[time-nuts] Re: DIY Low offset Phase Noise Analyzer (Erik Kaashoek)

[Bob kb8tq]

Hi

Indeed you can switch the gain of the amp. You still need to provide
a low gain output to feed the EFC input on your reference. The chain
to feed the sound card will be crazy high gain for the typical TCXO or
OCXO. Don’t even think of running that sort of gain into a VCO ….

Bob

> On Jul 8, 2022, at 9:16 AM, Erik Kaashoek <erik at kaashoek.com> wrote:
> 
> Bob,
> Clear, you have a lot more experience and knowledge. For me this is typical a case of "If you don't know about something it must be simple"
> So best would be to make it possible in the simple PNA to switch off the opamp gain, without changing the impedance the mixer sees,  so the offset tuned signal can be used to calibrate the slope.
> I found this picture very helpful to understand the relation between phase modulation depth and the strength of the side bands
> http://athome.kaashoek.com/time-nuts/PM_Sidebands.JPG
> It shows that below 0.2 radian peak phase modulation you can simplify to narrowband FM as only the 1st sideband has relevant power (certainly for the accuracy I am after)
> The whole presentation including the calculation can be found here:
> http://athome.kaashoek.com/time-nuts/Measuring_phase_modulation.pdf
> Written by Bob Nelson from Keysight.
> Very helpful presentation for people (like me) that are new to all this.
> Erik.
> 
> On 8-7-2022 18:58, Bob kb8tq wrote:
>> Hi
>> 
>> Like it or not, the mixer is a non-linear load. It also has a frequency
>> dependence. Even with “saturation” levels, the slope can and does
>> change. That’s the short list, as you dive into it, things get even more
>> complex in terms of “might be” sort of issues.
>> 
>> How can you be in saturation and have the slope change ( it does sound
>> unreasonable) ? The fundamental is not changing much (so you are
>> in saturation). The harmonics of the fundamental are changing. Since
>> the output is actually a triangle wave with rounded “corners” there are
>> indeed harmonics very much present.
>> 
>> The flat parts of the triangle wave are a “good thing” in this case. It
>> makes the device linear over a bit wider range than a sine wave would
>> provide. This gets you out of all sorts of nutty analysis concerning the
>> noise being “to much” to measure with the device. It also relaxes the
>> needed accuracy of the DC lock part of things. ( = slope of a sine wave
>> changes quickly ….).
>> 
>> You never really get away from the “to much noise” question. The
>> common definition of phase noise is that it’s more than 60 db below
>> carrier. That is really just the commonly used limit for  “you may need
>> to think about FM sidebands”.  Yes, that’s another rabbit hole to wander
>> down ….
>> 
>> Bob
>> 
>>> On Jul 8, 2022, at 8:32 AM, Erik Kaashoek <erik at kaashoek.com> wrote:
>>> 
>>> Bob
>>> This confuses me.
>>>> The calibration of the system changes ( or can change ) each and every time you swap
>>>> out signal sources. The levels are not going to be consistent setup to setup. Thus you
>>>> calibrate each and every time you change out either device.
>>> Assuming each source is saturating the mixer sufficiently (to be confirmed by measuring the output level of the source into 50 ohm) I do not understand how changing a source can change the calibration. Can you explain what is happening?
>>> Please keep in mind I'm not after 0.1dBc/Hz accuracy, +/- 5dBc/Hz would already be great.
>>> Erik.
>