Consistently low mash PH using EZ-Water Spreadsheet

[WharfRat]

Apologies, first, as I suspect this might have been asked once or ten thousand times before - I am not finding it. I finally moved from strips to a meter (http://www.myronl.com/products/ultrapen_pt2.htm) which I have calibrated and verified against the buffer solutions, etc. I am about four brews in using the meter and my PH is consistently low from ez-water's predictions. This morning, for example, predicted: 5.59; actual (20 mins into mash): 5.3. It always appeared low with the strips too, of course, but I did not trust them.

I always start with RO water and build according to style (loosely) and to get my mash PH into the suggested range.

My beers come out good, and I am not worried about a PH in the neighborhood of 5.3 (although I am usually a bit lower), but would love to be able to more accurately predict my mash PH. Any low-hanging-fruit suggestions for someone who typically finds the more in-depth water chemistry stuff a bit hard to grasp but would nonetheless like to predict/control PH with a bit more accuracy?

 

[mabrungard]

Are you cooling the wort sample to room temperature? The meter reading will be a little low if the wort is hot when measured.

 

[WharfRat]

It's a temperature compensating meter. That said, I do cool the sample a few degrees since the upper end of operating range is 166f or thereabouts, a bit close to mash temp for comfort.

 

[ajdelange]

In predicting mash pH you feed numbers into a model. To get an accurate answer you must have both a good model and good numbers to feed into it. We have neither. We know how to accurately model the effects of water alkalinity, the acid and base additions that we may make to adjust mash pH, and the malts but we don't know (or at least I don't know) how to accurately model the calcium/phytin reaction. It is easy to get numbers that accurately represent water (its pH and alkalinity) and the acids and bases (their normalities) but it is not easy to get good malt data i.e. data that actually represents the malt you are going to grind up and put in your mash tun adjusted for batch to batch, maltster to maltster, grower to grower, seasonal and moisture content variations without doing extensive testing on the actual malt. We all hope that the maltsters themselves may come to see the worth in taking such measurements and putting them in their spec sheets but no one is holding his breath on this.

Given the above it is not surprising that the numbers from a calculator will be off. The author of the EZ spreadsheet e-mailed me years ago and asked me how its predictions compared to my experience in the brew house. I replied that its predictions seemed to be low and I think he adjusted based on that. Perhaps he adjusted too far and my brewing isn't the universe of brewing either.

If you observe a bias (the average difference between the prediction and the actuality) over a respectable number of brews then the solution to your problem is simple. Subtract the bias from the prediction. This will not remove all the error however.

This is a basic shortcoming of modeling. It should be used only as a guideline and the real determination as to what is happening should be done by pH measurement as you are doing.

As I always do, I'll throw out the suggestion that you look at https://www.homebrewtalk.com/f128/ph-meter-calibration-302256/
for some tips on pH meter use, calibration and stability check.

 

[ajdelange]

It's a temperature compensating meter. That said, I do cool the sample a few degrees since the upper end of operating range is 166f or thereabouts, a bit close to mash temp for comfort.

The spec sheet says ATC works up to 25°C (77 °F). I don't know why that is as ATC is just a calculation based on the temperature reading and the operating range is spec'ed as up to 55°C.

Nonetheless, as the Sticky I referenced in my last post indicates, measurements should be taken at room/lab temperature. Not only is there a shift in the actual pH of the wort (which ATC that only compensates for the electrode's response cannot take out) but going from cool room temp to hot mash temp and back is stressful on the electrode and will shorten its useful life. Note that pH shifts lower (by 0.0055 pH/°C) at higher temperature so that may explain some of your bias.

 

[phuff7129]

I use the Brunwater.com spreadsheet. I recently bought a pH meter so I could measure my actual pH to the spreadsheet prediction. In the 7 batches I have done and measured with the pH meter I have been consistently +/- .1. Also my beer has never been better. I am very pleased with the Brunwater spreadsheet.

 

[ajdelange]

I think that's about as good as you can expect to get and I think that's good enough. Certainly I'd like to have 0.05 or less but I don't think that's realistic (can't really even measure to better than about 0.02 using ±0.02 technical buffers).

 

[WharfRat]

Doh! Thanks for pointing out the ATC limitation - I saw only the operating range and assummed it would atc throughout that range. I any case I will begin cooling my samples all the way to rt.

will certainly check the calibration sticky out. I did a two point calibration when I got it and once more... so far it is still at least indicating correctly in the two solutions.

Will also give brunwater a shot!

Thanks all for the advice and insight.

 

[Yooper]

Doh! Thanks for pointing out the ATC limitation - I saw only the operating range and assummed it would atc throughout that range. I any case I will begin cooling my samples all the way to rt.

will certainly check the calibration sticky out. I did a two point calibration when I got it and once more... so far it is still at least indicating correctly in the two solutions.

Will also give brunwater a shot!

Thanks all for the advice and insight.

I used to use EZ water, as there is a bit of a learning curve with bru'n water, but it was always giving me a too-high prediction. I found that Kai Troester's calculator always agreed with bru'nwater (yes, when I first checking pH of my mash I used all three to check and double check) and found those were almost always close to my actual pH so I stopped using EZ water.

Try using the two (or three if you're as anal as me!) side by side and see the predictions and if one of those matches your actual mash pH to see if it's simply the spreadsheet you're using.

 

[ajdelange]

As the two (Bru'n and Kai's) both use Kai's data and as his data requires a linear model any differences between the two would be in the way the phosphoric and carbonic acid systems are modeled. I would, therefore, be surprised if the differences in predicted pH were large.

 

[mabrungard]

From what I've seen, Bru'n Water and Brewersfriend DO NOT predict the same pH under all conditions. Under conditions where a very low mash pH may be predicted, the models diverge significantly.

 

[ajdelange]

Well I'm surprised but perhaps I shouldn't be. Clearly Kai's malt data gives us an average buffering capacity for each malt he measured which enforces a linear model for malt proton deficit. But then I think that DI pH and buffering capacity may be derived from color information and that assumes that some sort of curve fitting was done to the DI vs pH data and buffering capacity vs pH data and curve fitting is an art so that the results for a given input malt color may be different between the two models. But malt isn't the whole story. Water alkalinity can hand us an appreciable proton deficit at mash pH. An algorithm that estimates mash pH has to come up with a value for this proton deficit and I had assumed that this was a linear model. Given Ct (the total carbonic system carbon which is calculated from alkalinity, sample pH and alkalinity determination titration end point pH) this deficit can be approximated as a linear function of pH over a narrow range of pH values. This means that one can write the sum of the deficits of malts, water alkalinity and added acids or bases (as long as they are strong acids and bases) as a linear function of pH, set that sum to 0 and solve for pH without iteration. Again, I am assuming a lot i.e. that iteration is not used in either of the two but it could be hidden in a macro. I keyed off the comment that the differences are greater at lower estimated pH. The lower the pH the more water proton deficit deviates from linear. If sticking to linear there is again some art in picking the linear model to use. Best fit to the most likely range of mash pH or best overall fit to the range of possible mash pH's are two choices and they would result in different answers at low pH.

In further thinking about this it came to me that one isn't limited to linear models if closed form solution (no iteration) is a requirement. One can model the proton deficit of water quite nicely with a quadratic over the range of pH 4.9 to 6.8 and solution to the resulting sum of deficits quadratic in pH is easily obtained. One could even do a cubic this way but going to a cubic means a much more complex solution (though still closed form) and does not buy much in terms of the accuracy of the model. So if one of the spreadsheets has a quadratic model for water and the other a linear one there will be appreciable, I'd expect, differences in what they predict.

This is, of course, all pure speculation on my part.