[time-nuts] Noise Floor

[John Miles]

> Is there a _done_ solution to expand a Timepod to
> ~ 100 MHz, narrow-bandish is OK. I'm willing to solder,
> but I'm not ready for yet another development project.
> Not now.

A pair of double-balanced mixers driven by independent signal generators works well.  The mixer RF inputs are driven by a splitter, while the IF outputs go to channels 0 and 2.  Add gain/loss as needed.  

Independent bandpass filters at each mixer RF port can be helpful, since common-mode noise on the unwanted sideband can cause artifacts.  It will usually (but not always) be fairly obvious when you have problems with this.  

> When I've flushed my project pipeline, I might consider
> something serious with JESD-204B ADCs and such.
> That would require a larger FPGA with GTX transceivers
> and maybe a Beaglebone as a controller.

Some of the newer parts are pretty tempting, but note that the need for wideband acquisition is driven by offset frequency range rather than carrier frequency range.  Most noise measurement applications won't need 100 MHz-1 GHz of acquisition bandwidth, unlike the wireless base stations and high-end signal analyzers that the ADC manufacturers are targeting. 

-- john