[time-nuts] fast freq. synthesis schemes

Lux, Jim (337C) james.p.lux at jpl.nasa.gov
Thu Oct 15 00:19:36 UTC 2009

> -----Original Message-----
> From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On Behalf Of John Miles
> Sent: Wednesday, October 14, 2009 4:44 PM
> To: Discussion of precise time and frequency measurement
> Subject: Re: [time-nuts] fast freq. synthesis schemes
> Pretuning is the right strategy, but for microsecond agility, YIGs may be
> the wrong choice due to their main-coil inductance.
> If I were building an agile 12-18 GHz synthesizer I'd try a heterodyne
> scheme with varactor-tuned oscillators and a fixed (or very coarsely tuned)
> YIG or DRO.  Either way, you would probably use a sampler, such as the parts
> in the Aeroflex/Metelics catalog, to construct the outermost PLL.  Suitable
> counter and PFD chips exist as well (Hittite etc.) but samplers are cheaper
> and easier to use if you don't mind designing the IF circuitry for them.

Ooohh.. Sampling Phase Detectors or Harmonic Mixers.. The problem is that you have to hit them with a lot of power on the reference port (+20dBm wouldn't be unusual) Depending on your application, making that much LO power that is suitably quiet is a challenge.  Presumably, though, you're not DC power limited, so that helps.

Getting a PLL with simple single integer N is pretty easy with the Hittite parts, especially if you can tolerate N that is a multiple of 4 or 8.  There's a paper out there by S.K.Smith, et al., that describes a breadboard PLL we did at 8GHz, where we drove the reference input with the output of a DDS mixed with a fixed signal.
take a look at Figure 6 and 7

However, that won't change in less than a microsecond (the loop bandwidth is too narrow).. you could widen up the loop bandwidth, but the reference source would need to be quieter (not a challenge.. we didn't take any special efforts to make our DDS quiet, etc.)

The earlier paper by Cook, et al., 
shows a more traditional SPD DRO PLL, and gives some performance analysis of the loop.  Just as in the Smith paper, tuning speed wasn't a big deal for us, but the theory is there to generalize it.

> I would use a DDS, but only for fine tuning in a summing loop.  E.g., use a
> DAC to pretune the varactor or YTO to within 50 or so MHz, feed the sampler
> LO port with a clean 100 MHz crystal, then close the loop by comparing the
> sampler IF to the DDS-generated offset signal. 

This is a nice technique. The trick is in making sure you lock to the right comb and the right side.. But the external DAC can help with that.

 That way the PN is dominated
> by the lower N factor assocaited with the 100 MHz comb, and the resolution
> is determined by the DDS.


More information about the time-nuts mailing list