[time-nuts] TPLL secret reveled

[WarrenS]

There has been speculation on why the low cost, simple, TPLL tester compares 
so favorably with the TSC 5120A over a wide range of taus, even when  using 
different Oscillators. The theories range from pure luck to some special 
secret magic hardware.

If I was that lucky, I'd be doing Lottery and not ADEV.
And If I could build some secret simple Hardware, and was able to adjust it 
to give some desired effect on future random noise, I'd be in politics or 
the stock market, not making ADEV Breadboards.
There is another more likely explanation that is within my limited 
capabilities, that many refuse to consider.
That is this method just works correctly without doing anything very special 
at all.

The reason that the simple TPLL works so good but is hard for some "experts" 
to accept, seems to come down to the fact that this method uses Frequency 
and not Phase to make the raw data log used to then calculate ADEV data.

Most if not all other methods depend on Phase differences (i.e time 
differences) to calculate the Frequency difference.
The simple analog TPLL method holds the Phase difference to zero (with-in 1 
femtosecond) and calculates the average Frequency differences by  measuring 
the voltage on the EFC of the reference Oscillator. That way no variable 
phase is involved.
This filtered Frequency data log can then be used to calculate ADEV without 
first needing to convert from or to phase.
The fact that average Frequency differences do not start out as wide 
bandwidth Phase, means that all the fancy write-ups that tell how best to do 
that, do not apply.

Some have argued that Phase and freq are the same thing, so the fact that 
this uses freq would make no difference.
Others have argued that one can not average Freq, so you have to convert to 
phase first.
Both have a basis of truth for some methods, but neither is correct here, as 
proven by the end results.
 (unless one wants to also assume that I'm either the luckiest or smartest 
person in the universal for getting it right the first time.)

While it is true that the TPLL Freq data can be turned into accurate 
controlled Bandwidth Phase data, Wide band phase data can not be so easily 
turned into controlled limited Bandwidth, zero dead time, integrated Freq 
data, which is the thing that is needed to get good ADEV results.
The reason is, to get high resolution, such as 1 femtosecond (1e-15) phase 
data, it takes a  wide bandwidth, which means high noise and many other 
problems. To get 1e-15 freq data from the EFC (limited by the ref osc), this 
can be done with the B/W equal to tau0.

The short Over simplified summery is:
At 1sec Tau0, the up to 1e15 to one difference in bandwidth between Phase 
and Frequency measurements makes a big difference that some seem to be 
missing. (think Phase trigger jitter)
The simple TPLL has No secrets, just some basic differences that seem to of 
been forgotten or not considered.  The way that frequency is measured is 
what is different between the TPLL frequency method and the other more 
popular phase methods.

The fact is, that the simple TPLL in spite of it limitations, has been shown 
to work just fine in the real world.
 If one can not understand or accept that measuring frequency without 
needing to first measure phase is the real secret reason why the TPLL works 
so good, without needing any special H/W or S/W 'adjustments', then feel 
free to propose a better reason besides Luck.

ws
-------------- next part --------------
A non-text attachment was scrubbed...
Name: TightPLL_BB7.gif
Type: image/gif
Size: 40614 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20100609/4d9067b8/attachment.gif>