[time-nuts] Comparing 10 MHz Oscillators at 10 GHz

Bob Camp lists at rtty.us
Wed Jan 5 17:33:24 UTC 2011


My guess is that you could take a pair of the PLL+VCO chips that several
people now make (cheap) and whip up a pretty simple error multiplier

The Analog ADF-4360-1 is one example at ~ $3 each. You could use it to take
your 10 MHz up to 2250 with one and 2260 with the other. Mixing back down to
10 MHz would boost the apparent offset by 226. Switching to 1 MHz or 100 KHz
for the output would give you 2260 or 22600 as the multiplication factor.

The issue with any of these setups is going to be jitter. The tried and true
approach back in the 1960's was to run the output of the mixer through a
narrow band crystal filter. That let you strip off the "extra" wide band VCO
phase noise and keep the noise associated with the source. 

The results are the same in the time domain, but it's a *lot* easier to look
at it in the frequency domain. Assume your 10 MHz source is running -155 dbc
/ Hz at 100 Hz offset. The "correct" multiplied noise would be -108 dbc / Hz
at 2250 MHz. That's about 30 db better than the AD chips seem to do with a
PFD at 200 KHz. More or less, you would need to get into the < 10 Hz offset
range to be properly representing the noise on the OCXO. That's hoping
there's not another gotcha at the lower frequency. 

The noise situation gets "better" as the PFD frequency goes higher. That
gives you a trade off between high multiplication (mix down to 100 KHz for
example) with higher noise vs lower multiplication and lower added noise.  

This is not a knock on the AD chips. They just happen to have a pretty good
data sheet. Most if not all of the microwave bricks are going to have the
same sort of issues. None of them are really designed for super low phase
noise inputs at 5 or 10 MHz. You'll also find that the sampling detectors in
the bricks have some "interesting" time domain issues as well.... 



-----Original Message-----
From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
Behalf Of EWKehren at aol.com
Sent: Wednesday, January 05, 2011 11:53 AM
To: time-nuts at febo.com
Subject: Re: [time-nuts] Comparing 10 MHz Oscillators at 10 GHz

Strictly for adjusting and comparing two 10 or 5 MHz sources I use the  
circuit from the Austron 2110 that takes a 5 MHz input and through mixing  
generates 5.0005 MHz subsequently devided down to 1.0001 MHZ.  Mixed with
other input at 1 MHz the resulting difference is 100 Hz which any decent  
counter can resolve to nine digits. Works for me and often I just use the
At one time I used the multiplier chain out of a couple of FTS 4050  
Cesium's but found the above more convenient for strictly tuning purposes.
Bert Kehren  Miami
In a message dated 1/5/2011 3:18:52 A.M. Eastern Standard Time,  
df6jb at ulrich-bangert.de writes:


I had the same idea as you and my friend Frank and I  performed the 
experiment to check whether it is possible or not:  

Two brick oscillators (I believe to remember in the 8 GHz region)  were
locked to the same source (HP8660) in the 100 MHz region which in turn  was
locked to my local 10 MHz reference (Z3805). The two signals were mixed  
to DC with a M14A mixer. By means of a phase shifter in one of the  cables 
were able to change the phase between the signals and so to  determine the
mixer's sensivity as a phase detector. The mixer's output was  sampled with 
HP3457 at a one second sample rate. The voltage measurements  were then
re-computed into phase fluctuations and this data was fed into my  PLOTTER
utility to compute what must be considered the AD noise floor of  this
system. I have not documented the results but I remember that the  noise
floor indicated a clear improvement against a direct phase comparison  at 10
MHz for a given TIC resolution. 

What you suggest will produce  you a mixer output signal which (when looked
at with a scope) will easily  enable you to trim your LPRO within seconds. 
you can lock the bricks  directly to 10 MHz this is even better. 

I have been thinking to use  this scheme as a general tool for oscillator
stability measurements. Since  we must consider that two odcillators may not
always be THAT close to each  other in terms of frequency it would be better
not to mix to zero but to a  beat freaquency of say some 1-100 kHz (depends
of course on the brick's  pull range). This would involve a offset generator
for one of the signals.  I have drawn a circuit but not actually built that
uses a ADF4002 and a DDS  block to lock a 100 MHz signal to a 10 MHz signal
where the DDS will  provide the possibility to offset the 100 MHz signal in
small amounts. I  plan to lock two low noise WENZEL 100 MHz OCXOs to the 10
MHz sources with  one of them with a small offset. Then these two 100 MHz
signals are  compared after being multiplied by the brick oscillators (I 
two bricks  that translate 100 MHz to 10 GHz). 

Perhaps the group can comment on  the feasibility of the plan.

Best regards
Ulrich  Bangert
time-nuts  mailing list -- time-nuts at febo.com
To unsubscribe, go to  
and follow the  instructions there.

time-nuts mailing list -- time-nuts at febo.com
To unsubscribe, go to
and follow the instructions there.

More information about the time-nuts mailing list