[time-nuts] Z3805 two frequency maxima

Volker Esper ailer2 at t-online.de
Wed Nov 28 09:19:56 UTC 2012


...and I found a very interesting article on that topic:

http://www.micrel.com/_PDF/App-Notes/clk/PAN1202291%20-%20AFM%20Period%20Jitter.pdf

It's written by Eddy van Keulen (Micrel): Relation between Harmonics and 
Deterministic Jitter. His approach is the assumption, that frequency 
modulationt causes a spectrum with 10MHz and 5MHz (and of course 15MHz). 
Using the FM thesis and formulas give us a very graphic description of 
the observed two-maxima-histograms (a.k.a. bimodal or double peak).

I'll do more experiments and calculations at the weekend.

Volker



Am 28.11.2012 01:36, schrieb Volker Esper:
>
> Now, that you ask: the measurements are in that range, yes, though it's
> not exactly the values.
>
> time shift 12ns -> spacing 10ns
> 28ns 35ns
> 66ns 55ns
> 0ns 8ns
>
> and yes, the peaks get closer while reducing the amplitude of the 5MHz,
> it's almost exactly linear:
>
> With the 10MHz at 200mV and a time shift of 66ns, I measured the
> following spacings:
>
> 5MHz voltage in mV spacing in ns
> 200 55
> 100 27.5
> 50 13.0
> 25 6.5
> 12.5 3.3
> 6.3 1.6
> 3.2 0.86
> 1.6 0.44
> 0.8 0,24
>
> The functional relation of voltage ratio and spacing is quite obvious.
>
> I have to admit, that my counter is not at it's optimal calibration. I
> will adjust it first before I can tell you more.
>
> To be precise: all these findings were made while trying to adjust the
> trigger circuits of the counter. To do that I needed a well designed 180
> degrees power splitter - but I didn't have one at the time. I've
> received a Mini Circuits ZSCJ-2-2 recently, now I can go on with my
> adjustments.
>
> Though the SR620 TIC is a great instrument when hunting the pico seconds
> we have to realize, that it's a thermal design desaster (I have to
> apologize to all sr620 friends). I have to run it for at least 12 hours
> if not 24 to be shure, that every single part is at a more or less
> stationary thermal state. Some (NERC) say "...never switch it off".
>
> I can't do that during the week, we have to wait till weekend.
>
> Thanks for your response
>
> Volker
>
>
> Am 28.11.2012 00:25, schrieb Azelio Boriani:
>> Very interesting indeed. Two questions: after adding the 66nS phase
>> shift,
>> were the two peaks at 66nS when at the same amplitude? Then, while
>> reducing
>> the amplitude of the 5MHz, were they getting closer (until the 240pS)?
>>
>> On Tue, Nov 27, 2012 at 9:21 PM, Bob Camp<lists at rtty.us> wrote:
>>
>>> Hi
>>>
>>> Phase does indeed matter, it just messes up the math. Most multiplier /
>>> filter combinations have significant phase shift between the
>>> sub-harmonic
>>> and the carrier. You rarely know what the phase shift is, but you can
>>> read
>>> the sub-harmonic. The simple db to jitter ratio gets you close enough to
>>> make rational decisions on how much filtering you need. You could
>>> play with
>>> filter phase but I've never seen that done in practice.
>>>
>>> Bob
>>>
>>> -----Original Message-----
>>> From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
>>> Behalf Of Volker Esper
>>> Sent: Tuesday, November 27, 2012 2:34 PM
>>> To: Discussion of precise time and frequency measurement
>>> Subject: Re: [time-nuts] Z3805 two frequency maxima
>>>
>>>
>>> I followed your argument and tried to synthesize such a signal. I built
>>> a simple power combiner (3 times 18 ohms resistors) and combined the 10
>>> MHz reference output of my signal generator with a 5 MHz signal from the
>>> same generators regular output at the same amplitude. My oscilloscope
>>> showed the locked phase of the two signals.
>>>
>>> I applied this combo signal to the SR620 and observed a wonderful
>>> two-maxima histogram. When reducing the amplitude of the 5 MHz signal
>>> (while keeping the 10 MHz amplitude) the peaks distance decreased
>>> linearly with the voltage reduction, until the peaks melted togehter at
>>> about -20dBc.
>>>
>>> On the one hand it was a success, but why only 20dBc? My experiences
>>> with the Z3805 showed a 5 MHz subharmonic at 62dBc and the peaks spaced
>>> at 60 ps.
>>>
>>> So I startet to add phase delay to the 5 MHz signal by looping-in some
>>> meters of coax cable.
>>>
>>> When coming to a delay time of 66 ns I could distinguish the two peaks
>>> at a spacing of 240 ps down to an amplitude ratio of about 1000, that
>>> is, 60dB.
>>>
>>> Volker
>>>
>>>
>>>
>>>
>>>
>>> Am 18.11.2012 03:36, schrieb Bob Camp:
>>>> Hi
>>>>
>>>> Just good old Fourier series.
>>>>
>>>> Bob
>>>>
>>>> On Nov 17, 2012, at 9:12 PM, Volker Esper<ailer2 at t-online.de> wrote:
>>>>
>>>>>
>>>>> I'm impressed - but what law is behind this?
>>>>>
>>>>>
>>>>> Am 17.11.2012 21:26, schrieb Bob Camp:
>>>>>> Hi
>>>>>>
>>>>>> 60 db isn't to bad a number. More or less:
>>>>>>
>>>>>> 100 ns -> 100 ps is 1000:1. 20 log of that is 60 db. 100 ps to 60 ps
>>> is about 4.4 db. That would sum up to -64.4 dbc. The main gotcha is that
>>> you
>>> *might* also have some 15 MHz (and higher) energy in the signal as well.
>>> Also phase gets into the calculation. Still, pretty close.
>>>>>>
>>>>>> Bob
>>>>>>
>>>>>> On Nov 17, 2012, at 12:50 PM, Volker Esper<ailer2 at t-online.de>
>>> wrote:
>>>>>>
>>>>>>>
>>>>>>> So let's have a look into the machine... and what do we see?
>>>>>>> There's a
>>> nice little Symmetrcom oven, with the sign reading "5.000 MHz" - bingo!
>>>>>>>
>>>>>>> May be there's a time saving way to determine the energie of the sub
>>> harmonic: using my spectrum analyzer. It tells me, that there's a 5 MHz
>>> subharmonic at the level of -62dBc.
>>>>>>>
>>>>>>> How would you have calculated the energy? What would be your ansatz?
>>>>>>>
>>>>>>> Thanks so far
>>>>>>>
>>>>>>> Volker
>>>>>>>
>>>>>>>
>>>>>>> Am 17.11.2012 17:55, schrieb Bob Camp:
>>>>>>>> Hi
>>>>>>>>
>>>>>>>> That's what you get if you have "sub harmonic" energy in the output
>>> of
>>> your OCXO. I'd bet you a warm glass of beer that you have a 5 MHz /
>>> doubled
>>> to 10 MHz MTI OCXO in your Z3805. If you have a lot of time on your
>>> hands,
>>> you can calculate the likely level of the energy from the amount of
>>> jitter
>>> (spacing between the two peaks) you get.
>>>>>>>>
>>>>>>>> Bob
>>>>>>>>
>>>>>>>> On Nov 17, 2012, at 11:41 AM, Volker Esper<ailer2 at t-online.de>
>>> wrote:
>>>>>>>>
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> while playing with my recently aquired TIC (SR620) and
>>>>>>>>> measuring the
>>> period time of some oscillators I discovered something I hadn't
>>> expect at
>>> all:
>>>>>>>>>
>>>>>>>>> The output of my GPSDO (Z3805) writes two maxima in the period
>>> histogram (at a spacing of 60ps).
>>>>>>>>>
>>>>>>>>> I didn't believe that result and assumed an inherent error in my
>>> measuring setup or the counter itself.
>>>>>>>>>
>>>>>>>>> So I plugged another oscillator, the reference TCXO of my signal
>>> generator (R&S SMX), and that result made me happy and uneasy at
>>> once: The
>>> TCXO hat only one maximum.
>>>>>>>>>
>>>>>>>>> I havn't calculated the ADEV curve, yet.
>>>>>>>>>
>>>>>>>>> See pictures.
>>>>>>>>>
>>>>>>>>> Why does my GPSDO produce such a weird result?
>>>>>>>>>
>>>>>>>>> Cheers
>>>>>>>>>
>>>>>>>>> Volker - DF9PL
>>>>>>>>>
>>>
>>> <DSCF1437_bbb.jpg><DSCF1439_bbb.jpg>________________________________________
>>>
>>> _______
>>>>>>>>> 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.
>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> 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.
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> 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.
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> 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.
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> 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.
>>>>
>>>>
>>>> _______________________________________________
>>>> 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.
>>>>
>>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> 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.
>>>
>>>
>>>
>>> _______________________________________________
>>> 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.
>>>
>> _______________________________________________
>> 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.
>>
>>
>
>
>
> _______________________________________________
> 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 mailing list