[time-nuts] Advantages & Disadvantages of the TPLL Method

WarrenS warrensjmail-one at yahoo.com
Fri Jun 18 19:58:20 UTC 2010

Charles posted

> its operation needs to be characterized if technically oriented folks are 
> to be expected to take you seriously.

If a simple BB that works as well as that one, does not already do that, 
then nothing else I'm able or willing to do or say is going to change much.
It is a good thing that people can accept things when shown that they work, 
even when they do not fully understand how.
Or else there would no acceptance of much of anything including gravity, 
computers, email,  women or Windows.

So your point is?  and why do you think I care?  Maybe you are missing the 
point of my intended audience.
After all, I'm not the one that is lost or the one that does not know now it 
works. I have a working unit that, I know how it works and how to test it.

> how you determined phase locking down to femtosecond levels, & ...

You do indeed have many valid technical questions that I'd try and explain 
in a way that you could understand,
IF my goal was still to try and educate you, or if I thought there was ANY 
chance I'd be able to do so.
You seem to again of missed the point of my last answer,
One with enough technical ability can see what John's data says from his 
noise floor measurements.
If you do not understand that or how I'm doing it by measuring the PD 
output, then
I can not teach or show you how, even if I still wanted to try. My past 
failed attempts prove that.
So I'll just continue to do what I do best, and that is to make and test 
TPLL BBs for myself.

If you were thinking that I am one that can or am willing to try and teach 
you basic or advanced anything, then you are greatly over estimating my 
abilities and/or my patience. You need to look else where if you want to 
read a fancy math paper on how and why this all works or how to measure it.
If you can't find one to your liking and still want to know how to measure 
fs stuff, I'm sure if you ask nicely in a new thread, others would be more 
than willing to help you out with something that you could understand and 
Then again if I missed your point and your only goal is to verify if I now 
how to do it, then I can save you some trouble,
Yes the TPLL and I know how to measure fs phase differences, and we both 
know how to integrate along with a lot of other basic things including even 
adding two plus two.  (and I doubt that you understand that last point 
either, & Don't take it so personal, It is likely not just your problem)

BTW, one of the other points you seem to be missing in how I can measure & 
test things so easy that others can find so hard to do.
After all I do have a big advantages over most, I can use another one of my 
GP PLL BB as a tester.
As I've point out before, they can do much more than just ADEV.
But then you would not be expected to know that without an advantage list.

Seems like it is again 'Time to Push the Reset.Button' on this thread's 
subject, cause this has got way off the subject.
As long as it is so far off the subject of advantages & disadvantages, I'll 
add, in response to what others have said both on and off line.

IF others want to build a TPLL using buffer amps, or VFC, or difference Ref 
Osc, or multiple Ref osc, or a digital version or with cross correlation, or 
using different software, or different algorithms, or different connectors, 
or more parts,  or more expensive parts or cheaper parts, or that works 
better with some imaginary unreal data set, or over a wider freq range, or 
over a longer tau, or at a lower level, or with less injection locking, or 
any of the other thing's that have been brought up,
by all means, Go for it. I've tried to encourage others to do it their your 
own way. No single solution is best for all situations.

What several of the suggestions show is that many do not yet know how simple 
a TPLL can be made or do they even understand exactly how the 3 basic parts 
work together.
 I have consider all the suggestions and tried many of them and so far have 
found the variations unnecessary for my applications.
Also I have not heard about any H/W that others have built, only a lot of 
criticism from some about what I've done or said or not said.
No problem, If others do not like what I've done or the way I've done it or 
tested it, even when many of them do not know what it is or how it works,
by all means they should do it there own way.

What I've done is to test one of my simple TPLL versions and show that it's 
performance is good enough to be limited by the OCXO.
That is all, Don't read more into my comments than that.
The simple TPLL does not go down to 1e-15.  What I said is that is the limit 
of the low cost AMP that I choose, and it insures the amp has no significant 
negative effect.
Of course all is not perfect, nor is this is the best that a TPLL can be 
made, not even close. But it is good enough for me.
I know how to made it better, much better, with lower noise, more 
resolution, faster, lower tau, smaller or bigger, more costly or lower cost, 
and on & on.
So what?, the achieved goal of 'KISS' was to keep it small, simple and low 
cost,  and make it good enough to test the high end OXCOs that are 

This simple $10 version (my ebay cost) of a TPLL tester using just a single 
powered active part (not counting the Ref or the ADC or the PC) has plenty 
of limitations but it  has been show to be good enough to closely match a 
TSC 5120A over a wide range of signals.
By all means if your needs, skill levels, experiences, desires, goals, and 
junk boxes are different than mine, and you want to make something different 
or better, then by all means 'Go For It'.

I'm not trying to tell anyone what method to use, or what they should build 
or how to build one OR how to test it.
I'm just trying to get a list posted to show the many unknown advantages and 
disadvantages of the TPLL methods.
Because so few have had any experience with this method and therefore have 
little or no practical knowledge about it.
How can one go about deciding it is something is useful for them without 
knowing  what the advantages and disadvantages are?


[time-nuts] Advantages & Disadvantages of the TPLL Method
Charles P. Steinmetz charles_steinmetz at lavabit.com
Thu Jun 17 15:02:12 UTC 2010

Warren wrote:

>>I'm curious how you determined that the oscillators are being held
>>to within femtoseconds of each other.
>I done  it several ways including measuring the PD output.
>You seem to be missing how insignificant an 1-e6 injection locking
>to EFC gain ratio is.
>I can't detail, to your satisfaction, all the hundreds of test that
>show no significant effect of so many different things.
>For an independent test that may help you with things you missed see:
> http://www.ke5fx.com/tpll.htm

There you go again, accusing someone else of missing something.  I
have never said I thought your method doesn't work.  I said its
operation needs to be characterized if technically oriented folks are
to be expected to take you seriously.  "See, it works!" is just not
sufficient characterization to achieve this, IMO, and that is all you
seem to offer -- references to John's tests included.  I inquired how
you determined phase locking down to femtosecond levels, and all you
can say is that you "measured the PD output" among other
ways.  That's like a cosmologist saying (s)he "measured the positions
of some stars, among other things" to determine the gravitational
constant, which is fine for an interview in People magazine but
inadequate for a technical audience.  What phase detector did you
use?  What are its characteristics and limits?  With what did you
measure its output?  What results did you get (at the very least, a
descriptive summary of your dataset)?  How do those results support
an inference of locking to within femtoseconds?  What other ways did
you use to determine locking error besides measuring the PD
output?  You imply that the ratio of the injection gain to EFC gain
in your system is 1E-6.  How did you measure the injection
gain?  What were the results (at the very least, a descriptive
summary of your dataset), and how do they support an inference to the
1E-6 figure?

Best regards,


More information about the time-nuts mailing list