I never had that big difference, but this is what ajdelange answered when I asked same question:
Originally Posted by ajdelange
Wish I had a nickle for each time I've answered this one. Calculation of mash is theoretically fairly simple but there are practical difficulties in implementation. AFAIK neither of the two spreadsheets you are comparing uses a 'robust' model as this would required iterative solution and while that is easy enough to do with Excel it's probably not something the casual user will want to be bothered with. Even if they did use the robust model it would be very difficult to obtain the data the robust model requires (which among other things says the robust model isn't that robust). This is because the model needs to know the titration curve for each malt in the grist and that varies with malt type, barley cultivar, moisture content, maltings, crop, batch number and, most significantly, mash time and temperature.
Thus the spreadsheets and calculators must use a much simpler model which approximates the behaviour of, for example, a representative 40L crystal. The models are the province of the spreadsheet creators. Some use measured data, some use correlations with malt color, some have empirical tweaks etc. In modeling there are an infinite number of degrees of freedom. Different model, different results. The democrat's model of Obama care is that it will make excellent medical care available to everyone while dramatically reducing the costs. The republican model is that it will destroy our medical care system and take the rest of the economy down with it [just noticed you are in Croatia so this is probably not a terribly meaningful example to you]. Using that as an illustration makes the obvious question stand out in capital letters: which model do you use/accept/believe? Or should you average the results of the two. In the mash pH model the average of the two is 5.48 and that's about what you would get if you left out the baking soda. With baking soda, the mash pH will be higher - probably closer to what EZ predicts so in this case EZ is probably the better predictor but it is very likely Brun water will be the better predictor in other cases. Your best course is to use both spreadsheets while carefully measuring the actual mash pH you obtain when you brew. You will discover from this process which of the two better models your brewing materials and practices and this may depend on style. In the future you can then use that spreadsheet but by that time you will find the combination of your pH meter and experience more powerful than a spreadsheet and rely on them. Eventually the pH meter will go the same way.
The sodium calculation is very easy for a spreadsheet or calculator so if the two spreadsheets are giving you conflicting answers it's likely you are not entering the same data so check that.
Until such time as you get the spreadsheet thing sorted out you might want to just follow the recommendations of the Primer.
It is most important that you do not add any bicarbonate. This will drive mash pH as high or higher than the EZ prediction. Also there is no need for the magnesium sulfate. You can use some as an additional source of sulfate if you want that but beer generally tastes better with lower magnesium (though it may help you to live longer).
There is so much environment factors that affects pH and its pretty tough to make realistic calculation.
One of most variable factors is barely, it seems that the key to get right pH is reaction between Ca and Mg with malt phytin which produces acids to decrease alkalinity (left over alkalinity is RA). And result of that reaction would be part of the reason why calculators have diff. pH result.
Hope it helps.