OK, this is an odd one. I have a 34972A connected via USB to a Dell T-3500 Xeon PC running Win 7 32-bit in an automated test system. During the day, the system runs normally (~130 seconds/test), but sometime after 7PM, the data acquisition times and/or USB transfer speed is reduced by >40% of "daytime" normal (>180 seconds/test).
During the test, the 34972A performs various tasks, including voltage and frequency measurements. There are over 40 test steps, but not all tests are affected. Further, there's other USB devices outside the 34972A's world that also seem to be affected by this slowdown.
I'm unable (at present) to determine if some external (Windows related?) functions are doing background tasks after 7PM, such as McAfee or other programs.
I've used the 34972A, 34970A, and the 34980A in various test systems, but the USB interface is not very good for data transfer rates as compared to GPIB (yes I know, serial vs parallel), but the additional $500 cost for a GPIB interface forces me to USB.