0
\$\begingroup\$

I'm currently trying to test my ADC chip using two signal generators. Below figure visually illustrates how the test setup would look like:

test plan for 40 GS/s ADC

The traditional approach uses a 10 MHz reference port to synchronize two signal generators, but I am concerned about its long-term phase and frequency instability.

The best alternative approach would be using 1 GHz reference by skipping 10 Mhz -> 1 GHz freq synthesis, but unfortunately, PSGs in the lab do not have this option (H1S).

E8267D in the lab has an HCC option, which enables phase coherency by sharing LOs. However, it is unclear to me if this option allows for two signal generators to have different output frequencies. For example, the figure below shows how the HCC option is configured within the signal generator. As the internal PLL is positioned prior to the LO input ports, It seems improbable to have two different frequencies at the output of two signal generators synchronized via HCC.

enter image description here

For reference, I am attaching another figure showing the rear panel of E8267D that we have in the lab:

enter image description here

Any advice would be really appreciated.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ It's not clear why frequency drift even matters if you're testing using the same oscillator for both sample and data since drift will be common to both and cancel out. The point of this setup configuration is that the absolute stability of the source won't matter during testing, just how well the PLLs (which are hopefully pretty good in high end test equipment!) can lock to it. Am I missing something about what you're trying to do? \$\endgroup\$ Commented Jun 13, 2023 at 1:03
  • \$\begingroup\$ drift "over time" will not be common to both, especially when you are using only 10 MHz reference for timebase synchronization. Your reference frequency will be identical, but internal VCOs inside each signal generator will have different phase noise profile over the time. There are at least three VCOs I can think of in typical signal generators. As you are multiplying frequencies, e.g., from 10 MHz to 20 GHz, you will experience phase drift over time badly. \$\endgroup\$
    – Emm386
    Commented Jun 13, 2023 at 1:06
  • \$\begingroup\$ Reference: keysight.com/us/en/assets/7018-02474/application-notes/… \$\endgroup\$
    – Emm386
    Commented Jun 13, 2023 at 1:07

1 Answer 1

0
\$\begingroup\$

I'm answering my own question for people who may encounter the same problem in the future.

After some research and asking questions to Keysight engineers, having different output frequencies using HCC options (sharing a common LO) is impossible. In this case, the cheapest way to solve this problem is using 10 MHz since 1 GHz REF IN/OUT cannot be retrofitted after the fact. The best way to solve this problem is purchasing a signal generator supporting multiple output channels with full coherency.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.