to be perfectly successful calibrations, only to find the post-calibration readings differed by more than that, resulting in a second calibration being needed.
I imagine the factory curve (slope and offset) on the device had to pass some type of testing before being accepted as the factory setting. Whether it still aligns with today’s expected values may not be the case, which is where an additional offset I mentioned comes in to play.

My point was simply that every time you calibrate, you’re changing those settings anyway. So I was just wondering whether placing more emphasis on the factory curve might result in more stable numbers over time.

 

[Waters]

For what its worth, I haven't calibrated any of my probes in years. All still check against manual testing.

 

[BeanAnimal]

My point was simply that every time you calibrate, you’re changing those settings anyway. So I was just wondering whether placing more emphasis on the factory curve might result in more stable numbers over time.

Your example shows that the factory assumed slope happened to be close on this probe at around pH 8.

As mentioned above, you are still checking two buffers and calculating a correction. That is you still doing the calibration work just less precisely, with the added step of applying the correction manually and adjusting programming to match. It makes no sense.

In this case the average somewhat works because the errors at the two buffers are close and the tank reading falls between them. That does not mean that the factory slope will stay accurate as the probe ages.

If repeated calibrations give inconsistent results, that points to other issues to diagnose. I am not sure how you are verifying failed calibrations.

 

[Randy Holmes-Farley Verified Community Expert]

What purpose is measuring pH if it has an unknown error to it that may be large compared to the range of pH that reefers encounter?

In other words, why bother at all?

 

[Randy Holmes-Farley Verified Community Expert]

Would not be calibrating anything though. the probe would being using the factory preset calibrations. All I would be doing is adding a number to the final reported value. So back to my example. If place in the calibration solution with the factory preset calibration and was reporting a value lower by 0.15 is it wrong to say that the probe then placed in the tank now reading 8.00 is not really a value of 8.15 after factoring in how off the preset reading is.

Yes, that is wrong.

Sinplified, A pH meter calibration sets the pH 7 point AND the slope as it rises or lowers to other values. Your method only adjusts the fixed 7 point, not the slope, which is why 2 or more point calibration is used.


a pH meter is exactly accurate at pH 7 will not necessarily be accurate at pH 8 if the slope is not accurate.

I’d also caution that probes are not “factory calibrated”. They are merely manufactured.

 

[Tangdora]

So moral to the story’s just keep calibrating

 

[DaJMasta]

So there's error from the real probe performance and the default preset but then there is also drift over time, and that drift cab be in the linearity of the probe, not just a drifting offset.

If you don't calibrate, it is likely your short term readings can still be of some value, but what high pass filter is needed before medium and long term drift is eliminated? If your pH goes up by 0.01 arbitrary units per day, is that your probe drifting or your system getting more basic? If it was going up for 3 months then levels off and starts going down for a few months, is this meaningful data or has there been a slight positive overcorrection with then a slower long term drift downwards? You have no way of knowing.

Since we, as hobbyists, also generally don't understand the physical construction and properties of the measuring probe and system well enough to be able to assume a normal aging behavior of the system, we have no baseline of what drift behavior to expect or how much drift could potentially effect things (is a year old probe usually 0.1 off or 1.0 off?)

Then the initial calibration could change the linearity of the response, not just a fixed offset from the value, but whether 1mV at the probe head is 0.001 or 0.01 pH units on the controller, so without calibration you also could have your measurmeneg scale wildly differently than reality.

The real answer is, if you want stability and not to have big jumps in the measurement, you calibrate more often and moderate changes in your dosing adjustments over weeks to have smaller spikes and slower corrections towards nominal.