pH Meter Accuracy

Bought a Hanna pHep5 98128 and have used it on two beers trying to get my mash pH right. I'm not entirely happy with it because right after I calibrate it, if I put it back in the 7.01 solution it can be off by as much as .1 (specs say that it's accurate to .05) then drift over the next 5 minutes.

I've never used a pH meter so I'm hoping someone can take a look at this and tell me if what I'm seeing is normal or if there's something wrong with this one. I will say that when I measure actual mash pH, it does not seem to drift, but because of what I see when measuring the calibration solution, I'm wondering if those readings are accurate.

The measurements below were taken after calibrating the meter in 7.01 and 4.01 solutions. After each test the probe was rinsed with water & as much liquid as possible blotted off.




Edit:Not sure if everyone else is seeing the red x instead of the screenshot, but here's what's shown when measuring 7.01 solution after calibrating:

Time pH

Initial 7.11
5s 7.13
10s 7.11
15s 7.10
30s 7.09
45s 7.08
1m 7.07
1.5m 7.05
2m 7.04
2.5m 7.04
3m 7.03
3.5m 7.02
4m 7.02
4.5m 7.01
5m 7.01
10m 7

pH meters take a while to respond. Note that what you have here is an indication which is initially off by quite a bit but which gradually migrates more or less monotonically to the correct answer. The rate at which an electrode comes to equilibrium is one measure of the quality of the electrode. One would expect an electrode which costs several hundred dollars by itself to come to equilibrium faster than 10 minutes i.e within a minute or so but it is usually always necessary to wait at least a minute. Ten minutes does seem like a long time but note that you are within 0.02 pH of the correct answer within 3.5 minutes or so. More expensive instruments monitor the rate of reading change over time and typically signal with a beep that the rate of change of pH per unit time has slowed to a value which can usually be set by the operator thus signalling that equilibrium has been approached. In the lab I have the meter connected to a computer which plots the readings over time so that I can eyeball the approach to equilibrium.


A slowing of the electrode response as the electrode ages is normal and a sign (along with greater offset and reduced slope) that it is aging. It can also signal that the electrode is dirty. Neither should be the case with a new instrument.

So yes, what you are seeing is normal in the sense that electrode responses drift finally settling out close to the right answer but it does seem a bit odd that it is taking so long. OTOH you are within spec at a minute and a half.

How do clean an electrode if it's dirty?

Thanks for the response AJ. There were a couple other trials in the screenshot that didn't show up and I'm wondering if the 4.01 test still fits with this meter being ok but slow. In this case it seems to start close to the proper reading then move away from it.

Also checked vineagar and the differnce in readings over time is much larger

4.01 solution
Time PH
Initial 4.03
5s 4.02
10s 4.01
15s 4.01
30s 4.00
45s 3.99
1m 3.98
1.5m 3.98
2m 3.96
2.5m 3.96
3m 3.95
3.5m 3.95
4m 3.94
4.5m 3.93
5m 3.92

Distilled white vineagar
Initial 2.5
5s 2.46
10s 2.45
15s 2.45
30s 2.42
45s 2.4
1m 2.37
1.5m 2.34
2m 2.31
2.5m 2.29
3m 2.27
3.5m 2.26
4m 2.25
4.5m 2.24
5m 2.23

sudsmcgee said:

How do clean an electrode if it's dirty?

Click to expand...

I used the manufacturers cleaning solution and soaked the electrode in it for an hour then rinsed it. Is there anything else should be done to clean it?

WRT to cleaning electrodes: the best method is to do what the manufacturer recommends. That said, in brewing deposition of protein on the electrode can be a problem so a soak in an protein lysing enzyme is often a good idea. There are products made especially for cleaning brewing electrodes but things like Zymit should work.

WRT drift: I have tentatively concluded that one mark of the less expensive meters is that they drift more than the more expensive ones. Looking at the data for 4 buffer and vinegar note that after a minute or so the readings may be subject to variation but that they are reasonably tightly grouped. The standard deviations are less than the specified accuracy of the meter. Keep in mind that the accuracy of a sample reading depends on threereadings: one for each of the buffers and one for the sample. If you know that your electrode response is slow you should wait a good long time in each buffer before accepting the reading as part of the calibration process.

May be of some use - I bought a cheap PH meter off EBAY with very poor performance. Put in some better quality batteries and the performance was vastly improved. Might be worth a shot.

Regardless of the fact that the meter is not quite as accurate as the specs would indicate...the variation it is showing is still small enough that is it OK for brewing purposes, even good enough for research laboratory standards.

I wouldn't worry that much....it is probably due to electrode quality as ajdelange stated. Is the 0.05 spec listed by the meter or by the electrode?

pH meter accuracy is a pretty involved subject as it depends on several factors
1. Drift in electrode offset
2. Noise in electrode response
3. Drift in electronics
4. Noise in electronics (including quantizing noise in the A/D)
5. Accuracy of the buffers used for the calibration (i.e. are they 4.01 pH ± 0.02 or ± 0.05?).
6. Accuracy, noise and bias in the temperature measurement. Either ATC is used or the buffers and sample must be held at the same temperature. In either case there is error associated with the temperature measurement.
7. The pH of the sample relative to the pH of the buffers. Best accuracy is obtained when the pH of the sample is half way between the pH of the buffers. It falls off quickly when the pH is outside that span.

In a good laboratory instrument the design is such that the buffers' accuracy is the limiting factor (at least for the usual NIST ± 0.02 pH buffers) so that a reading made mid span is usually accurate ± 0.01 pH.

Note that the mean of the pH 4 buffer readings in this case is 3.9755 and the standard deviation 0.034. Thus the standard error in the mean is 0.009. Thus the pH estimate based on the simple average of the measurements is 3.9755 ± 0.001 i.e. between 3.974 and 3.976 i.e. within 0.05 of the actual buffer pH so the meter is within spec. OTOH there is a definite downward drift over time. One would ordinarily discard any readings taken in less than a minute. If you discard those here you would get an average reading of 3.956 ± 0.007. Meets the 0.05 accuracy at the high limit and just misses it at the lower. Had the 4 buffer calibration been done after waiting 5 minutes I'm guessing the 4 buffer reading would be within the accuracy spec.

As you can see it's a bit of a numbers game. Experience teaches users how to interpret pH meter readings properly. Until then it is quite possible to be led down the garden path by one.

broadbill said:

Regardless of the fact that the meter is not quite as accurate as the specs would indicate...the variation it is showing is still small enough that is it OK for brewing purposes, even good enough for research laboratory standards.

I wouldn't worry that much....it is probably due to electrode quality as ajdelange stated. Is the 0.05 spec listed by the meter or by the electrode?

Click to expand...

Thanks for the reply and the reassurance, I was kind of considering selling it and getting the Milwuakee MW102, but I think I'll keep it for now and see how it goes. Still, it would be nice to know when it says my mash pH is 5.3, is that really 5.35 or 5.25 or somewhere in the middle?

The .05 accuracy is from the meter's spec sheet. I can't find anything on the manufacturers site showing the specs of the probe itself.

ajdelange said:

Keep in mind that the accuracy of a sample reading depends on threereadings: one for each of the buffers and one for the sample. If you know that your electrode response is slow you should wait a good long time in each buffer before accepting the reading as part of the calibration process.

Click to expand...

Unfortunately this meter autocalibrates and does not let you choose when the reading is accepted. Your explanation does help me understand (I think) why I'm seeing different behavior measuring the 7 buffer vs the 4. Likely when it autocalibrated on the 7, it was close to equilibrium but with the 4 it was not. When I later measure the 4 solution as the meter is approaching equilibrium, the reading moves past the calibration point and keeps going because the calibration point was accepted too soon.

ajdelange said:

As you can see it's a bit of a numbers game. Experience teaches users how to interpret pH meter readings properly. Until then it is quite possible to be led down the garden path by one.

Click to expand...

That's what I'm trying to avoid and I appreciate your help. At this point I think the way I'll look at it when I measure mash pH is that the reading after 1-2 minutes is likely within .02-.03 of the actual pH and more likely to be slightly higher than actual pH (based on the 4.0 calibration point being accepted too soon).

I guess another thing we have going for us is that typical mash pH fits in nicely near the halfway point between calibration points.

I only mention this because as you took the trouble to monitor pH readings over time it is clear that you have an intrinsic curiosity about how these things work but you can do your own calibration and ATC if you can put the meter in millivolt mode. You can do this with may meters but I am not sure you can do it with the cheapies. The general concept is that you measure mV in 7 buffer, mV in 4 buffer, compute the offset and slope and then use those to convert the mV reading of the sample to a pH value. As you are doing the slope and offset calculations you can decide when to accept the mV readings for the calibrations. If you are interested in this I can post the formulas (or at least I think I can - I'm out of the country at the moment.)

The manual doesn't mention millivolt mode and there's no obvious way to switch to it: mode button only controls Cº/Fº and buffers used. No other switches in battery compartment or anywhere on the thing.

I've sent a message to tech support asking if it can be done with this meter and will post back once I hear. I'm definitely interested in doing this if it's possible.

You don't really need mV mode though it makes understanding of what is going on easier. The electrode produces a voltage

E = E0 + s*T(7-pH)

where E0 is the offset, s is the slope and T is the temperature (in Kelvins - add 273.15 to centigrade). When you use a pair of buffers of known pH you get two voltage observations and can solve 2 equations for E0 and s and then in response to sample voltage Es calculate the pH responsible as pH = (E0 - E)/(s*T) + 7. If you turn ATC off the meter will use 293.15 (20 °C) as the temperature. What is important is that the estimates of E0 and s are fixed if the meter is not recalibrated and for our purposes it does not matter that the values are incorrect. You will calibrate the calibration.

1. Turn ATC off
2. Measure the first buffer's temperature and pH. Convert temperature to Kelvins. Call the measurements T1_ and pH2_.
3. Do the same for the second buffer. Call the results T2_ and pH2_.
4. Determine the actual pH's of the buffers at T1 and T2. There is usually a table on the packaging or the values can be found on the net. Look for NIST technical buffer data. I have some simple formulas but not sure I can dig them out (I'm not at home).Call these pH1 and pH2.
5. Compute beta = (T1_*pH1_ - T2_*pH2_)/(T1_*pH1 - T2_*pH2). If the meter were calibrated perfectly pH1_ would equal pH1 and pH2_ would equal pH2 so beta would be 1.000
6. Compute alpha = T1_*(pH1_ - beta*pH1). If the meter were calibrated perfectly beta would be equal 1 and alpha would be 0.
7. Measure a sample. Call the pH measurement pHs_ and the Kelvin temperature Ts_
8. Compute (pHs_/beta) - (alpha/beta)(1/Ts_). This is the temperature corrected calibrated pH estimate. When alpha = 0 and beta = 1 (meter perfectly calibrated) the estimate is identical to the reading.

I promise to check this tomorrow for algebra errors and edit as necessary.

This meter will not display raw voltage measurements from the probe per mfgr tech support.

ajdelange said:

1. Turn ATC off

Click to expand...

Unfortunately it is also not possible to turn off the ATC. I did some research and learned a little about the Nernst equation and how ATC is calculated hoping I could factor it in to your equations given that the meter readings will always show up as temp. adjusted, but TBH doing so is beyond me at this point. Seems to me I would need to know the meter's calculated calibration slope and offset numbers to be able to adjust for the ATC reading, but I'm probably missing something here, my math is not great...

I guess I could use the method and formulas you provided and do all my readings at 25C. That would work to get more familiar with the readings, but I'm not sure how practical it would be to try and hold all my samples at 25C on brewday.

Brewed a batch today and this meter was working much better. After calibrating it, measured both the buffer solutions and within 30-45 seconds the meter was within .02 of the correct reading. Leaving it for a couple minutes it moved exact or within .01 and stayed there.

Not sure why it's working better, all I can think of is that the electrode benefitted from soaking in the storage solution for some more time. Before my initial post, it had already been in storage solution for over a day, and calibrated multiple times, but always showed the inaccurate, slow pattern. Seems fine now and I hope it stays this way!

I'm not sure how new your meter was when you began using it but the "work" is done in a "hydrated layer" in the glass membrane. This needs to be formed for the electrode to work stably and I suppose it is possible that in your case it was not fully formed for those first couple of brews. Extended soaking in the storage solution is definitely a good way to form that layer and be sure the appropriate ions are held within it. pH meters are definitely subject to vagaries and it just takes some experience to be able to tell when things are kosher and when they are not.

Even though it now seems less important in light of better behavior it appears that it is possible to "calibrate the calibration" in a meter that does not display mV and in which you cannot disable ATC. It's a bit more complicated than what I posted earlier but not that much. I'm writing it up and think it would be better to transmit a .pdf than to try to post it here so if you are interested pm me an e-mail address. Don't know when I'll finish the analysis.