noise figure of mixers

[exces3]

is there someone here that can help me understand
the difference between the dsb and ssb noise figures,
and more importantly, in a typical situation, which applies.

the typical story is that NF is usually about 0.5 db
worse than conversion loss.

( hard to put into writing here)

situation 1.
normal ham situation, you only want one sideband.
what is best expected NF, when WORKING WITH COMPONENTS
at normal room temperature, ..assume LO is well filtered
to eliminate wideband noise.

situation 2, same as 1, except the RF port of the mixer
is reactively terminated, at the ( INPUT's) image.

situation 3, zero IF
( seems to not make sense, as if you do a measurement
with a noise diode, and a 'Y' measurement, you will get
more 'delta', but the system is not any better than ssb.)

thanks for any comment, i THINK i COULD could come
up with more precise questions,
after some thinking about it.

sorry about qrm on this post, the baNDS ARE NOISY

((((((73))))))

 

[G0HZU]

I assume you are referring to VHF/UHF designs if you are concerned about modelling noise figure accurately?

Not sure if the following is of any relevance or use to you but here goes anyway...


Situation 1 with a regular superhet design at VHF with an IF at maybe 21.4MHz.

A typical VHF design will consist of an LNA then a preselector (for image rejection at 42.8MHz away) and then the mixer.

So the issue about unwanted image noise contribution (i.e. unwanted sideband noise 42.8MHz away) becomes less significant with this topology because the LNA will create a pedestal of noise ahead of the mixer at the wanted band but the unwanted sideband image noise 42.8MHz away will not be as big.

eg if the LNA has 15dB gain and a 3dB noise figure and there is maybe 3dB loss in the preselector after the LNA then you get about 15dB excess noise arriving at the mixer RF port at the wanted channel compared to the image noise as we can assume the image noise that got amplified by the LNA didn't make it through the preselector.

So the noise contribution from image noise is often negligible in a typical VHF superhet design?

 

[G0HZU]

However, if you removed the preselector (obviously a bad thing to do) then the image noise at 42.8MHz away would also arrive at the mixer at the same amplified level as the noise in the wanted ham band.

So the the noise performance of the receiver would degrade because the image noise would add 3dB more noise as it would mix 'on top' of the wanted signal.


However if you converted the receiver to a zero IF (DSB) receiver using an I and Q demodulator then the noise performance would be 3dB better than the badly designed superhet case above that had the preselector removed. This is because with a zero IF there is no image to worry about (although with a zero IF/DSB receiver you can get a tiny amount of noise from a band at 3 times your wanted channel if it is not filtered because a typical mixer will only have about 11dB rejection of noise/terms at 3F)

But if you 'do' fit the preselector to the original superhet design then the noise performance of both receiver types (Superhet SSB vs I and Q DSB) should be very similar because the LNA and preselector will hide the issue of unwanted image noise for the superhet?

Hope I've got all that correct as it's late here in the UK and I'm soon off to bed...