There were some questions about the Rigol DS1000Z PLL in another thread, and most of the arguments were coming from this topic, so I read the PDFs posted at the beginning, then the whole thread, and decided to ask here, even if it's been a while since the last post.
By the way, very good findings and well documented, thank you very much for posting them,
Bud.
For testing my DS1054Z jitter at 5us distance from the trigger point, I used the latest firmware (which is supposed to fix the PLL issue in software by using better values for programming the PLL), and indeed at 5 us distance the Jitter is not visible any more on the screen.
Still, if I increase the distance from the trigger point long enough, there still is some jitter, and the jitter is apparently increasing linearly with the distance in time from the trigger point.
I measured the jitter by setting the display to memorize the trace on the screen for the last 10 seconds, and try to guess the best spread in time of the edges by averaging 5 different captures. Both instruments were on for a couple of hours at constant temperature before measuring. Here are some values measured with a square signal from a Rigol DS4102 DDS (that is supposed to have +/- 2ppm frequency accuracy, and a max jitter of 500ps according to its specs) using a 50 ohms cable and an external terminator at the oscilloscope's BNC.
As can be seen in the attached captures, the jitter is increasing almost linearly with the time distance from the sync point. Why is that? Isn't the PLL jitter supposed to be random and independent of the test signal, so the jitter of the displayed test edges wouldn't accumulate with the distance from the trigger point?
The PLL is trying to sync with the internal oscillator, which is running continuously and independently from the test signal, so if the PLL jitter is randomly distributed, then I expect to see on the screen the same jitter for any edge of the test signal, no matter how far from the trigger point is the displayed test edge.
Why does the displayed jitter of the edges from the test signal increases (almost linearly) with the distance (in time) from the trigger moment?