[time-nuts] Practical considerations making a lab standard with an LTE lite

Bob Camp kb8tq at n1k.org
Sun Nov 23 21:45:01 UTC 2014


Hi

If your target frequency error is in the < 1x10^-10 to the “hopefully 1x10^-11” range, You should consider your very requirements carefully. I tossed up some frequency plots of the KS boxes and of the Z3801 a while back. They are OCXO based boxes running in a very good thermal environment.   Their OCXO’s ADEV is roughly 1x10^-12 at 1 to 10 seconds. That compares directly to the TCXO’s apparent ADEV posted by Tom earlier at 5x10^-11 in the 1 to 10 second region.  

The OCXO based parts (with a very loose interpretation of 1x10^-11 frequency accuracy) do not hit < +/- 1x10^-11 frequency accuracy. If the plots are to far back to dig up, I can re-post them. They will hit a < 1x10^-10 frequency accuracy limit without any quibbling over the definition of the spec. The frequency accuracy of a TCXO based part is not going to measure up to an OCXO based part. That’s not because the TCXO part is in some way flawed, it’s just the way things work on a GPSDO. 

If you are going with a TCXO, concrete bunker construction is not needed. With an OCXO based part, it just might help a bit. This may be a bit counter intuitive. It’s a function of where the (much better) ADEV of the OCXO intersects the (constant slope) ADEV of the GPS receiver. The control loop on the OCXO based part will be running at a *much* longer time constant. If the OCXO ADEV is 10X better, it will be 10X longer. If it’s 100X better it will be 100X longer.  In both cases (TCXO and OCXO) the ADEV at 1 or even 10 seconds will not be improved by thermal this or that , once drafts are eliminated. The filter will still track where it needs to track. If the OCXO is running a filter out at a thousand seconds, you will *will* see slow thermal variations. The TCXO based part’s output running at (say) 10 seconds will not see the same variations, they will be corrected out by the GPS before they hit the output. 

—————————

Why tie these things together? 

Where you wind up depends very much on where you are headed. Starting with the right gear for the application will matter in the end. Putting a lot of effort into a project without considering the ultimate goal may not be as economical as it could otherwise be. 



Bob


> On Nov 23, 2014, at 3:04 PM, Jim Sanford <wb4gcs at wb4gcs.org> wrote:
> 
> All:
> 
> I appreciate all the responses to my post earlier today.  Very informative.
> 
> First:  DownEast Microwave sells a nice kit for distributing 10 MHz.  Specs are on their website, but basically, one in, four out -- each individually buffered and filtered.
> 
> Second:  I will use the 20 MHz from the LTE-Lite to lock a 100Mhz TCXO which will be the LO for a high performance 2meter amateur software defined radio.  (OpenHPSDR.org for info on the SDR)  I may multiply it to help with some of the microwave LOs.  It will also use the 20 Mhz to lock a 1GHz TXCO to be multiplied for microwave LOs.
> 
> Third:  I will have three of the LTE-Light units.  The first will feed some LOs as described above, and the synthesized 10 MHz output will be my lab frequency standard.  The lab is in a cinder block room off the basement, with 2" of foam insulation under 2 inches of concrete which is the floor for a covered porch above.  I'd never thought of it, but the "put it on the floor next to a brick wall" idea fits here.  Actually, I can put it next to 2 buried brick walls, and will surround it with cinder block on the remaining sides.  Can probably cover it with a few 12x12 paver stones.  NOW, this involves drilling a hole through cinder block and drywall between the office/ham shack and the lab.  Would rather not, but have to anyway.  I have been "informed" that the fan noise from the ham shack gigabit ethernet switch will become politically unacceptable in about 72 hours.  (Office/ham shack share a guest bedroom.)  I would like to get 1E-10 or 1E-11 accuracy out of this setup.  Thanks for this suggestion!
> 
> Fourth:  The second unit will be in a building at the base of my antenna tower, about 350 feet from the house.  This building is above ground, and will be allowed to swing from 45F to 80F over the course of the year.  Hence my interest in insulating and heating.  I might consider putting something in the ground here, the problem would be access for servicing....  I would like to get 1E-10 or 1E-11 accuracy out of this unit.  Considered shipping 10 MHz in coax out from the house, would rather not, and would like some redundancy, anyway.
> 
> Fifth:  I get that the /efc/ vs. /temp/ relationship is very complex and accept that trying to characterize it is not worth the effort.  Thanks for this bit of information.
> 
> Sixth:  My third LTE-Lite will drive a 10MHz reference for a mobile ("rover") microwave setup, providing the reference for a bunch of GHz LOs.  This station will see motion, and temperature variation. Ultra low power will not be a concern, so heaters are acceptable. I would be happy with 1E-9 accuracy out of this unit.  That translates into 10Hz frequency error at 10 GHz.  This kind of frequency accuracy has been demonstrated to provide 3+db improvement in the ability to detect weak signals -- very significant for microwave weak signal work.
> 
> Finally:
> I have pondered all the suggestions about measuring output impedance, etc.  For now, I have decided to default to Said's expertise with the units and will use one of his suggested circuits as buffers.  Hopefully, these will be  on a board inside the HAMMOND box with the LTE-Lite.  That buffer will drive one of the MMICs to provide additional power to drive a filter and then output to the distribution amplifier.  I will continue to look for a better idea from one of you smarter than me.
> 
> Thanks again for all the insight and ideas.  You guys type and I learn.
> 
> 73,
> Jim
> wb4gcs at amsat.org
> 
> On 11/23/2014 4:46 AM, Dr. David Kirkby (Kirkby Microwave Ltd) wrote:
>> I would like to make a unit with multiple 10 MHz 50 Ohm outputs to feed my
>> various bits of test equipment.  I am thinking about some practical
>> considerations.
>> 
>> 1) It would be great if there was a circuit published which can give 50 Ohn
>> output impedance from a 12-15 power supply,  which
>> 
>> a) Doesn't load the TCXO
>> b) Doesn't degrade the phase noise.
>> c) Powered the LTE lite.
>> 
>> Ideally one for both 10 & 20 MHz crystals.
>> 
>> Better still if there was a PCB available.
>> 
>> 2) How should I mount the components?
>> 
>> My preference would be a metal box with
>> * IEC mains socket
>> * antenna input socket
>> * 9-pin D for reading dats
>> * 15  BNC's outputs
>> 
>> With a power amplifier to provide the output for 15 sockets, some
>> ventilation possibly requiring a small amount of forced air cooling would
>> be needed. But given the TCXO"s sensitivity to temperature changes, I don't
>> know whether it might be preferable to mount the LTE lite in its own box
>> without any power supplies in it - perhaps with some thermally insulting
>> material around the LTE lite so the crystal doesn't experience any fast
>> temperature changes. Then have the power hungry bits completely separately.
>> 
>> I don't have a particularly big lab, so wherever I mount the LTE lite, the
>> temperature is going to change with the air conditioning unit blows hot or
>> cold
>> 
>> There are fairly large temperature changes when I am not using the lab, as
>> I don't run the air conditioning unit 24/7.
>> 
>> I am interested in people's thoughts on the best way to go about this.
>> 
>> For testing I have a couple of  signal generators that have ovens that are
>> powered 24/7. Also I should soon have the SR620.
>> 
>> Dave.
>> _______________________________________________
>> 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.
>> 
> 
> 
> 
> ---
> This email has been checked for viruses by Avast antivirus software.
> http://www.avast.com
> _______________________________________________
> 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.




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