Hi
After reading in Suffer1981de's post about the DS1000Z firmware 00.02.03.SP5, I e-mailed my scope dealer and immediately received the new firmware. Although "new" is probably relative: seems to have been released on Jan. 26th, already. Haven't yet looked for changes or bug-fixes but noticed that the measurement menu on the left side has gained the 8 new measurements that are documented in the User Guide from Jan 2014, which is online on Rigol's website for quite some time, already.
Here is a link to the firmware: http://www.filedropper.com/ds1000zupdate-v000203sp5
Have fun. :-)
This link seems dead. Does anyone have a working one? Is this update legit?
EDIT: Also, I know that the 500µV/div option doesn't seem to work for anyone, but my understanding is that you can set it, but the signal is offset wrong. I enabled the option (or, at least, I fed it a code it took that should have), but I can't seem to go down that low, my lowest vertical setting is 500mV. Incidentally, that setting works.
Oh. Duh. Probe attenuation factor. Set to 1, I see 500µV. Okay, so I've definitely noticed the problem where it flies off screen never to return, but depending on tongue angle it seems to work fine.
That said, I don't think there's anything special going on here: I notice that, with my vertical at 500uV, the discrete ADC steps are clearly visible. They look to be about 60µV tall. When I switch to 1mV/div, I still see steps, about half as tall on screen, which is to say they still look to be 60µV tall. It seems that, in both 1mV and 500uV/div modes, the ADC is scaled to 15.36mV p-p full scale, which actually is close enough that it seems believable that the 2mV/div range is the lowest range where full scale is actually all 8 divisions on screen (and the descretization steps are actually 62.5µV and not 60 as I roughly measured with cursors).
Not to say I'm too bummed out about that, a closer view is still helpful. But I don't think your input resolution is actually improving any when lower than 2mV/div, you're just seeing a digitally zoomed-in picture. If that's true, why not 200µV/div as well? Though at that point you'd only be seeing 3.2 steps per division, which would be almost more misleading than useful.
EDIT again: By the same token, I kind of which I had a 1ns/div timebase setting. Though of course it'd only represent one sample per division.