[time-nuts] TPLL secret reveled
Tom Van Baak
tvb at LeapSecond.com
Thu Jun 10 20:14:11 EDT 2010
Where does the "1e-15" figure come from? Is that phase and
are the units seconds or radians? Or is that relative frequency
at the mixer, if so over what tau does it refer to?
I'm curious how you measured or calculated the 1e-15 number.
Saying "very nearly zero" doesn't mean much to me. In science
we use numbers; often scientific notation. So in this case does
nearly zero mean one microvolt? One nanovolt? One picovolt?
Sorry if I'm dense here.
----- Original Message -----
From: "WarrenS" <warrensjmail-one at yahoo.com>
To: "Discussion of precise time and frequency measurement" <time-nuts at febo.com>
Sent: Thursday, June 10, 2010 4:16 PM
Subject: Re: [time-nuts] TPLL secret reveled
> The ADC is measuring the EFC averaged voltage signal.
> I think all understand that the EFC controls the Frequency of the Ref OSC.
> This EFC voltage (freq) is what the external S/W uses to convert the two
> oscillator differences into ADEV.
> The Phase difference at the phase detector output is held to zero (within
> 1e-15 or so) by the gain of the AMP.
> The phase detector output looks like a summing junction, nothing is going to
> happen to make the point change from VERY nearly zero.
> So there is NO phase change, because phase is always at a constant 90 deg
> between the two oscillators.
> sorry I can not explain it better than that, I'm sure other can.
> Tom asked:
>> Fact #1) the TPLL uses Freq not phase,
> Warren, can you please clarify. From what I can see it actually
> uses a Minicircuits phase detector, right? That suggests what is
> being observed is a phase difference between the DUT and the
> REF, not a frequency difference.
> That analog phase measurement is then filtered and amplified
> into a pre-calibrated EFC correction to the REF which over a
> short time reduces the phase difference as seen by the phase
> detector. The digital stream comes from an ADC is connected
> to the EFC.
> So what does it mean to say "TPLL uses Freq not phase"? It
> looks to me like it uses both.
More information about the time-nuts