[time-nuts] Another "atomic" clock question

Chris Albertson albertson.chris at gmail.com
Sun Mar 2 08:38:07 UTC 2014


Yes, there is some satisfaction is measuring to fine tolerances.  Of
all the physical measurements we can take, frequency can be measured
with the most accuracy.    You say 1ppb is enough.    Let's see a
billion is 1E9.  A 10Mhz signal is 1E7 cycles per second 9-7 is 2.
You can get to 1ppm simplify counting cycles for 100 seconds or less
than 2 minutes.

Had you needed 1E-12 level then  7-12 is five which means 100,000
seconds or about 30 hours.  So you get your answer in some amount of
time but likely it takes several measurement cycles to converge and
you oscillator must be good enough to be stable over that long of a
period.

You can speed up the convergence by a lot if you compare clocks using
a finer measure than full cycles, You can measrure small fractions of
cycles (phase) and then make an adjustment every second.

My opinion:  For 1ppm you will need any reasonable GPS receiver that
has a 1PPS output.  You can find them that are really good and put the
pulse out within single digit nanoseconds of the correct time but you
can use a GPS that gets to 100ns or even a few hundred ns.   An older
Motorola "encore" sells for about $20 on eBay.  Then you need antenna
and power  supply.  The guys who are going for the last ns will buy a
better GPS, place the antenna on a tall mast on the roof, measure all
their cables and look for an ultra clean power supply.  You don't need
any of that.

Look for the thread here on this list about the Arduino based GPSDO.
I think this will be the most economical one ever.   I'm thinking
"under $50". and that include a $25 ovenized crystal oscillator from
eBay.

Five years ago people would tell youth just buy a Thunderbolt but
prices have gone up 3X on those.  Maybe that was a good thing, now
peole will have to think a little harder.

On Sat, Mar 1, 2014 at 10:48 PM, Bob Albert <bob91343 at yahoo.com> wrote:
> Chris,
>
> Okay you want numbers.  Well, I think 10 ppb or thereabouts should do it.
> Somewhere there is a discontinuity in accuracy plotted against cost and I
> don't want to cross that barrier just yet.  If I can get 1 ppb without a big
> increase in cost, I'll take that.
>
> My need for this is nonexistent.  I am only interested in doing it for the
> fun of seeing all zeros on the counter and having it give me that
> repeatedly.  The pleasure of knowing I am as close as the equipment is
> capable is what I seek.
>
> I'm sure many time nuts feel the same.  I am not interested in offering a
> calibration service or tracking spacecraft or measuring the diameter of the
> moon.  How do I get accurate frequency from GPS?
>
> I have the same fetish regarding components, resistors and capacitors and
> inductors.  I have lots of good meters but am always looking for a better
> one.  I am trying to get six useful digits of voltage and resistance
> measurement and eventually want to do it with current as well.  Not so sure
> about temperature, mass, and force.
>
> Once I get where I want to be, I'll probably go into basket weaving.
>
> Bob
>
>
> On Sunday, March 2, 2014 5:46 AM, Chris Albertson
> <albertson.chris at gmail.com> wrote:
> On Sat, Mar 1, 2014 at 8:05 PM, Bob Albert <bob91343 at yahoo.com> wrote:
>> Paul, as I said I just want to know how close my crystals are and be able
>> to adjust them as well as they can be.
>
> Don't say "as well as can be" that can get expensive and time
> consuming.  You need to use numbers.  Say "and be able to adjust them
> at the 1E-8 level."
> Then you will get advice to just use WWV.  But what if you need
> 10,000 times better?  Then use GPS  After that it starts getting
> harder but you still are not up to "as well as they can be."
>
> I admit to a few years ago using a  50 cent TTL can oscillator as my
> "lab standard"  The part was salvage from some junk and was good to
> about 5 digits accuracy.  It worked actually better than I needed.  My
> RF signal generator was from the 1960's with a hand turned dial to
> adjust the frequency.  The TTL can let me calibrate the dial.
>
>
>
>>
>>
>> I probably will never go rubidium (note that I qualified that) but still
>> somewhere one has to decide where to set the frequency.
>>
>> I did WWV at 20 MHz for a beat of somewhat slower than one per second.  I
>> know the phase changes but probably not much in a few minutes, as the path
>> length doesn't vary very quickly.  And I don't need phase lock to them
>> anyway.  In the old days they had 25 MHz and even 30 MHz for a slight
>> improvement in settability if not stability.
>>
>> Bob
>>
>>
>>
>>
>> On Saturday, March 1, 2014 7:38 PM, Hal Murray <hmurray at megapathdsl.net>
>> wrote:
>>
>>
>>> I am trying to understand how this is done.  Should I ever get a rubidium
>>> standard, I'd want to check its calibration, and that's not a trivial
>>> exercise.
>>
>> If you assume your rubidium is stable, then it's pretty easy to check
>> and/or
>> calibrate.
>>
>> The trick is that you need someplace to stand.  A PC running ntp is good
>> long
>> term.  There is a tradeoff between good and long.  Good is ambiguous, but
>> both how-good is your PC clock and how good/accurate a measurement do you
>> want are appropriate.
>>
>> Probably the simplest way is to get one of tvb's preprogrammed PICs.
>>  http://www.leapsecond.com/pic/picdiv.htm
>>  http://www.leapsecond.com/pic/picpet.htm
>>
>> One approach is to use a picDIV to make a PPS and then monitor that.
>>
>> If you have Linux, you can feed the PPS to a serial port.  My hack for
>> counting 60Hz will work fine at 1 Hz.
>>  http://www.megapathdsl.net/~hmurray/time-nuts/60Hz/60Hz.py
>>
>> Another approach is to use a picPET and connect a modem control signal
>> from
>> the monitoring PC to the Event input on the picPET.  Then the data
>> collection
>> program grabs the time, flaps a modem control signal, grabs the time
>> again,
>> then grabs the text from the picPET and logs everything.
>>
>>
>>
>> --
>> These are my opinions.  I hate spam.
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at febo.com
>> To unsubscribe, go to
>> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>> and follow the instructions there.
>
>
>
>
> --
>
> Chris Albertson
> Redondo Beach, California
>



-- 

Chris Albertson
Redondo Beach, California



More information about the Time-nuts_lists.febo.com mailing list