I can offer some measurements for the SDG6000X AWG.
To cut a long story short, for virtually all practical applications, the shape of the reference signal makes no difference.
For those who are interested in the boring details
, here's some more information on what I did:
I used my
Rb frequency standard based on a LPRO101 oscillator that provides a very good sine, and my homebrewn buffer / square trigger circuitry, as a reference.
It provides a sine output level of +7dBm. Alternatively, I used the 5V TTL square output. Both outputs have an impedance of approx. 50 ohms.
This signal was fed directly to the REF input of the SDG6000X via an RG58 cable of approx. 90cm length without additional termination at the SDG end.
The SDG6000X was configured to provide a 10.1MHz sine on channel 1 to an SSA3000X spectrum analyzer which was running on its internal reference. I always did three measurements:
- Trace A external sine reference
- Trace B external square reference
- Trace C internal reference
The attached screenshots of the SSA are self-explanatory by the file names though not really interesting. No intermodulation or the like could be observed, the worst that happens is a slight coupling of the harmonics of the REF signal to the output in the frequency ranges of 50~250MHz and then a little stronger around 1GHz. Considering the "forest" of peaks the SDG6000X produces when running on the internal reference, and the absolute amplitude of this noise which is close to -90dBm (except the clock feedthrough at 2.4GHz which is more like -70dBm), it's not worth worrying about.
FYI, I also recorded a set of traces with the output disabled which almost didn't change a thing. In this screenshot, I also added a fourth trace (D) with the generator completely powered down.
A test with the modulation domain analyzer (sorry for the crappy photos...) revealed a very slight improvement of the jitter / phase noise (standard deviation) if a square wave is used as a reference signal. An external sine ref signal and the internal reference are on par to each other once the instument has reached thermal equilibrium (except for the absolute frequency accuracy of course).
If all this information is also valid for the SDG2000X series is something that I cannot answer. So maybe someone who owns this AWG and the additional required gear may want to repeat the tests?
Cheers,
Thomas
Edit: Messed up one of the MDA photos...