I figured out the VSM dumping issue- I was waiting the extra time (80us at 200KHz = 16 clocks) after the dummy read in the initialization, but not after the dummy read after setting the address. I guess starting to read early interrupts the address load. Now all the VSMs are dumping as expected. I'll dump the 2 chips on the working S&S just to make sure there are no surprises, and I'll blowtorch the '351 and '352 from the other S&S just to take a look at the dice.