need help picking a ph meter

It's time for me to buy a PH meter , I would like a hand held
with detatchable probes , and trying to keep it under/around
$200.oo ish,this is going to be used here at home (brewing)
and at work taking ph readings (water)don't really want a pen
style, Any help would appreciated.

While pH meter technology has advanced in leaps over the last couple of decades both in the electrode and electronics it should be clear, given that a good lab instrument will cost around $1000 (with electrode), something you pay $59 for is going to have some shortcomings. Reading back over some of the posts here will give you an idea of the kinds of frustrations some have experienced with these inexpensive meters. A good electrode alone costs about $200. Here good means that it has rapid response, is stable (drifts slowly) and lasts a long time (more than 2 years). While a 2 year lifetime may seem short it is a great improvement over the typical less than one year lifetime that we used to live with and it seems to be getting better. I think the electrodes I'm using now have been going for over 3 years and I know one guy who has been using his for over 5.

You can get by with the inexpensive meters if you check the stability of their readings frequently (see the pH Calibration Sticky) and some brewers have been luckier than others in this regard. Given the importance of mash pH I think it is worth dealing with the frustrations but what I suggest you do if it is at all possible is to find a good used meter and buy a new, high quality electrode for it. The electronics don't age. They'll hang in there without degraded performance until they break and obviously the odds of a break occurring sooner rather than later are enhanced with a used instrument but electronics are pretty reliable these days. I know two professional brewers who got free meters from labs that were upgrading and obviously this is a great, riskless way to get a good meter if you are so lucky. Otherwise used instrument dealers in your area or e-bay/Craig's list are thoughts but none of these are without risk as you will be paying something for an unwarrented piece of gear.

If you buy something new in the low price range you are also taking somewhat of a risk because the specifications are not forthcoming. You will see a meter advertized with 0.01 resolution and 0.05 accuracy. What does that mean? Does that mean that the standard deviation of your readings of samples is 0.05 pH or that when connected to an electrode simulator at its QC check that it read less than 0.05 units error? My conclusion, after looking at a few inexpensive meters, is that it means that the electrode will drift by that much in a relatively short period of time (minutes). This is why it is important to do the stability check in the Sticky both to get a sense of how stable your meter is and as a means of correction of readings you may wish to improve the accuracy of.

Things to look for:

1. Resolution of 0.01. You probably will not find a $200 meter that doesn't have this. With this resolution, proper calibration and fair stability you will be able to determine pH to an accuracy of about 0.02 if you do the stability check even if the meter is specified at 0.05 accuracy.
2. True digital meter. The analogue ones with a d'Arsonval meter are pretty hard to come by these days but there are still hybrid ones - analogue electronics with a digital display. Your clue that you are looking at one of these is that there is a temperature compensation knob and the calibration is done by adjusting trim pots (with a little screwdriver or knobs). Digital circuitry does the temperature compensation and calibration in a microprocessor. There is none of the temperature and mechanical instability, backlash etc. associated with the analogue components.
3.This is less important than the first 2 but is very nice to have: display of slope and offset after each calibration. If you note these in your logbook you will have a record of your electrode's aging.
4.Again not so important as the first 2 is automatic buffer recognition. The meter, based on voltage alone, determines which buffer it is in during the calibration procedure so that which buffer you use first is immaterial.
5. Automatic stability detection: This is a mixed blessing. pH electrodes take some time to stabilize so there is some question as to when you should read the meter. In the automatic system the meter keeps track of how much the reading changes in a given time interval. When this is less than a certain amount the meter will either freeze the display or beep to signal that it thinks the reading is stable or both. This can be very handy when brewing as you can go off and do other things while the meter is stabilizing and, when you hear the beep, go back and write down the reading (and temperature - be sure to do that). It can be a problem if, during calibration, the meter decides it has a stable reading before it really is stable. If a meter you are looking at has this feature be sure that you can at least adjust the sensitivity of the stability detector or, better yet, be able to turn it off. I think this may be a fairly common problem with cheap meters that have the feature but won't let the user disable it.
6. Be sure that the meter can be calibrated with the NIST traceable technical buffers (4.00 and 7.00). There is another set, the NIST buffers, at 4.008 and 6.865. If your meter lets you choose one or the other of these sets, as some do, that's fine. Just be sure you don't get a meter that takes only the NIST set as opposed to the NIST traceable technical set as the latter are sold by every home brew shop and lab supply house whereas the former may be harder to find.
7. Millivolt display:
7a. Millivolt mode becomes useful as your electrode ages. Many of the meters will not allow you to calibrate if the slope drops below a certain percentage of Nernstian. This is typically 95% and while low slope is an indicator that it is time to replace the electrode if this unhappy event occurs on Sunday you are out of luck. With the millivolt display you can do calibration and pH determination in a simple spreadsheet by entering the millivolt readings and temperatures of buffers and sample. The spreadsheet calculates slope and offset and determines sample pH. You can keep going this way until the new electrode arrives.

7b. It has occurred to me, cynical bastard that I am, that one could sell more replacement electrodes by setting the rejection threshold to a high value. IOW, the millivolt mode which gets around the manufacturer's limitation could potentially squeeze more life out of an electrode but modern ones seem to indicate their deaths by intolerably slow response before their slopes collapse.

7c. There is actually a third parameter descriptive of a pH electrode called the 'isoelectric pH' it should be 7.00 or close to it. This is fulfilled so closely in most cases that meters simply assume the pHi = 7.000 and proceed from there. If you know that pHi is not 7 (I have an electrode where it is 8.4) you would have to use millivolt mode (and I do) to use the electrode unless buffers and samples are at exactly the same temperatures.

7d. Millivolt mode allows the connection of ORP electrode and ISE (ion selective [or is it 'specific'] electrodes) should you wish, at some time in the future, to explore the redox aspects of beer staling, for example, or measure the sodium content in your water or beer.

You will doubtless see discussion as to whether ATC is 'worth the extra money' and some will argue vehemently that it isn't. It's true that the main value of ATC is that it lets you get away without having to regulate the temperature of samples and buffers precisely (near room temperature is fine with ATC) or measure separately (an ATC meter will have the temperature sensor built into the electrode in most cases or come supplied with a temperature probe) and enter into the meter the temperature and it is also true that the corrections it supplies are small (0.01 - 0.03 pH depending on temp. differences and pH) but why have things in your error budget you don't have to have? The absolute limiting factor in your ability to measure pH is the accuracy of your buffers. With ATC temperature related uncertainties become less than buffer uncertainties and that is as it should be. Temp. effects can be controlled and eliminated. Buffer uncertainties can't.

To my way of thinking the best reason for having ATC is that it signals that you have a full digital meter.

There are some thoughts which I hope are helpful but keep in mind that the question 'what pH meter should I buy?' is like the question 'what car should I buy?' As with automobiles there are so many makes, models features and prices that it is nearly impossible to answer the question. Unfortunately there is no 'pH Consumer's Report' to help you. I think the best you can hope for is that several others will post with 'I have the pHenomenon meter and it's fantastic' or 'I've had nothing but trouble with my pHlub'.

I can't really do that because I use laboratory meters that are out of your price range by an appreciable factor. I have experimented with the Hanna pHep just to see what one might expect with an inexpensive meter and have found it to be ustable but not so unstable as to be unuseable with frequent calibration/calibration checks.

I’m with you on the separate probe thing. I like to leave the probe in solution while doing other things. The pen types are a little top heavy and are harder to read.

I definitely agree about the .01 resolution. Be careful though, that’s not the same as accuracy. With ± .05 accuracy you will be within a tenth, good enough for our purposes. My meter, the MW101 is spec’d at ± .02, a little better.

I disagree with AJ about the need for ATC, though I’m not vehement about it. If you’re working at room temperature a few degrees is pretty insignificant. It’s needless cost and complexity.

I don’t know where to start with digital vs analog. It’s two different ways of doing the same thing. We are, after all measuring an analog quantity. At this level of precision, trim pots are the least of your worries.

If I were dropping a grand on a meter, I’d want all the bells and whistles. For brewing in my kitchen, I want something cheap, reliable and accurate.

There’s a lot of good information on braukaiser.com. The Milwaukee SM101 he’s talking about has been replaced by the MW101, the one I use. If you want ATC and digital calibration, check the MW102.

http://braukaiser.com/wiki/index.php?title=PH_Meter_Buying_Guide

Wynne-R said:

I’m with you on the separate probe thing. I like to leave the probe in solution while doing other things. The pen types are a little top heavy and are harder to read.

I definitely agree about the .01 resolution. Be careful though, that’s not the same as accuracy. With ± .05 accuracy you will be within a tenth, good enough for our purposes. My meter, the MW101 is spec’d at ± .02, a little better.

I disagree with AJ about the need for ATC, though I’m not vehement about it. If you’re working at room temperature a few degrees is pretty insignificant. It’s needless cost and complexity.

I don’t know where to start with digital vs analog. It’s two different ways of doing the same thing. We are, after all measuring an analog quantity. At this level of precision, trim pots are the least of your worries.

If I were dropping a grand on a meter, I’d want all the bells and whistles. For brewing in my kitchen, I want something cheap, reliable and accurate.

There’s a lot of good information on braukaiser.com. The Milwaukee SM101 he’s talking about has been replaced by the MW101, the one I use. If you want ATC and digital calibration, check the MW102.

http://braukaiser.com/wiki/index.php?title=PH_Meter_Buying_Guide

Click to expand...

Thanks both of ya guys ,i was looking at both of the milwaukee's
you mentioned , is manual calibation hard to get use to ? and is
ATC worth it ( i can cool the wort easy) and i am not in a hurry.
i see it just may be (the ATC( worth it )

Wynne-R said:

I disagree with AJ about the need for ATC, though I’m not vehement about it. If you’re working at room temperature a few degrees is pretty insignificant. It’s needless cost and complexity.

Click to expand...

The cost is marginal and is not in fact the cost of ATC but the cost of having a fully digital implementation. A fully digital meter must have ATC or it must have temperature information put into it manually as it cannot calibrate without it. I don't agree that the cost is needless because what you are really buying is the improved performance of the digital implementation but I do understand that on a limited budget one may have to sacrifice the advantages digital grants. As for added complexity: there is none. It makes use of the meter simpler. The RTD is usually incorporated in the electrode so that the brewer is relieved of the burden of having to take temperature measurements of buffers and samples or, the alternative, placing samples and buffers in a water bath. At worst there is a separate temperature probe which must be placed in the solutions with the electrode (and similarly rinsed between solutions). This still seems simpler than having to use a separate thermometer. In either case you are relieved of having to adjust knobs or pots - again a simplification.

Wynne-R said:

I don’t know where to start with digital vs analog. It’s two different ways of doing the same thing. We are, after all measuring an analog quantity. At this level of precision, trim pots are the least of your worries.

Click to expand...

This is sort of like saying that the analogue computers of the days of yore are as good as the digital ones of today. That simply isn't so. In an analogue pH meter we must have a high impedance variable gain instrumentation amplifier with adjustable input offset. The design must emphasize temperature stabilization of the gain i.e. the design must be such that the gain does not change with the temperature (of the electronics). Behind the instrumentation amp there must be another variable gain amp, this one for temperature compensation, and it too must be temperature stabilized. The adjustable gains and the offsets are controlled by potentiometers - a mechanical electrical device. These are hard to set exactly and eventually become noisy as they accumulate dust and wear over the years. As they are mechanical shock or vibration can cause a change in the setting and as they are resistors their resistance changes with temperature too.

In the full digital implementation you have a fixed gain amplifier followed by an A/D converter. While you wouldn't want gain and offset to be wildly dependent on temperature you don't have to compensate nearly as tightly because the calibration is done not by adjusting gains but by computing slope and offset. Small changes in gain simply show up as small changes in slope. If the gain vs temp. characteristics of the amplifier are known the measured voltages can be corrected simply by measuring the temperature of the electronics.

Beyond the implied accuracy improvements having a microprocessor on board opens up all kinds of possibilities for 'bells and whistles' such as automatic buffer recognition, automatic stability detection and various logging functions.

A pH meter is sort of like a hi-fi. In the hi-fi the the limiting element in the performance of the system is the speakers. You therefore improve the electronics in the amplifier to the point where their distortions are well below those introduced by the speakers. In a pH meter the limiting element is the buffers. You improve the performance of the electronics until the buffer uncertainty dominates the error budget.

Wynne-R said:

If I were dropping a grand on a meter, I’d want all the bells and whistles. For brewing in my kitchen, I want something cheap, reliable and accurate.

Click to expand...

And you'd get them but many are really not that useful. Lots are aimed at helping laboratories with the documentation requirements imposed by GLP. These will record all the details of each measurement including voltage, temperature, the time and date of the last calibration, the calibration constants, the serial number of meter and electrode (there is a tendency these days towards 'smart' electrodes which do most of the processing in the electrode itself and pass only the results to the 'meter', the operator's initials etc. all in a from directly importable into Excel. Many have more than one input channel so that one may simultaneously measure, for example, pH and ORP. Some allow direct connection to a computer. This is great if you want to graph pH over time or do your own 'ATC'.

I've had good luck with this meter (link below). I use the cleaning solution monthly, storage solution in the cap without drying out, and calibrate before every measurement with pH 4 & 7 buffers poured before each brew. I'd suggest buy the $50 of solutions right away with a meter.

http://www.eseasongear.com/hahi98waphtt.html

I've recently visited a 3000+ barrel/year micro brewery that uses a very similar Oakton brand meter.

overdbus said:

is manual calibation hard to get use to ?

Click to expand...

No. People did it as a matter of course for years.

1. Bring buffers to the same temperature
2. Measure that temperature
3. Turn temperature pot to indicate that temperature
4. Rinse, blot... (see pH Calibration sticky) electrode and place in 7 buffer
5. Adjust offset pot until meter reads pH on buffer package label for that temperature.
6. Rinse, blot electrode and place in 4 buffer
7. Adjust slope pot until meter reads pH of 4 buffer at that temperature.
8. Measure temperature of sample.
9. Turn temp pot to that temperature
10. Rinse, blot electrode and insert into sample. Read pH


By comparison with ATC you do steps 4, 6 and 10. Everything else is taken care of by the algorithm. It will beep (or somehow signal that it is finished) with each buffer and you will have to push a button to tell it to accept the reading and continue and you should also bring samples and buffers to approximately the same temperature but it doesn't have to be exact.

Nice to hear from you AJ, lots of good information there.

The MW101 in Amazon is $65, the MW102 is $95. That seems significant to me. The MW102 uses a separate probe for temperature. That is added complexity.

The way I use my meter is once the sample is cool to the touch I put the probe in. When the reading stops moving I’m done. If I’m not sure I’ll use a quick reading thermometer.

As far as analog vs digital I still don’t see it. I took the cover off my MW101 pH meter and it had three little chips and one big one, and a few transistors. I’m guessing the little chips are dual op-amps, probably the same ones used in the MW102. Op-amps are analog, unless there is some digital version I don’t know about. Performance on the two meters is identical, ± .02.

I’ve been adjusting pots in all sorts of conditions and they just don’t move, unless they burn up or take a lightning hit. Potentiometers are variable resistors, for people that don’t know. Think volume control. If they get noisy you clean them.

As log as I’m taking pot shots (pun) at AJ I’m going to take issue with the HiFi analogy. Good speakers sound good with almost any amplifier. Crappy speakers sound crappy regardless of the amplifier. The distortion of most any amp is less than 1%, while even a really excellent speaker is a few percent.

Oh AJ, I still have my analog computer. It’s a Pickett slide rule. I don’t use it much anymore, I prefer my HP calculators. So there’s that. Then again I remember the idiot in Radio Shack trying to convince me that I needed a digital antenna to go with my digital converter box. There’s no such thing as a digital antenna.

Slainte

Wynne-R said:

Nice to hear from you AJ, lots of good information there.

Click to expand...

Please keep in mind that throwing out information is what I'm trying to do. I'm not trying to push people into buying digital or analogue but rather trying to make sure that they understand the difference and what the relationship between digital and ATC is. Obviously there are advantages to digital. Were there not the high end meters (and now even the under $100) wouldn't use the digital design. It can be used as a marketing ploy, of course, (as in the case of the digital antenna) but in pH meters it isn't purely that.

Wynne-R said:

The MW101 in Amazon is $65, the MW102 is $95. That seems significant to me.

Click to expand...

It does represent a nearly 50% increase in the cost of the unit but $30 really isn't that much money these days. I understand that there are people on tight budgets and for them $30 could be significant. I would say that if money is really tight one should not spend the extra $30 as the added convenience and stability are probably not worth it. People that want convenience and digital could look at the pHep units but their electrodes are unstable.

Wynne-R said:

The MW102 uses a separate probe for temperature. That is added complexity.

Click to expand...

Relative to units that have the RTD in the electrode, such as the pHep, yes. Note that I'm not trying to push Hanna. They do seem to make some decent, if not top line gear, however. I don't know anything about Milwaukee.

Wynne-R said:

As far as analog vs digital I still don’t see it. I took the cover off my MW101 pH meter and it had three little chips and one big one, and a few transistors. I’m guessing the little chips are dual op-amps, probably the same ones used in the MW102. Op-amps are analog, unless there is some digital version I don’t know about. Performance on the two meters is identical, ± .02.

Click to expand...

I can't say what the chips are but I can assure you that there are as many parts in common between the two meters as it is possible for their engineers to design in. It is not totally beyond belief that they would be identical with the digital features 'crippled', the pots removes and $30 knocked off the price. Though I rather doubt that's the case here manufacturers do that sometimes. My guess would be that the instrumentation amp (two op amps interconnected so as to give very, very high input impedance) are probably the same with the gain, offset and temperature pots replaced by fixed resistors. The A/D converter could be the same so that the difference would lie in the replacement of the display driver with a microprocessor.

Wynne-R said:

I’ve been adjusting pots in all sorts of conditions and they just don’t move, unless they burn up or take a lightning hit. Potentiometers are variable resistors, for people that don’t know. Think volume control. If they get noisy you clean them.

Click to expand...

All true but we've all dealt with noisy pots and we've all been frustrated trying to deal with backlash. To be honest about it I like analogue controls (probably because that's what I grew up on). I particularly hate digital focus controls on cameras. But I digress. I think the point here is that a pH meter doesn't need controls. A graduate TA once wrote on critique of a circuit design in which I had lots of pots that I should "think with my brain, not my fingers". That's what the digital meter does.

Wynne-R said:

As log as I’m taking pot shots (pun) at AJ I’m going to take issue with the HiFi analogy. Good speakers sound good with almost any amplifier. Crappy speakers sound crappy regardless of the amplifier. The distortion of most any amp is less than 1%, while even a really excellent speaker is a few percent.

Click to expand...

Exactly - no disagreement at all. Here the 'speakers' are the buffers. Buffers with great uncertainty associated with them will be the determining factor in the performance (accuracy) of the meter no matter how good the electronics. If you use perfect electronics with the typical NIST traceable ±0.02 pH buffers at pH half way between the buffer pH's your accuracy will be about 0.02pH. If your electronics (and thermometer) are less than perfect but still good enough that they induce less than 0.02 uncertainty the overall uncertainty will be better than 0.028. If, OTOH, electronics are worse than 0.02, say 0.03, then overall uncertainty climbs (0.36 at 0.03). My philosophy is simply that just as you wouldn't buy an amplifier that worsens the performance of the speakers you wouldn't by a meter that degrades the performance available from the buffers.

This is why I'm interested in what it means when Milwaukee says "accuracy 0.02 pH". Clearly that doesn't include the buffers because the electronics would have to be near perfect to get that level with ±.02 buffers. And what if you use ±0.01 buffers?

Wynne-R said:

Oh AJ, I still have my analog computer. It’s a Pickett slide rule. I don’t use it much anymore, I prefer my HP calculators.

Click to expand...

I was just wishing yesterday that I still had one of my old slip sticks. The kids destroyed them years ago. It's hard to believe that I had a summer job designing transformers with one many years back. Today, that laborious process would be done with at worst an Excel spreadsheet and more probably with an iPhone ap so the salesman could tell the customer in his office on the spot how much the thing would cost.

Wynne-R said:

Then again I remember the idiot in Radio Shack trying to convince me that I needed a digital antenna to go with my digital converter box. There’s no such thing as a digital antenna.

Click to expand...

Well it is receiving digital signals.

It turns out AJ was right, the ATC is a tiny bit more accurate. I got an email from the company and followed up with a phone conversation.

The MW101 is a slightly faster unit than the MW102 and the difference in accuracy is extremely small.

Click to expand...

It’s actually ± .01 for the MW102, vs ± .02 for the MW101. The difference is the ATC reads the temperature to a tenth of a degree C, better than is possible with the manual adjustment. Other than that the meters are essentially the same.

The company is going to change the spec next year so both meters will be rated ± .02, which he said is good even with the ± 1digit error of the display for Least Significant Digit. As far as the buffers, they use insanely super accurate buffers.

If you use perfect electronics with the typical NIST traceable ±0.02 pH buffers at pH half way between the buffer pH's your accuracy will be about 0.02pH. If your electronics (and thermometer) are less than perfect but still good enough that they induce less than 0.02 uncertainty the overall uncertainty will be better than 0.028. If, OTOH, electronics are worse than 0.02, say 0.03, then overall uncertainty climbs (0.36 at 0.03).

Click to expand...

AJ, Really? Would you please explain how 2+2=2.8? I’m sure you’re right; that there is some sort of counterintuitive math goin’ on. Please elucidate me.

The error from buffer uncertainty is independent of the error from temperature measurement error. We are assuming for this example that each introduces an error which is, on average, equal to 0 but dispersed to the extent of 0.02 pH. This means that 68.1% of the errors are between -0.02 and + 0.02; 95.3% of them are between -0.04 and + 0.04 and 99.5% of them are between -0.06 and + 0.06. Only 0.5% are bigger in maginitude that 0.06 pH. We are saying that these errors are "0 mean gaussian distributed random variable (GRVs) with standard deviation 0.02"

When the buffer introduces error eb and the meter error em then the total error is eb + em. If two 0 mean GRV's are added the result is another GRV with standard deviation given by sqrt(sdb^2 + sdm^2). Thus
sqrt(0.02^2 + 0.02^2) = 0.02*sqrt(2) = 0.02*1.41 = 2.82.

Standard errors are said to 'root sum square' or RSS.

Thanks AJ, for confusing me yet again. I appreciate the lesson and you showing your work. I’m not sure Gaussian distribution is appropriate to model the uncertainty of the pH relative to the meter indication. The maximum deviation is .02 so that’s rectangular. Of course it gets gassy (Gaussian) when you add the buffer so maybe it’s still a good model. I’m guessing it’s less than 100% confidence, which would be back to my 2+2=4.

It’s all over my head. As long as we’re making better beer, I’m happy. My best batch ever is the next one.

Wynne-R said:

Thanks AJ, for confusing me yet again.

Click to expand...

Sorry about that. When someone asks a question of this sort I'm never certain as to what level of technical training/experience that person has. There is also the question of how much space to take up answering a question like this one. It's fundamental statistics but there is a lot more to fundamental statistics than I can put in here. My posts are long enough as it is.

Wynne-R said:

I’m not sure Gaussian distribution is appropriate to model the uncertainty of the pH relative to the meter indication. The maximum deviation is .02 so that’s rectangular.

Click to expand...

You are talking about the meter here I assume. Where does it say the maximum error is ±0.02? I've never seen that in a spec sheet. Remember a few posts back I said I would like to know what ±0.02 actually means. Given that they don't explicitly say I have to assume that errors are normally (Gaussian) distributed with standard deviation 0.02 as errors usually are. This is because they are caused by the simultaneous action of multiple agents. In this case the actions of atoms and electrons in the semiconductors in the meters innards. Thus the voltage measured by the electrode is corrupted by noise and the electronics add more. Similarly, the RTD and temperature measurement circuits are noisy. Electronic noise is gaussian. That leaves quantizing noise. A unit with precision of 0.01 pH is subject to error of 0.01/sqrt(12) just from this effect alone (and then there is the quantizing noise of the A/D converter in the meter). Biases you can calibrate out. Another error source is non linearity. If the meter exhibits non linearity to the extent of 0.02 at peak then that is really a bias and can be calibrated out with closely spaced buffers (piecewize linearization).

Besides all that I don't know the conditions under which they measure accuracy. Electrode simulator or actual electrode?

Wynne-R said:

Of course it gets gassy (Gaussian) when you add the buffer so maybe it’s still a good model.

Click to expand...

Keep in mind that a pH determination requires that you know 7 things. Three temperatures, 3 voltages (one pair for each buffer and the sample) and the isoelectric pH of the electrode. Each of these is statistically independent of the others so you have 7 independent error sources. Central limit theorem comes into play here and the result is gaussian even if the individual error sources aren't.

Wynne-R said:

I’m guessing it’s less than 100% confidence, which would be back to my 2+2=4.

Click to expand...

Errors only add algebraically if they are completely correlated. If they are completely uncorrelated they RSS.

Wynne-R said:

It’s all over my head.

Click to expand...

Obviously some of it isn't as your comments and questions show. Now I know I tend to get way too professorial in some of my posts and do that both so that as much of the information as possible is available to anyone who wants to pursue it and as I can't really simply declare that this is the way something is and you should accept it because I say you should. I have learned over the years that anyone who uses the phrase 'trust me' isn't to be trusted.

The reason I’m assigning a limit to the deviation is that Milwaukee told me that the meter would indicate within .02, as in give it a 6.000 buffer and it will read 5.98 - 6.02. So that’s the range based on lab testing. Can we have Gaussian with limits?

The math is over my head, but I still don’t know why the buffer and the meter both couldn’t read +.02 simultaneously. I know it’s wildly improbable, but is it impossible?

Oh well, it’s been fun, but I give up. We lost the thread and the OP a long time ago.

ajdelange said:

... Given that they don't explicitly say I have to assume that errors are normally (Gaussian) distributed with standard deviation 0.02 as errors usually are. ...

Click to expand...

That's kind of a mind-blowing statement for me. Do you have a particular basis for assuming the stated error is ONE std dev? When I hear a "+/- error" I would assume they're saying this error won't typically be exceeded, I.e. 2-3 standard deviations. If your assumption were correct, I think most of us would consider the Hanna with its error of +/- 0.05 to be totally unacceptable since you'd often get readings well outside that range.

Wynne-R said:

The reason I’m assigning a limit to the deviation is that Milwaukee told me that the meter would indicate within .02, as in give it a 6.000 buffer and it will read 5.98 - 6.02. So that’s the range based on lab testing.

Click to expand...

You have more information than I do. They are telling you the peak error is 0.02 pH. They are foolish not to state this as if they don't say anything one assumes it is rms. I am skeptical about this as actual accuracy depends on so many things including the pH being measured (accuracy is better half way between buffers' pHs than it is close to a buffer pH and much better than it is for a pH outside the span of the buffers). A high quality meter won't even specify accuracy. Rather it tells you how accurately the meter can measure voltage and how accurately it can measure temperature. You plug these numbers in to formulas that do error analysis to come up with an estimate of the accuracy.

Wynne-R said:

Can we have Gaussian with limits?

Click to expand...

No and that's one of the reasons for my skepticism.

Wynne-R said:

The math is over my head, but I still don’t know why the buffer and the meter both couldn’t read +.02 simultaneously. I know it’s wildly improbable, but is it impossible?

Click to expand...

The answer is contained in the question. The key words are 'wildly improbable'. If you have a set of electronics that reads 0.02 pH high and a pH 4 buffer that is really at pH 3.98 and calibrate the meter with it a sample that is actually at pH 4.00 will read 4.04. But the probability of this is vanishingly small. On average the meter induced error is between - 1 sd (standard deviation, sigma) and plus 1 sd 68% of the time between -2 and + 2 95% of the time and the meter error is in no way related to the buffer error. In this case the errors add as if they are on the legs of a right triangle and the magnitude is that of the hypothenuse.

Wynne-R said:

Oh well, it’s been fun, but I give up. We lost the thread and the OP a long time ago.

Click to expand...

OP should have gotten some general ideas about what to look for and the worth/cost of analog/digital - ATC/noATC. I did think there would be more posts about individual experiences with particular meters.

SpeedYellow said:

That's kind of a mind-blowing statement for me. Do you have a particular basis for assuming the stated error is ONE std dev? When I hear a "+/- error" I would assume they're saying this error won't typically be exceeded,

Click to expand...

An error of 1 sd won't typically be exceeded - it happens a bit less than 1/3rd of the time.

SpeedYellow said:

I.e. 2-3 standard deviations.

Click to expand...

Which is it? Two or 3? That's the problem. If you don't say then the reader is going to assume it's rms error i.e. 1 standard deviation. If I want to measure and report error, what do I do? Stick the numbers in a spreadsheet and use the stdev (or whatever it is called) function. I don't, usually do a histogram and integrate it over some interval or take magnitudes, sort them and see how many fall into a particular percentile band at least without telling you what that band is.

If I give you the standard deviation you can easily compute the number of measurements that fall outside any band. That's why the standard deviation is usually given.

SpeedYellow said:

If your assumption were correct, I think most of us would consider the Hanna with its error of +/- 0.05 to be totally unacceptable since you'd often get readings well outside that range.

Click to expand...

Well, not that often. Only 32% of the time but this is something to think about. 5% of readings will be in error by .1 pH unit. But then if you have a good meter (defined as one whose electronics induced errors are less than the combined effect of the two buffer uncertainties) and make measurements about half way between 4 and 7 pH (which, fortuitously, is the region of most interest to us) based on the usual NIST traceable buffers with accuracy ±0.02 pH you can expect right around ±0.02 pH accuracy in the reading. This means that 5% of your readings will be off by 0.04 or more and 0.2% off by 0.06 or more. This is why I feel it is important to get the best accuracy you can afford.

WRT the Hanna pHep. I believe most of the error in that, and other inexpensive units, is electrode drift - not noise from the elctrode or quantizing noise in the 'instrument'. Drift can be beaten to some extent by frequent cal checks and I encourage people to use those. See the pH Calibration Sticky.