As for ISO standards becoming freely available, I doubt that that will happen any time soon.
8.1 Special requirements
. . .
Rule 3.
. . .
a) In modern vernaculars, anusvara before a stop or class nasal shall be transliterated as the corresponding class nasal; in other languages, anusvara before a stop or class nasal shall be transliterated as the corresponding class nasal unless it arises from sandhi (euphonic combination) of final m with that consonant.
EXAMPLE 1 Sanskrit संग is transliterated as saṁga when it represents the noun formed from sam + root gam, but as saṅga when it represents the noun derived from the root sañj.
8.2 Recommendations
Where word boundaries are not shown in the original text (as happens commonly in Sanskrit) and a word ends in a consonant, the transliteration should show word division by a space; but when phonological processes result in two words sharing a common vowel, no attempt should be made to separate them. This will require a good knowledge of the language in question.