[time-nuts] Two types of GPDSO / Rubidium

Fri Sep 27 00:46:27 UTC 2019

Hi

The gotcha is that (unless you have something very exotic) there *is* no one standard
for the lab. You may have a device that is quite good on phase noise at some offsets and
horrible for long term. You might have another device that is good medium term and also
pretty good for close in phase noise. Since a Cs is a primary standard, it will always win in
the “long term / free running” category. In the disciplined category a GPSDO will always 
come out the winner, if you wait long enough.

So if this is headed off in many directions, first step would be to set up a phase noise measurement 
system. Start comparing standards. Best to start with things you have more than one of. First 
approximation is that they each are just a bit better than the comparison of the two would indicate. 
Then start cross comparing devices. Eventually you will come up with “this one is best at these
offsets” sort of decision. Just how good it is …. time to buy another :). Why start with phase noise?
it is the quickest of the measurements to do. Start with gear that will handle the 10 MHz sort of 
frequency you now are dealing with. Expand the range as needed. Also be sure to watch the 
floor of your gear vs the measurement. You might need better gear :)

Next step would be to start looking at long term drift. That might start at 1 second or it might start
a bit longer. Again it’s a compare similar to similar at first sort of thing. Then start doing cross comparisons. 
If you stick with ADEV measurements, you can more easily compare to published data. There also is
a body of ADEV data in various posts to Time Nuts and on web sites run by Time Nuts. ( The same is
true of phase noise).

Test gear wise, something like the TimePod or the new “Son of TimePod” would be a fun device to
play with. Neither one is in the “dirt cheap” category. They don’t seem to show up on eBay for < $200….
That’s not in any way a knock on the retail price of either device. There are other pieces of gear out there
that do way less and cost way more. 

Bob

> On Sep 26, 2019, at 11:46 AM, Taka Kamiya via time-nuts <time-nuts at lists.febo.com> wrote:
> 
> Bob, et al.
> First, current line up of timing source:OCXODOCXOEfratom Rb of various kindPRS-10GPSDO of various kindDatum FTS4040/AHP5071A high-performance (DEAD tube)
> Time measurement gear:HP85132AHP5335AHP5370ATIC
> 
> My first and at most problem is, I don't know which one to trust and in what circumstances.  They all settle around 10E-10 to 10E-12 but since nothing stands out, (like Hydrogen Maser) I have no ways of knowing if the drift/error/noise I see is coming from DUT or the clock source.  I need to establish "the" lab standard which everything can be compared against.
> So what's my ultimate goal?  Just that.  There is no way for me to tell if I will get into microwave or something else.  Either way, having one reliable standard is a must.  I am aware, there is no such thing as "the" standard.  Various DUT and purpose calls for different standard.  It could be OCXO or Maser.  But still, I need to characterize them to know how bad/good DUT is.
> NUT case is already taken care of.  I'm a certified NUT.
> 
> --------------------------------------- 
> (Mr.) Taka Kamiya
> KB4EMF / ex JF2DKG
> 
> 
>    On Thursday, September 26, 2019, 11:56:56 AM EDT, Bob kb8tq <kb8tq at n1k.org> wrote:  
> 
> Hi
> 
> There are some “simple” answers to this:
> 
> 1) You can (and many of us do) set up to measure things yourself. For most setups this means
> comparing devices, but indeed Time Nuts seem to have piles and piles of devices. You can do this
> with shiny new gear (possibly for a lot of money), eBay versions of the same stuff (hopefully for 
> less money) or with home brew / DIY gear. Phase noise and the various xDEV’s (like ADEV) can 
> be measured this way. 
> 
> 2) You *do* need to figure out what the objective is. If it’s microwaves then the test gear needs to 
> run to those frequencies. If it’s just 10 MHz then not so much. If you do indeed anticipate having
> gear at the 1x10^-15 @1 sec level then your setup needs to be a bit fancier than if “only” 1x10^-11 
> is what you are after. 
> 
> 3) If this is cost constrained, be realistic about the costs. It is *very* easy to say “max $100” and then
> start talking about many thousands of dollars of gear. It is also easy to start with a budget and have it 
> creep on and on and on. Even projects funded by national governments (with far larger charge cards 
> than any of us have) hit limits in terms of costs. It is part of this at any level. 
> 
> It’s easy to say that you are setting up a full lab. If so, a hydrogen maser (possibly two or three of them)
> *could* be part of the mix. There *are* Time Nuts who have set up that way, but very few of them. Short
> of setting up that way, you will be in a compromise situation. (Indeed even a maser is a bit of a compromise
> compared to other more exotic gear). Some degree of focus will be needed. Few of us have a really good
> handle on that “focus” thing …..:)
> 
> Another wrinkle to this is that there are indeed measurements and terminology that are a bit unique to
> the field. Indeed the measurements and gear that were common in the 1950’s have evolved into something
> a bit different today. The same is true if you go back into the 1930’s. Different disciplines focus on different 
> measures, even today. Tying papers, spec sheets, or measurements from one era or discipline to another 
> can be challenging. There is no one universal single number that measures “best”. 
> 
> So yes, it’s complicated and it *can* be confusing. Welcome to why it’s called Time Nuts :)
> 
> Bob
> 
>> On Sep 26, 2019, at 1:29 AM, Taka Kamiya via time-nuts <time-nuts at lists.febo.com> wrote:
>> 
>> Tom,
>> I know you are absolutely right.  When I buy from eBay, I have no idea where they came from.  Reject pile, junk yard, or pristine clean room.  Unfortunately, I can afford up to certain amount.  I have lots of PRS10 but their origin is unknown.  Even if I buy 5065A, origin will be unknown and so as condition.  Recently, I called around local cal-lab to see who can help me validate these devices.  No one could.  I'm very tempted to buy T-bolt or PRS10 new but at $1500, it's quite hefty.  I may just have to.
>> I'm at learn as much as I can stage.  I have a Cesium but it's old.  4040/A  I have no idea how it was used or stored.  It would be nice to have one known good standard.  The best I can do for now is my array of GPSDOs.  (from unknown sources)
>> 
>> --------------------------------------- 
>> (Mr.) Taka Kamiya
>> KB4EMF / ex JF2DKG
>> 
>> 
>>     On Thursday, September 26, 2019, 3:05:19 AM EDT, Tom Van Baak <tvb at LeapSecond.com> wrote:  
>> 
>> It might be a bit of an over-simplification to split oscillators in 
>> those two camps, "fixed" and "agile". There are often many trade-offs in 
>> performance that you have to deal with. Moreover if you are getting your 
>> oscillators from eBay, or especially parts from China, you may also have 
>> to assume that original factory specs are no longer valid.
>> 
>> This is one reason why the time nuts list is so preoccupied with making 
>> accurate measurements of frequency standards. You only know what you 
>> have if 1) you buy it new from a reputable company, or 2) when you buy 
>> it cheap and measure it yourself with reputable test equipment, or 3) 
>> you get it from a friend who has measured it for you.
>> 
>> If you're asking specifically about Rb, I think the data shows that hp 
>> 5065A has best performance, followed by PRS10, followed by all the cheap 
>> telecom Rb. Sorry, I don't have plots or URL's handy at the moment to 
>> support this with real data. My main take-away is to start with your 
>> requirements, see which products / specs meet that, and if you're buying 
>> dirt cheap surplus, make your own measurements to validate your 
>> requirements.
>> 
>> /tvb
>> 
>> 
>> On 9/25/2019 3:45 PM, Taka Kamiya via time-nuts wrote:
>>> This is a cross post from EEVBLOG.  I'm hoping there is someone who's familiar with this subject would help me out here.
>>> I am hoping someone can help me understand this.  I've seen similar discussions take place both for GPSDO with Crystal Oscillators and Rubidium modules.  It appears there are two types of each.
>>> 1)  fixed frequency type (less jitter)
>>> 2)  frequency agile type (more jitter)
>>> 
>>> I've read frequency agile Rb modules (ones you can change output frequency) is one kind of Rb (sa.22c and fe5650, etc), and there is another one that you cannot change frequency. (ie. T-bolt, PRS10, etc).  Words like phase noise and PLL are thrown out often in discussions.  I vaguely remember frequency agile types are less suitable if ultimate in stability is needed such as multiplying the output into GHz range.  This discussion was about 10GHz transverter.
>>> Is this because frequency agile type has the ultimate output from PLL (subject to jitter) and fixed frequency type is from OCXO?  If this is the case, why frequency agile type even exist?  It's not like it can be used as a VFO (on radio).....
>>> 
>>> I'm sorry this is SO vague but that's the reason for this post.  I need to understand this.  There was a wiki page on this, but it doesn't go into this discussion deep enough.
>>> 
>>> Would someone help me gain knowledge in this?
>>> 
>>> ---------------------------------------
>>> (Mr.) Taka Kamiya
>>> KB4EMF / ex JF2DKG
>>> _______________________________________________
>>> time-nuts mailing list --time-nuts at lists.febo.com
>>> To unsubscribe, go tohttp://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
>>> and follow the instructions there.
>> 
>> 
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com
>> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
>> and follow the instructions there.
>> 
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com
>> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
>> and follow the instructions there.
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.