[time-nuts] OCXO PLL gains?

[Magnus Danielson]

Hi Rick,

On 2019-02-02 17:49, Richard (Rick) Karlquist wrote:
> On 2/1/2019 3:22 PM, Magnus Danielson wrote:
>>
>> The Gardner book is the one book I recommend. Few books has core 
>> knowledge so well compressed. If one only gets one book, this would 
>> be the one I would recommend.
>>
>> Cheers,
>> Magnus
>>
>
> I'm glad someone mentioned this classic.  45 years ago, this
> was my first training in PLLs.  In the first 10 or 20 pages,
> I learned just about all I needed to know about designing PLL's.
> Design for a natural frequency around 1% of the sampling rate,
> set damping factor to at least 1, and I was off and
> running.
>
Exactly. If I would have to settle with only one PLL book, the Gardner 
book is it for sure. It may not be the best one for all people to start 
with, but they should mature up to it pretty quickly after the Best book 
if that is where they start.
> As far as PN "bumps" are concerned:  they are generally unavoidable,
> but you want keep them from being very peaky.  You either need to
> reduce loop bandwidth or reduce extraneous poles in the tuning
> voltage path.
>
Well, if your damping factor is too low, you will have a more 
prodominant peak due to the resonance. With a well-damped system, the 
noises add on top of each other at the cross-over and well, you can only 
do so much on a second degree system.
> You can easily build a SPICE model where voltage represents phase
> and simulate the loop, both open and closed.  The open loop
> response should have one and only one dominant pole.  That is
> what gives you a stable loop without a lot of peaking.
>
> Gardner has an interesting chapter on 3rd order PLL's utilizing
> double integrators.  Unless you are tracking out doppler shift
> on a moving signal source, you should never need double integrators.
Indeed. The improved slopes may be a reason to raise the level thought.
>
> We used a double integrator on the oven control loop on the HP1938A
> OCXO.  It was tricky to get it working correctly, due to
> a phenomenon known as "wind up".
The danger is that it is no longer stable for all settings, so that can 
catch you badly.
>   We also used a double integrator
> on the 5071A cesium standard, but cesium loops are NOT PLL's.

Indeed, double-integrator in FLL is a separate class of problems.

Didn't know it had double-integrator.

Cheers,
Magnus