Hi,
That is a very interesting question indeed. It is a massive, extremely complex, and very fascinating question and whole research topics have been written on the subject. Even the greatest minds in the field have not fully resolved that one! Please bear in mind I am not a palaeontologist and can only go on what I have read in popular science books.
Certainly the earliest known cephalopods were from the Late Cambrian and were tiny cap shaped shelled animals and are known from China dating to about 512 million years old, give or take a million. We only have the shells of these animals but they contain chambers and the siphuncle connecting these chambers, a characteristic feature of shelled cephalopods. You might like to try a search under
Plectronoceras and
Palaeoceras for images of these ancestral cephalopods. As we have no soft-bodied preservation it is unresolved as to whether these early cephalopods crept along the sea bed in a manner similar to a snail or drifted in the current using arms to feed on plankton.
The molluscs go back somewhat further but from what I’ve read the fossil record in the early Cambrian is very patchy and very hard to interpret containing bizarre forms of life that may or may not be related to later forms we are more familiar with. Go back far enough and I’m sure the arthropods and molluscs had a common worm-like ancestor, probably in the late Precambrian, but it seems that the precise nature and date of this primitive animal is still a matter for conjecture and debate. The affinities of the annelids, arthropods and molluscs has not been fully resolved but some researchers have grouped them together under the
Protostomia based on embryological similarities in extant species (highly complex stuff involving cleavage planes, and position of the mouth in initial embryo development; please don’t ask me to explain, because I can’t!). The chordates and echinoderms, on the other hand, are sometimes grouped under the
Deuterostomia, the other ‘superphylum’.
I believe that it has been speculated that the annelids and molluscs were closely linked but the molluscs split from the annelids before the development of segmentation (see the 550 million year old mollusc
Kimberella ). At a middle Cambrian date (525mya-ish) we have date peculiar forms of life such as
Wiwaxia seemingly demonstrating this; although the animal had a mollusc-style creeping foot and a radula-like structure, it was covered in scales (sclerites) that has a basic structure resembling that of the Cambrian polychaete annelid
Canadia (Conway-Morris 1997). It is thought that
Wiwaxia was on the path that led to the polychaetes.
Travelling back in time even further to the base of early Cambrian about 570 million years ago we have possible traces of molluscs in the Siberian Tommotian faunas which consist of small shelly forms and tiny shells a few millimetres across. Arthropods are not present at this date but that does not mean that they were not there, primitive soft bodied worm-like arthropods may simply not have preserved, although it is equally possible that they were not present and what we would define as an arthropod had not yet evolved. It is possible that the common ancestor may be present amongst the earliest complex faunas we have, bizarre
Ediacaran animals, dating to 600 mya (approx) but these animals are so strange that many of them cannot be allied to any modern group.
The arthropods seem so different in construction that on first glance they could not be closely tied to the molluscs but, as mentioned, it is thought that their origins were closely tied to the annelids, much as the molluscs seemingly were. One can imagine a worm like ancestor with lobopod legs (in a manner similar to the modern
velvet worm ) slowly evolving a mineralised carapace in the early Cambrian possibly as a response to predation pressures, and thereby developing jointed appendages.
In short, that was an awfully long-winded reply to say we simply don’t know! We have no fossil of a common ancestor of all these groups but hints in later fossil groups and modern embryological evidence suggests that there must have been one at some far distant point…..
Ref: Simon Conway Morris (1997) 'The Crucible of Creation)
(Gawd, that took ages to write).
. 
