# In brief how to calculate mismatch? Part 1

## Recommended Posts

Hello Team,

How you usually calculate mismatch? (i mean the old tool not the new by the moment)

For instance, i got the flash report of a current installation of 81 kWp over a rooftop. We didnt put in order the fieleds neither in current nor in voltage.

So i did this:

1) I calculate RMS error of currents

2) I calculate RMS error of voltages

In both cases this RMS = 0,3% aprox. I set the values and, as this percentage is so low the gaussian distribution says there is 0,1% of mismatch, so almost neglective

What am i doing badly? Is it possible that flash reports are "magically" almost perfect?

##### Share on other sites

This is indeed the result of this Mismatch calculation: such a low discrepancy in modules provides extremely low mismatch (it is not linear).

Now what is the reality ?

0.3% RMS means 0.9Wp for a module of 300 Wp. This is a "sub-class" of the usual class powers (0..5Wp).

The PV modules are indeed measured at the factory output, and "sorted" in order to get the sample with measurement results (on the paper) as narrow as possible.

But what is the meaning of these measurements:

- According to all Flash-test manufacturers, the accuracy of these instruments cannot be ensured with a better accuracy of +/- 3 % (for usual accuracy classes),

- Probably all the modules don't degrade quite uniformly with the LID.

Therefore I really doubt that the real performance (not the Flash-test result) of each module can be ensured to stay within RMS = 0.3% when installed in the system !

This is the reason why PVsyst proposes a default value of 1% (many other people/software propose 2%).

Probably nobody knows the exact reality.

##### Share on other sites

I agree with you, i m pretty sure that if we measure string by string after LID (for instance) the dispersion must be much higher.

Anyway what i would like to recheck is that the conversion from RMS to Gaussian (or square) distribution and to the electrical mismatch is reliable and contrastable,i m not doubtful about your algorithms but need to be sure positive for future calculations as my job depends of this accuracy.

BR Jose