Monte Carlo simulations and returns suspicious results - error in the setup of results?

Hi all,

I am using ICADVM20.1 and running Monte Carlo (100 iterations) simulations and getting some results which make me think I am mistaking something in the way I set it up.

What I try to do is simulate the mismatch of currents of an array of current sources. These current sources are binary weighted, therefore each element of this array is composed of iterated instances, one double of its previous one.

Therefore to measure the current going through them I calculate in maestro (IDC("/i_pSW_B9<3>/src") + IDC("/i_pSW_B9<2>/src") + IDC("/i_pSW_B9<1>/src") + IDC("/i_pSW_B9<0>/src")) where i_pSW_B9 is the iterated instance collecting the current to be evaluated. The resulting standard deviation is 80nA.

Now if I try to measure the same but on a single instance, therefore calculating just IDC("/i_pSW_B9<3>/src"), I get a standard deviation which is about 3 times higher ( 220nA). This is totally unexpected as for each halving of the current sources I would expect the standard deviation to be divided by sqrt(2), being therefore around 40nA = 80nA/(sqrt(2)*sqrt(2)).

My suspect is that the Monte Carlo analysis would not be compatible with using IDC function? (I tried operating point as well on the source of these current sources but the result is basically the same)

Thank you

Best regards