[time-nuts] Practical Limit - Comparing 10 MHz Oscillators at 10 GHz- an alternativemethod?

Bob Camp lists at rtty.us
Tue Jan 11 18:19:54 UTC 2011


What is the desired goal? 

It's a rare list member that has multiple parts to check that are in the <
1.0 x 10^-13 at one second ADEV category. There are indeed a few that do,
for the rest of us, that's a pretty reasonable limit.

If my sources are at that point, a measurement system that does 5X better
probably is more than adequate. That would be 2.0 x 10^-14. It's likely that
> 90% of us would be doing just fine with a system that did 1.0 x 10^-13. 

DMTD is the obvious way to do this, but let's go the old fashion way

If you want to use a "straight counter" and look at 10 MHz, you need an
error multiplication of 1.0 x10^5 or more. Not going to happen for less than
the cost of upgrading the counter.

A number of cheap "computing counters" are out there in the ~$100 range.
More or less they come up with 1x10^9 resolution regardless of the input
frequency. That gets you to a multiplication of 10,000 to hit 1.0 x 10^-13.
That is in the "we can do that" range.

You could also spend a more and add a digit (or more) to the counter. That
would get you a multiplication of 1,000. My guess is that you haven't saved
as much on the error multiplier as you spent on the counter. 

Looks like something like a 5335 is the target counter in this case. 

You get multiplication from taking the frequency up to something high from
10 MHz. You also get it from mixing down to something below 10 MHz. It's all
part of the same process, but it's easier to look at it in two chunks.

Going up, simple stuff will get you into the low GHz. That's 100X or more.

Going down, how low do you want to go? 100 KHz will give you 100X. 

Simply done, a single stage system can give you roughly what you need to get
your 5335 to the 1.0 x 10^-13 level at one second. Other than the output
filter, parts cost would be sub $50. You likely are limited to about 25,000
with simple parts. Output to the 5335 would be at 100 KHz in that case. 

Taking that to a two stage system opens up a few choices. You can do 250X
and 250X with 10 MHz out. The output filter might be easier in this case.
I'm guessing cost would be $100 including the hand made filter. It would get
you to 62,500 and the 1.6 x 10^-14 level at 1 second. 

Since all this is below the noise level of the sources it's only really
useful for measuring noise. There is another way to look at it.

A 1.0 x 10^-12 / 1 second ADEV source not going to be settable to that level
at one second. A lot depends on the noise process involved, but let's (for
simplicity) assume that it's settable to 5.0 x 10^-12. Our same 5X better
number gets us to a counter that will do 1.0 x 10^-12 at one second. 

In this case you need a multiplication of 1,000 to do the job. Multiply to 1
GHz and mix down to 1 MHz. Maybe costs you $70 or so, including a hand made
crystal filter. 

Lots of ways to do it. 

Of course you could also just go buy a Tracor 537 for next to nothing and
get it running. Finding one is probably harder than fixing it. I doubt you
would spend more than the cost of building something. 

What ever way you do it, you are still stuck with the same old problem. The
input needs to be on a very specific frequency. With some designs it can be
a sub multiple of that frequency. With other designs it can be a multiple.
Either way, 7.352 MHz is going to give you trouble. That's pretty much why
the error multiplier boxes all went into storage in the 1970's...

Another problem you are stuck with is the output filter. You would like it
to be nice and narrow (under 100 Hz). That's 10 ppm at 10 MHz. With a 10,000
multiplication you are down to 1.0 x 10^-9. If it's centered, you get +/-
5.0 x 10^-10 of range on your system. If the parts start out as close as the
5335 will let you set them, they may or may not be in range on a 10,000X
system. Of course you can turn off the filter. 

Wonderful stuff.



-----Original Message-----
From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
Behalf Of Perry Sandeen
Sent: Monday, January 10, 2011 9:35 PM
To: time-nuts-request at febo.com
Subject: [time-nuts] Comparing 10 MHz Oscillators at 10 GHz- an


While comparing oscillators at 10 GHz might work for some it is indeed a
daunting challenge at the very least.

I would like to propose what may be cheaper and more readily achieved method
for Comparing 10 MHz Oscillators by more of us time-nuts.  (This is still

What I propose is this:

Make two separate, but exactly the same, circuits, one for the reference and
the other for the unknown to be measured. They would be located parallel to
each on a circuit board.

The input would be divided by a decade counter giving us a 1 MHz signal.
Then we use this signal to phase lock the oscillator section of a CD 4046 to
10MHz.  The 10 MHz signal goes into one port of a LM 1496 double balanced
mixer.  The other input would be from a reasonably stable 9 MHz oscillator.
Taking the DBM output into a simple low pass filter we use that 1MHz signal
to phase lock the next CD 4046 oscillator to 10 MHz.  Then we start the
whole process again.  

At any point we chose along the chain or at the end we feed both 10 Mhz
signals into a single CD 4046 or similar phase detector and obtain our

Since each set of circuits is fed with the same 9 MHz subtraction
oscillator, any deviation of its performance is cancelled.  

The required chips are inexpensive and these frequencies are far more
manageable than microwaves.

Constructive criticism? Comments?  Improvements?




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