Var diff on some pools starts low and goes up, on others it may start high and go down. It should only be a problem on pools that start with an unusually low diff.
From observation, Westhash start Diff for the S5 is 512.
I don't know about westhash specifically but a lot of pools are running var diff, which starts with very low diff shares and increases as the miner reports higher hashrate. If the miner can't run with a 'shit tonne' of low diff shares, its hashrate never climbs. As you say it would be great if the controller doesn't eventually lock up, but that's the reality. If you use it inappropriately then its not going to work as we want it to.
I've experience as a software engineer, and for the hardware to lock up due to excessive workload translates into faulty software to me. There should be safeguards against such eventualities -- the work queue can either be successfully met or it can't. Lockup indicates a defect (ie bug), I can't easily believe any real engineer would consider it otherwise. There have to be cleaner solutions than GUI lockup for work overload handling!
I've decided to regularly sample the Diff reported by my two S5s. So far the high has been 4.1K(4096?) and the low 512 with only a brief excursion to 256. I'm hoping to see just how low it gets. This recognizing I'm limited to periodic manual checks.