[time-nuts] Measuring phase with an HP 3456A?

Bob Camp kb8tq at n1k.org
Sun Oct 2 02:02:10 UTC 2016


Hi

DMTD = dual mixer time difference, a system publicized in a paper by NIST in the late 1970’s. You mix two “same frequency” DUT’s with an offset oscillator. The beat notes out of the mixers are measured (often with a “computing counter) to get high resolution measurements. Basically an enhancement on the single mixer everybody had been using to make it work for two devices on the same frequency. 

DBM = double balanced mixer. Something you need in the DMTD. You also need isolation amplifiers to get into the (roughly) 120 to 140 db isolation range. 

3456 = A fine old HP digital volt meter. Capable of 6 1/2 digit resolution. Be sure to warm it up for at least 5 hours before you start doing anything with it.

TIC = time interval counter. One way to do the period measurement to flip over (invert) and give you a “computing counter” output. (= measure period not frequency). 

I suspect that OCXO and LPF don’t need explanation. 

Bob





> On Oct 1, 2016, at 8:53 PM, Bob Albert via time-nuts <time-nuts at febo.com> wrote:
> 
> You are firing those acronyms at me too quickly.  I don't know what most of them signify.
> If you have two oscillators, I assume that somehow you lock them to the same frequency; otherwise the phase difference will be changing.  Which one is the reference, and why?
> If you want to measure phase angle I would think a Lissajous pattern would be one way.  Another would be to synchronize a sweep from one channel and use delaying sweep to compare the other.  Another would be to build a calibrated phase shifter.
> 
> Using a voltmeter requires a discriminator.  Plus, the voltage for 180 degrees or zero degrees must be calibrated to find the ratio and thus the sine or cosine of the phase difference.
> And while I never asked why we want to do this, I am reserving that question for later.
> Bob
> 
> 
>    On Saturday, October 1, 2016 5:17 PM, Bob Stewart <bob at evoria.net> wrote:
> 
> 
> Hi Bob Albert,
> 
> I'm having trouble following it, as well, and I started it.  So, let's back up.  Completely eliminate the use of the term "DMTD".  I didn't post that.  I'm only interested in measuring phase with the 3456A, which is a voltmeter.
> 
> So, I have two disciplined OCXOs, a DBM, an LPF, and a 3456A.  The question is this: "Can I use the 3456A to measure the phase angle (not phase noise) more precisely than the 5370A" under very specific circumstances; i.e. the frequencies are known to be very, very close between the two inputs?  One of the obvious limitations on getting a directly convertible phase output is that the mixer only allows me to directly measure 180 degrees of phase difference, and without some external referent, I won't know which 180 degree spread that is.  So, I suspect that I will need part of the answer from the 5370A (or even a 5335 or 5334).  And even then there's the problem of getting the two measuring devices (the DBM and the TIC) to agree on where phase zero is.  But I think I can easily get over the phase zero problem with software, and this will need software in any case.
> 
> Bob
>  -----------------------------------------------------------------
> AE6RV.com
> 
> GFS GPSDO list:
> groups.yahoo.com/neo/groups/GFS-GPSDOs/info
> 
>      From: Bob Albert <bob91343 at yahoo.com>
> To: Discussion of precise time and frequency measurement <time-nuts at febo.com>; Bob Stewart <bob at evoria.net> 
> Sent: Saturday, October 1, 2016 6:46 PM
> Subject: Re: [time-nuts] Measuring phase with an HP 3456A?
> 
> I am having trouble following this thread.  I assume we are trying to measure phase noise, but of course the result includes the noise of the local oscillator(s).  Isn't the 3456A a voltmeter?  I have one of those.  In AC mode it has a bandwidth of more than 100 kHz and measures true rms.
> 
> Someone please take the time and trouble to explain what is being done here.  At this point I am imagining a mixer and a local oscillator and some unknown source.  Quadrature?  How do you accomplish that?  Doesn't that require a 90 degree phase shift?  With an analog phase shifter, or some more modern scheme?  A diagram would be helpful.  I am interested in this partly because I also want to measure phase noise (I can do it with a communications receiver or a deviation meter, both of which I have), and partly because I feel as though I am being left behind technologically speaking and want to keep up better than I have.
> 
> Ever since data streams seem to take less bandwidth than that of the signals whose information they contain, I have been thoroughly confused.  My classic textbooks don't have stuff like this.  When someone talks about quadrature, it looks to me as though they are waving hands in the air, saying some incantations, and then come up with miracles.  I know it works but I surely would love to understand it.
> Bob
> 
> 
>    On Saturday, October 1, 2016 4:24 PM, Bob Camp <kb8tq at n1k.org> wrote:
> 
> 
> HI
> 
> 
> DMTD = Dual Mixer Time Difference
> 
> Single Mixer = what is commonly used for most things. 
> 
> If you have a single mixer setup, just put the two inputs in quadrature, attach to a sound card and you have all you need for phase noise.
> 
> Bob
> 
>> On Oct 1, 2016, at 4:30 PM, Bob Stewart <bob at evoria.net> wrote:
>> 
>> Hi Bob,
>> I don't have a DMTD breadboarded up for testing.  This was just a test of the new LPF using only a single Mini Circuits ZLW-1H DBM, and things kind of progressed from looking at the output of the LPF on the scope to "I wonder what I would see on the 3456A?" sort of thing.  I'm running a holdover/recovery test on the code and hardware changes to get a reliable 1PPS from my GPSDO, so there is some very slow movement over the range of 0 to 100ns.
>> 
>> Bob
>> 
>>       From: Bob Camp <kb8tq at n1k.org>
>> To: Bob Stewart <bob at evoria.net>; Discussion of precise time and frequency measurement <time-nuts at febo.com> 
>> Sent: Saturday, October 1, 2016 3:14 PM
>> Subject: Re: [time-nuts] Measuring phase with an HP 3456A?
>> 
>> Hi
>> 
>> What is the beat note coming out of the DMTD?  
>> 
>> Put another way:
>> 
>> DMTD involves three oscillators. Two are on roughly the same frequency and the third is 
>> offset from the other two. The difference frequency is typically something like 10 Hz. 
>> It does not *have* to be 10 Hz, but that is one way to do it. 
>> 
>> So, moving on using 10 Hz (which may be wrong):
>> 
>> If you are at (say) 10 Hz, you get a 1x10^6 “error multiplication” on the output. One cycle 
>> at 10 MHz gives you one cycle at 10 Hz. The one cycle is 10% of 10 MHz, it’s 0.1 ppm 
>> of 10 MHz. You get a 10 degree phase change at 10 Hz for each 10 degree phase change
>> at 10 MHz. 
>> 
>> The 10 Hz offset limits your phase noise process. The upper (or lower) sideband wraps around
>> at 10 Hz and then starts dumping back into the other sideband’s data. You also need to have a 
>> signal processing chain that will tolerate the carrier being “in band”. Between the two … not
>> such a great way to do it. 
>> 
>> It’s *much* easier to simply hook up a single mixer (half of what you have already) and look at 
>> the two sources in quadrature. Then the sidebands line up. The carrier is gone. The dynamic
>> range can be *much* less. 
>> 
>> Bob
>> 
>> 
>>> On Oct 1, 2016, at 2:32 PM, Bob Stewart <bob at evoria.net> wrote:
>>> 
>>> I've been spending a small amount of my time looking into making a sort of hybrid DMTD with a pair of DBMs up front feeding the stereo input to a sound card.  So, I got the 100KHz LPF back from Oshpark and hooked it up to my scope for verification - an obvious step.  Then I hooked it up to my 3456A just for grins.  (The two DBM inputs are 10MHz outputs from two different GPSDOs).  So, as I watch this, I think the obvious question: can this measure phase angle better than the 5370A?  I guess I need to send it through a full 100ns of phase change to get a calibration value.  So, who's been down this road and what did you discover?
>>> Bob
>>> _______________________________________________
>>> 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_lists.febo.com mailing list