Attila

I have measured the PN of the LTC6957-4 at 10MHz.

I could measure the PN of the TAPR variant of the Wenzel circuit as well as the PN of the comparator based circuit (with CMOS output buffer).

Bruce

PN measurements for LTC6957-4 here:

http://www.ko4bb.com/getsimple/index.php?id=phase-noise-and-other-measurements-with-a-timepod

NB HX2410 plot below is for a Holzworth sine to CMOS converter.

Bruce

Moin,

On 10/04/2017 12:38 AM, Attila Kinali wrote:

I modified my TAPR TADD-2 to let one output driver be driven from the
clock input circuit. Driving it from the 5 MHz sine out of a OSA 8600
BVA gave me 2 ps of jitter on the DTS-2070C, which is about as low as it
ever goes, so more or less at it's noisefloor. Not to shabby.

Bruce will probably be able to measure it up before me, as I'm in
Washington DC right now on travels, but I should actually do that
exercise as I get home.

As you have 5 MHz, there isn't very deep Collins squaring needed.

Cheers,
Magnus

Hi

Things can get a bit crazy doing this:

The Collins paper pretty well shows that edge speed matters. A survey
of comparator data sheets will also suggest the same basic thing. Up to
some point (clipping maybe) increasing the levels into a chip improves things.

Not all chips will take the same max P-P input, so you can’t go to the same point
with each of them.  Some have strange breakpoints as you drive them.

Boosting an RF sine wave amplitude can be done with a transformer or with a
matching network. Each has it’s advantages. Both change the source impedance.
That may (or may not) impact the way the chip works. Ultimately things like broadband
phase noise must be impacted ….

It’s tempting to say that a fixed level test will give you a first cut at the various chips.
That may be true in some cases. I’d guess that it’s not a good idea to sort down very
far on that basis ……

Bob

These plots came from the Wenzel diff-amp shaper:
http://www.ke5fx.com/wenzel_resid_PN.png
http://www.ke5fx.com/wenzel_resid_ADEV.png

I don't recall many details of the test setup, or how faithful my implementation of the shaper was to Charles W.'s app note, but this is representative of several plots I have lying around. Not bad performance at all for the cost/complexity involved.

-- john, KE5FX
Miles Design LLC

John
Do you recall what the amplitude of the input signal to the Wenzel shaper was?

Since I used a 1:2 (3dB) terminated splitter for my measurements the input to the splitter is the same as the Timepod reference signal amplitude

Bruce

I was using a similar setup, so the shaper input would have been +4 to +5 dBm.  My intention at the time was probably to feed it a signal comparable to the lowest levels commonly seen from typical 10 MHz sources.  It may or may not do better at higher input levels, but it'll obviously get worse once the level goes below a certain point.

-- john, KE5FX
Miles Design LLC

John

Apparently I've already measure the PN of a comparator driving a few ACMOS buffers.

Attachment (slightly over 100kB) compares this with HX2410(Holzworth).

Bruce

Hoi Bruce,

Thanks for the link. That's some nice data.

On Wed, 4 Oct 2017 12:02:35 +1300 (NZDT)
Bruce Griffiths bruce.griffiths@xtra.co.nz wrote:

This would be very helpful. I have wondered how the modified Wenzel
circuit compared to an LTC6957 or a simple comparator.

		Attila Kinali

--
You know, the very powerful and the very stupid have one thing in common.
They don't alters their views to fit the facts, they alter the facts to
fit the views, which can be uncomfortable if you happen to be one of the
facts that needs altering.  -- The Doctor