[time-nuts] time-nuts Digest, Vol 186, Issue 19

[Dan Kemppainen]

So, Just a few numbers to look at.

I'm currently starting at a GPSDO output. The DAC is a PWM DAC with 20 
bit resolution. Note the previous replies in regards to the PWM. Also 
note, PWM is used here as the DAC in the micro was so poor it was not 
usable to drive the EFC lead.

The DAC covers the whole tuning range of the OCXO. In 60 hours it's 
required on average 0.61 counts per hour of correction for long term 
drift (temp, aging, etc.).

In reality, since this is using a GPS as a reference and due to changes 
in temperature, the DAC has been 'yanking' the EFC to keep phase lock. 
That peak to peak has been +/- 40 counts in 60 hours. Again, that's at 
20 bits, with about 6.4V full scale EFC range, and about .645 Hz/Volt.

Nothing stellar here in terms of OCXO, GPS, GPS antenna location, etc. 
Depending on your OCXO and source, 20 bits may not be enough.

Hopefully this gives you some numbers that are meaningful in your search 
for a DAC.

Dan



On 1/14/2020 12:00 PM, time-nuts-request at lists.febo.com wrote:
> Message: 3
> Date: Mon, 13 Jan 2020 16:24:56 -0800
> From: "Lifespeed"<lifespeed at claybuccellato.com>
> To:<time-nuts at lists.febo.com>
> Subject: [time-nuts] DAC for OCXO disciplining
> Message-ID:<009001d5ca71$0ccc52d0$2664f870$@claybuccellato.com>
> Content-Type: text/plain;	charset="us-ascii"
> 
> I have been looking into all-digital PLL designs to discipline an OCXO, the
> scope of which would of course include GPS, as well as other reference
> sources.  A component of the block diagram would be the DAC converts the
> digital control loop output for application to the OCXO tune port.  In order
> to use such a circuit to improve Allan deviation, rather than degrade it,
> concerns like step size, Differential Non-Linearity (DNL), and possibly to a
> lesser extent voltage noise, are concerns.  Some thoughts that have occurred
> to me are coarse and fine DACs, possibly sigma-delta or pulse width
> modulation (PWM).  Or some combination of all the above.
> 
> Any suggestions on topologies and/or candidate parts?  One possibility for
> fine tuning might be the AD5791.
> 
> Lifespeed