[time-nuts] Tight PLL Tester
WarrenS
warrensjmail-one at yahoo.com
Wed Feb 10 21:29:29 UTC 2010
If there are any Nuts out there interested in helping to make available to
other Freq-Nuts a SIMPLE tester that I have found to be a VERY useful low
cost tool, contact me off line.
warrensjmail-one at yahoo.com
The tool is based on an OLD but seldom used method called the "Tight
Phase-Lock Loop Method of measuring Freq stability".
For a block diagram and short description see Figure 1.7 at
http://tf.nist.gov/phase/Properties/one.htm#oneone
What I have made for my own use is a bread board of a simple analog version
of the NIST's block diagram.
There are of course many different ways to actually build it, depending on
ones preferences, skills, and junk box.
It can be done using a DVM, or a high or low resolution ADC, or a freq
counter, or counter IC chips, a Pic or any simple micro, or a sound card,
Or many other ways.
The nice thing about the method is that it takes no expensive or critical
parts to get performance as good as most anything out there.
Its main performance limitation is ONLY the single EFC OSC used as the
reference.
My unit works 'Good enough' to be able to test many of the things that
Freq nuts are concerned with.
Basically it is nothing more than a high speed freq difference detector that
can detect VERY small freq changes in a very short time.
What one then does with that data is where the flexibility comes from.
I've used mine for AVARs plots, detecting very small freq modulation due to
PS noise, freq offset plots, setting an osc on freq in seconds instead of
what can take hours, GPSDO Noise and TC test, etc. etc. The list is
almost endless.
Some advantages of a tight PLL method are:
1) It is very simple, cheap and easy to build, and small
2) Works well for comparing an Oscillators Freq offset, Freq Noise, Freq
modulation, over short time intervals
3) It provides very good sub pico Second Phase resolution even with simple
setups.
4) Its noise floor is low enough so that its limitation is the Reference
Osc.
5) The NIST says it can be used to one part in 1e14. I'm getting better
than 1e12 from it, limited by the HP10811 Ref Osc I use.
6) Would be easy to make into a PC board project for Time nuts that don't
access to all the high end equipment.
7) I have a working breadboard that I built from just my junk box parts that
has worked great for me for several different things
Some of its disadvantages:
1) It is not the best way to take long term phase drift differences, where a
simple phase difference device will work great.
2) It is not a DMDT and is not as flexible in many ways, but can be just as
accurate and a lot easier to build and less to go wrong and a whole lot
cheaper.
3) It is basically an analog device and does not have digital accuracy. But
for small freq differences, it is more than accurate enough to provide great
results.
4) For those luckily enough to already have a TSC5120A or better, you don't
need one, That is unless you want to verify its performance at short Tau.
5) And maybe the biggest disadvantage is that many of the leading Freq nuts
on this site don't like it and seem to believe that it should not work.
But maybe that is just because they don't have one and have not even tested
one and seem unwilling to give it any consideration.
6) A search of past post on the subject will show that many do not all agree
that this is a good idea,
but they don't have a working unit like I do, and I don't have the expensive
high end equipment that they have.
ws
PS
Sorry for the long post.
This is the best I can do to respond to the off line request I received to
find a way to make this subject more useful, constructive, cooperative and
less confrontational, and do it with less words and give more details.
I am not looking for a list of all the possible ways that it can be done
wrong,
Or guesses on why it should not work as good as it does.
I'll leave that for others to discuss.
But if what they say does not agree with my experimental results, you can
bet I'll still comment on it Again.
*************
More information about the Time-nuts_lists.febo.com
mailing list