Is there any reason, other than a defective product, why the logarithmic sweep on the 33521 would sweep at way less than logarithmically for periods over 20 seconds, and then go to the Stop Frequency in the last second. I just used the unit out of the box. It is particularly egregious for long sweeps. When doing a 20o second sweep from 10 Hz to 10kHz it didnt even make it to 100 Hz before the last second, and then it goes right to the Stop Frequency.
I had one of my colleagues test this out since I'm out of the office and they did not have any issues.
What waveshape is being used? If it isn't a sine wave, please switch to a sine wave and repeat the experiment.
Are the hold/return times set to 0 seconds?
How is it being measured? Best to use DC coupling and a fixed trigger level, because at slow frequencies (<100 Hz) you can get some strange behaviors with AC coupling or auto triggering. This is especially true if using a counter to make the measurement.
If you still have the unit maybe you could send me a state file, so I can try and repeat this? So far I am not seeing these problems, so it definitely could be your unit, but maybe I am just missing a setting somewhere.