[time-nuts] Flicker Noise Reduction

[Magnus Danielson]

Hi,

On 2020-11-19 05:27, jimlux wrote:
> On 11/18/20 7:15 PM, Richard (Rick) Karlquist wrote:
>>
>> On 11/18/2020 5:27 PM, jimlux wrote:
>>
>>> The proposed mitigation technique relies on tracking the
>>> rapid gain variations in the radiometer due to 1/f noise and
>>> correcting them by generating a baseline state in the first
>>> amplification stage of the low noise amplifier (LNA). The
>>> proposed 1/f noise mitigation method can be applied to any
>>> receiver, but it is especially valuable for THz receivers since
>>
>> I'm confused here.  My understanding of 1/f noise was
>> that it was additive as in "AWGN".  Can you clarify this?
>> You also imply it works for ANY receiver.  Now I'm REALLY
>> confused.  Unless the "direct conversion receiver" is actually
>> an envelope detector.  Then I think I see how it can work
>> at any frequency.
>>
>> Not arguing, just trying to understand :-)
>>
>> Thanks.
>>
>> Rick N6RK
> I was just copying from the intro to the paper. I'll read it tomorrow.
>
> My first glance is that they are talking about 1/f noise as gain
> variations, and stabilizing that. Maybe for THz amps that's a thing.

It seems that they aim to reduce the amplitude variations, and
considering their detector that will for sure be visible, but it should
also tell you about which frequencies they observe and thus which noise
affects them. Not being able to use normal techniques to mitigate 1/f
noise in amplifiers may be a major design hurdle at 670 GHz, who knows,
that would probably take several articles to clarify.

Personally, I find that 670 GHz is somewhat out of reach for my lab, so
I can't say I have much experience up there.

Oh, and for Johnsson noise and frequency, it's not white you know, it
starts to slope more and more. The thermal noise obeys the Max Plank
equation, but I have yet to see the proper derivations from that over to
electrical noise, because you need to properly show the physics of a
solid object and translation into electrical properties such as charge,
volt and current. I'm sure someone done the work, I have just not found
a really good reference for it, but it's fairly obvious from the
hand-waving level. As you do radios in 67 GHz range I know that the
slope is causing them issues, as it becomes noticable. There goes the
low-frequency approximation. This is also relevant as the noise level
and noise shape depends on temperature, which is used in radio astronomy
on regular basis.

Cheers,
Magnus