[time-nuts] 3048A Expert Needed

John Miles john at miles.io
Tue Jan 20 18:28:40 EST 2015


> However, the 3048A won't achieve lock.  The 100 MHz PLL's loop bandwidth
> (about 0.2 Hz) is too slow for the 3048A to achieve lock.
> 
> Does anyone have any ideas?

Since the DUT's loop bandwidth is (presumably) below your minimum offset of interest, I'd try the "Phase noise without PLL" measurement type, and establish quadrature manually with a phase shifter or a string of cables and adapters.  You'll need to determine the Kphi factor manually, using one of the techniques described in the 3048A manual.  

Another possibility would be to temporarily hack one of the synthesizers to provide a wider loop bandwidth, so it can respond quickly enough to make a traditional phase-locked quadrature measurement.  As usual, if the 3048A's control loop BW is greater than your minimum offset, you will need to select the option to characterize the loop and remove its influence from the measurement.

> On the timepod, I've used the technique whereby I have two oscillators @ 70
> MHz and two double balanced mixers.  So I actually measure the 30 MHz
> offset.  In theory the phase noise of the two 70 MHz references should be
> cancelled out.  But I just don't achieve lower enough floor levels.

With a TimePod or 3120A, you will see better results if you can use downconversion LOs in the 90-110 MHz range to yield an IF of 10 MHz or less.  Under perfect conditions you can typically measure less than -175 dBc/Hz at offsets beyond 1 kHz.  But if, for example, you use a 10 MHz source at the REF IN jack to measure 30 MHz IFs at the Ch0 and Ch2 inputs, your 10 MHz reference will suffer an effective penalty of 20*log10(3), or about 10 dB.  Noise at the REF IN jack can't be reduced by averaging since the Ch1 and Ch3 inputs are tied together internally, so you need to use the cleanest available reference for downconversion measurements.  

For instance, you'd need a hypothetical noise floor of -185 dBc/Hz from a 10 MHz reference to measure a 30 MHz DUT at -175 dBc/Hz, and that will still come with a 3 dB penalty.  At the same time, the TimePod's real-world noise floor gets worse above 10 MHz (actually the best results I've seen are achieved around 5 MHz.)  So 30 MHz IFs are undesirable when you need to measure ULN-class sources, even if you happen to have a 30 MHz ULN reference. 

-- john, KE5FX
Miles Design LLC




More information about the time-nuts mailing list