[time-nuts] Re: Question about ADEV for tau < 1000s

[Carsten Andrich]

Hi James,

author of the referenced paper here. Back then, I needed to compare time 
sources and I opted for SDRs for two reasons: I already had experience 
with SDRs and such devices available. Admittedly, the resulting 
measurement performance surprised me, so I ended up writing multiple 
papers about it. We're still using the measurement setup today to 
torture GNSSDOs: https://arxiv.org/abs/2406.16037

If you're looking for a cheap and simple solution, the tinyPFA may be 
your best good choice (disclaimer: I have never used the tinyPFA; 
judging purely on published information). Its performance is within less 
than an order of magnitude of my USRP N210 setup. It also comes as a 
ready-to-use product, whereas with the SDR approach – depending on your 
level of expertise – you're probably looking at several months of 
software development before you have a usable and reliable measurement 
setup. Back in 2016 I had to resort to hand-written vector code [1] to 
get the complex-valued sample processing to run in real-time on now 
ancient CPUs. Compilers' auto-vectorization and libraries have come a 
long way since then, but I'd still expect it to be an uphill battle today.

Best regards,
Carsten

[1] https://gist.github.com/carstenandrich/facd709aa409cae4053b2b8d6848fc98

On 17.08.24 07:36, Erik Kaashoek via time-nuts wrote:
> Jim,
>
> For accurate phase comparison without much effort and cost you may 
> have a look at the tinyPFA : 
> https://www.tinydevices.org/wiki/pmwiki.php?n=TinyPFA.HomePage
> An example of the expected performance when connected to Timelab: 
> https://www.tinydevices.org/wiki/pmwiki.php?n=TinyPFA.TimeLab
> Some measurements use the name tinyDMTD instead of tinyPFA but its the 
> same device.
> Here is are some examples measuring OCXO/DOCXO and Rb: 
> https://www.tinydevices.org/wiki/pmwiki.php?n=TinyPFA.Examples
> You can search this list for more info
>
> Erik.
>
>
> On 16-8-2024 23:00, James Spottiswoode via time-nuts wrote:
>> Hi Time-nuts Experts,
>>
>> I’ve found the time-nuts archive to be incredibly useful over the 
>> last year - thank you all!  I have an unusual question.
>>
>> I’m interested in getting the lowest possible ADEV at tau’s ~< 1000s 
>> and am not concerned about longer term drifts.  The protocol I’m 
>> using is based on comparing phase or frequency between two 10 MHz 
>> frequency standards a few meters apart in the lab.   Currently I am 
>> comparing a pair of OSC5A2BO2 OXCOs using an Agilent 53230A running 
>> with a 1 sec gate time and using the undocumented RCON command.  I am 
>> seeing a short term ADEV of 3e-11 at tau = 100s. I would like to get 
>> an order of magnitude or more better than this if possible without 
>> breaking the bank!
>>
>>  From the research I’ve done it looks like high end OCXOs are the 
>> best bet for lowest ADEV but many are designed for high acceleration 
>> environments, which I do not need for static references. The best 
>> I’ve found so far is the Connor-Winfield OH320-LA.  Does anyone here 
>> have a better suggestion? Are double oven controlled oscillators a 
>> better way to go?
>>
>> For frequency or phase comparisons between the pair of OCXOs an SDR 
>> looks like a better option than the 53230. This paper: 
>> "High-Precision Measurement of Sine and Pulse Reference Signals using 
>> Software-Defined Radio” which can be found here: 
>> https://arxiv.org/pdf/1803.01438 states in the abstract:   “… The 
>> measurement system is implemented and verified using the Universal 
>> Software Radio Peripheral (USRP) N210 by Ettus Research LLC. Applying 
>> standard 10MHz and 1PPS reference signals for testing, a measurement 
>> precision (standard deviation) of 0.36ps and 16.6 ps is obtained, 
>> respectively.”  This is considerably better than is achievable with 
>> the 53230A.
>>
>> So I have two asks:Â  Advice on the lowest ADEV OCXO for tau < 1000s 
>> and your thoughts on the SDR based method described above for high 
>> precision phase measurements.
>>
>> With many thanks in advance,
>>
>> James
>>
>> James Spottiswoode  KG6SMH
>>
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com
>> To unsubscribe send an email to time-nuts-leave at lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe send an email to time-nuts-leave at lists.febo.com