Rigol DG4162 phase noise

[Matthias Toussaint]

Just in case someone is interested, I've attached some phase noise measurements from the Rigol DG4162 waveform generator. Actually not too shabby. First harmonic is 50dB down, which is ok. There's kind of a hill around 100kHz. Not sure what that is. Maybe the PLL for the internal clock.

[JackOfVA]

Just for curiosity, if you have the equipment, check the internal 3rd order intermodulation products.

Set Ch 1 for some frequency, say 3 MHz and Ch 2 for 4 MHz and combine the two outputs in a hybrid combiner. Probably will need an isolation pad between each output and the hybrid combiner.  The look at the output with a spectrum analyzer for 3rd order products at 2 and 5 MHz.  This will tax your spectrum analyzer as well, and it may be necessary to use a notch filter to remove the 3 and 4 MHz test signals, as well as narrow RBW and extra attenuation on the SA input.

One of my main products lines is ultra high performance low noise figure radio preamplifiers with 2nd order intercepts of +100 dBm or more and 3rd order intercepts > +45 dBm.  I had hoped to use my DG4100 as a source of test tones and retire, for that purpose, the pair of HP 8657A generators, auxiliary amplifiers and attenuators, etc. Alas, it seems that is not possible as the DG4100 has an internally generated 3rd order intermodulation product that is above that produced by my amplifiers.  This is a "problem" that probably affects half a dozen people in the world who own a DG41xx series generator but I am curious whether my observations can be duplicated with a different instrument.

[Matthias Toussaint]

Not sure if that is what you wanted to see. I'm back home (The FSUP is at work) so I have to do the test with my DSA815. This is probably more a SA test. The 3 and 4 MHz signals are padded down by 10dB before the combiner. At the input of the SA is another 6dB attenuator. The 2Mhz seem to be around -85dBc and 5MHz around -72dBc. I don't have a notch filter at hand to knock down the 3/4 MHz carrier, so this measurement might be useless