All I know is that I tried the following signal generators (all at 20 MHz):
HP 3314A
Wavetek 288
HP 3335A
HP 8782A
The first two are generic function generators and the last two are RF synthesizers with ovenized oscillators.
These four were all tried against the following digital scopes:
Rigol DS1104Z
HP 54610B
HP 54112D
Fluke PM3390B
Tek TDS350
Every scope but the Tek showed the following:
no jitter at 0 delay with any of the four sources
plenty of jitter at sufficient nonzero delay, but ONLY with the first two function generators
no jitter at sufficient nonzero delay with the two RF ovenized synthesizers
Now, all the non-Rigol scopes use equivalent time sampling and it seems clear that in the absence of a stable source, such as the synthesizers, the combination of sampling and delay leaves one with a "jittered" presentation after the wandering waveform builds up.
Apparently, the Rigol is doing some analog of equivalent time sampling (
) since I get exactly the same behavior - OK only with the synthesizer sources.
Incidentally, the Tek scope passes the test because it cheats - it won't let you set the delay large enough to cause a problem.
Anyway, I feel a lot better now knowing that my 1980-1990 scopes are just as good as my 2014 scope!