Brugmann's Law

A question on Brugmann's Law. It is widely known that sometimes there is
no Brugmann's Law in Sanskrit when we would expect it and this has to be
explained in various reasons.
Mayrhofer's solution is that Brugmann operates only in front of *m/n/l/r
but I find that impossible since we have forms like acc. pá:dam, nom. pl.
pá:das, nom. pl. gá:vas etc.
However, that gave me a new of the top of the head idea (probably not new,
it's probable that it's old and that I've never seen it) - maybe Brugmann
operates only in front of voiced consonants? This would explain rasá: ~
Slavic rosa (no need for the laryngeal), pátis (but this is easily analogy
to *poty-), ápas ~ Latin opus (but this could be due to *h3e-, and not
real *o-) and, most convincingly to me: katarás ~ Greek póteros, Slavic
kotorU, Lith. katra`s etc.

Now, I cannot think of an example with unvoiced consonant and Brugmann's
Law? Any examples? I haven't really looked seriously for examples, one
obvious candidate is nápa:t which indeed has -a:- as from Brugmann in all
the right places (nápa:tam etc.) but according to Macdonell there is no
*napat- attested in Vedic so one cannot see the oposition short-long.

Mate