As Ruth notes, things are more complex than just looking for a line that once had a healthy and long-lived ancestor as the breed has not been very well served in the interim, with so many breeders now given the breed's popularity, and by any measure, few following even the basic heart protocols and much resistence now to testing for syringomyelia (SM) as well.
Unfortunately the lifespan of dogs since even a decade ago has changed considerably. Since the time when your old stud would have been active, many breeders have focused almost entirely on using the same small number of sires (causing a high concentration of a few genes, an issue known as 'popular sire syndrome'). In addition, the two serious health conditions of MVD and SM have grown more widespread and at least with SM, more severely symptomatic. The reason is that both these conditions are believed to be polygenetic and sadly, it seems some of the tiny handful of founding dogs for the breed left by the 1940s had the basic genes for these conditions. As the breed is so closely interrelated and as these dogs were bred over and over to each other it has meant that the range of genes needed for the conditions to express themselves have become ever more likely to crop up and ever more likely to produce more sever cases of these conditions.
That means you best bet is definitely NOT to simply rely on what may have been good lines in the past -- many of these, some statistically research is indicating now, have long since lost whatever strength they had in general across the lines and now perhaps just certain branches of those lines, combined with other lines, will produce the pups most likely to live long and healthy lives.
The fact that you have had bad luck with so many dogs of different lines is a testimony to how huge this problem now is.
The best tool breeders have at the moment is both their knowledge of good historical lines, and health testing, not just for their individual sire or dam but ideally in related dogs. For SM, for example, there are clear indications that less affected or clear dogs come in clusters of related dogs.
Let's use Ronnoc as an example of how complicated genetics now are. According to a couple of researchers, it is likely that Ronnoc is one of only a tiny handful of older lines that has a very good probability of producing dogs at very low risk for SM. However according to a computer scientist and researcher running complex statistical testing on pedigrees, this is only for a certain BRANCH of Ronnoc and it entirely depends on what other lines may have been bred in since then.
The evidence? I have two Ronnoc dogs. One is a very rare all clear dog without either the skull malformation or syringomyelia. There are almost NO cavaliers with this profile amongst hundreds that have been MRId now. Yet this dog's half brother -- same father -- which I also own, has symptomatic syringomyelia. They have the exact same pedigree on one side and share some immediate ancestors on their mothers' sides -- but one is of the probable very good lineage and one has SM. Ronnoc True Luck is BTW considered to be one of the best dogs in the SM probability stakes but again it all depends on how his lines have developed since then. Sadly that line has ended more or less with the death of the breeder this year.
My personal feeling is that the best approach is to work with breeders who follow the MVD protocol and also MRI their breeding stock and work with the recommended SM breeding guidelines:
http://sm.cavaliertalk.com/breeding/breeding/breeding.html
Research so far is showing that litters from dogs fitting into the guidelines are producing litters with a very low incidence of SM and mild rather than severe skull malformation. Likewise breeders who work with the MVD protocol have found they can produce far lower incidence of early onset heart murmurs. However neither of these sad conditions will probably ever be eradicated because they are 1) almost certainly polygenetic and 2) so widespread.
Personally I think SM is a far more worrying development than MVD as MVD can be managed to a fair degree but SM is costly to diagnose and to treat and can be extremely painful for the dog. And it appears to be as genetically widespread as MVD.