The calibration guide gives a concrete adjustment rule: new factor equals old factor times the reference reading divided by the program reading. It gives separate voltage and current versions of that relationship. That’s much more useful to me than treating 769.0 and 0.85 as numbers that simply belong in the sketch. A factor connects an observed comparison to a calculation; it needs a meaning as well as digits.
Calibration makes much more sense to me when I think of a relationship instead of a magic multiplier. The sensor path gives the program information that has to be related to the quantity being measured. A reference comparison can help establish or adjust that relationship under stated conditions. It doesn't mean that any multiplier which produces a familiar-looking answer is justified. I want to know what the reference was and which part of the path the constant is meant to correct or describe.
A small example shows one reason. In this case, a reference says 2 units and a provisional reading says 1.8. Multiplying by about 1.111 would make those two agree at that point. Now suppose another reference condition is 4 units and the provisional reading is 3.4. The same multiplier would give about 3.78, not 4. Those numbers aren't results from the meter. They show why matching one point doesn't automatically establish the behaviour over a wider range. The disagreement could contain information that a single adjustment hides.
It might be tempting to solve the two points with both a scale and an offset. Mathematically, a line can pass through two points. I would need more knowledge of the circuit and more appropriate comparisons. The example is useful because it separates obtaining a fit from understanding the system. An equation can describe the points I gave it without proving a general measurement claim.
The reference instrument also has to be understood. A multimeter reading is useful information, but the instrument has its own range, method and limits. The particular model and measurement conditions matter. I don't have the raw comparison table or a reference specification to use here, so I can't turn the constants into an accuracy percentage. More decimal places in a calibration constant don't create more certainty in the comparison. They can simply make an unexplained choice look more exact.
The saved versions make that point practical. The report and smaller RMS sketch use 769.0 and 0.85. The final-named file has 766.0 and 0.85, while another saved copy has 625.0 and 30.0. I can locate those values and follow their multiplication without pretending that the filenames give their experimental history. The guide explains a method, and the report says a digital multimeter was used. The individual comparison readings would be what connects a particular pair of factors to a particular physical setup.
I want to keep the voltage and current calibrations separate enough to understand. The two sensor paths don't have identical scales simply because both enter the ADC. A multiplier may apply after a particular sequence of conversion and offset removal. If I move it into a different sequence or use it with another arrangement, the same digits may no longer represent the same relationship. The constant needs the code and the circuit version with it. Otherwise I might preserve the number while losing the conditions that made it meaningful.
This is also why changing a multiplier shouldn't be my first response to every unreasonable output. A connection problem, a wrong range or a misunderstood offset can affect the reading before calibration enters. Adjusting the final scale might make one display agree while leaving the earlier problem in place. I want to trace the stages first, so the comparison is being used to calibrate the intended relationship rather than compensate blindly for something else. The difference matters even if both approaches briefly produce the same visible answer.
The useful record would connect the arrangement, the reference reading, the unadjusted result and the calculation used to obtain the constant. Repeated or additional conditions would then have something clear to compare with. I'm not inventing that record just because it would improve the post. I want to recover the actual comparisons where they exist and leave the missing information visible where they don't. It would be much less useful to create a neat table of plausible values than to admit which part of the explanation still needs evidence.
The guide makes the reason for a scale adjustment explicit. If the program reads low relative to a suitable reference, the reference-to-program ratio raises the factor; if it reads high, the ratio lowers it. That explains the direction of the change without claiming a particular comparison was performed. The factor still belongs beside the circuit, input settings and stage where it is applied. Its role is a relationship, rather than a set of digits to carry unchanged into every later version.
The useful explanation follows the reference ratio and the place the factor enters the code. That gives each constant a job, while keeping a precise-looking number from standing in for the comparison behind it.