[time-nuts] Phase noise measurements on the cheap with an Agilent E4406A VSA Transmitter Tester?

John Miles john at miles.io
Sat Jun 18 16:01:59 EDT 2016

> and looked at what was used to make the phase noise measurements. It was an
> Agilent E4406A. The noise floor is nowhere near as low as the more
> expensive instruments, but the E4406A is available for under $500, which is
> more than two orders of magnitudes cheaper than an E5052B.

It appears that Leo's using a notch filter to remove the carrier before measuring it with the E4406A, so it's not quite a plug-and-play sort of measurement.  But yes, the E4406A is a really cool piece of gear given the prices they sell for.  It's not meant to be a general-purpose spectrum analyzer -- and Agilent went well out of their way to make sure of that -- but it can still handle many common SA measurement tasks including SSB noise.

The plot on that page came from my freeware phase noise app from http://www.ke5fx.com/gpib/pn.htm , which is a (very) distant ancestor of TimeLab.  The last two FAQ entries at http://www.ke5fx.com/gpib/faq.htm offer some hints for suppressed-carrier measurements that can be used with the E4406A and other analyzers.  There's nothing special about the E4406A with respect to this type of measurement, except that it's an unusually cost-effective way to get the job done.  

I've also heard of people opening up the E4406A and feeding HF signals directly to the ADC, eliminating the LO noise contribution but not the ADC's white noise, 1/f noise, or clock jitter.
> I'm wondering if there are other more suitable commercial instruments
> around that don't cost a fortune, yet would allow lower levels of phase
> noise to be measured. I tend to preference HP/Agilent kit, as it is better
> supported, both by the manufacturer and places like the HP/Agilent Yahoo
> group.

The E5052A/B's most immediate predecessor was probably the HP 4352A/B.  They were made specifically for VCO and PLL transient analysis and noise measurement, and they've been selling in the $1K-$3K neighborhood for several years.  Their measurement floor is better than a conventional spectrum analyzer, but still not adequate for "time nuts"-class measurements on 10 MHz sources. 

-- john, KE5FX
Miles Design LLC

More information about the time-nuts mailing list