Difference between NF and NFmin

some one could explain the difference between NF and NFmin in detail?
i know the NFmin is achieveable minimum NF of circuit, but i want to know how to use it when adjust the LNA?
is NFmin unrelated to matching(input power matching and noise matching)? if it is, whether i can use it to find the optimum bias point of LNA through sweeping the gate voltage, and regardless of matching conditions?

thanks a lot


tiger

Hi

When you have a transistor designed for LNA design, it generally comes with data like NFmin and Γopt. NFmin is the minimum achieveble NF of the amplifier WHEN the source reflection coefficient(Γs) is equal to Γopt.

As you properbly know, to have maximum power transfer of an amplifier you need to have conjugate match at input and output or expressed in another way: Γin=Γs and Γout=Γl.
You can make a perfect match at the output, but the input is where you have to make a compromize between low noise and high gain as the source reflection coefficient that maximizes your gain is not the same as Γopt which minimize your noise!

A procedure which can help, is to plot the noise circles and available gain circles in a Smith Chart and then do it for different bias points. You can do this by hand or use som software(ADS, MWO, etc..). You can plot Γopt and also Γs. I know ADS has a function that can plot the Γs that gives the maximum gain. These two points should be as close to each other as a first starting point.

I hope this helps, it is somtimes a bit difficult explain things only in writting.
But give a message if you have more questions.

Regards

referring to noise Factor:
F=Fmin+(rn/gs)|ys-ymin|^2
where:
Fmin is the min noise obtainable from the 2-port at a normalized source admittance of ymin.
rn is the normalized noise resistance for the 2-port.
ys=gs+i*bs is the normalized source admittance.

as you can see there Fmin is not linked to power matchin but only to noise matching.

I hope it can help.
Mazz

NF is the practical noise figure of a pratical rf network and NFmin is the ideal or a therotical noise figure of a circuit, NF=NFmin+matched network factor!
that means the matched condition influence the noise factor, but the NFmin is the best conditions that a matched network can arrive!
i doont know whether this explaination is clear or not!

thanks for your patient explanation, perhaps i don't explain my meanings clearly. i concern about RFIC, so i can't get data of MOS transistor.

Added after 11 minutes:

Added after 3 minutes:

to mazz and springf2000:

thanks a lot for your help!
i know that if i mismatch noise, the NF will deviate from NFmin.
i see some paper, NFmin is only inversly proportional to cutoff frequence with minimum gate length, so only bias condition can influnce NFmin, am i right?
in other words, i can simulate only one MOS transistor to determine the optimal bias point through watching the NFmin varying along with the bias conditon, but i can't, i want to the reason!
NFmin is related to noise mathcing? specifically is it influnced by Bs=Bopt?

thanks a lot!

To your last questions:
don't confuse a theory based on linear networks (the one that defines NFmin) with circuit issues.
NFmin is related to noise mathcing?: yes, it's NF when you have noise matching.
specifically is it influnced by Bs=Bopt? yes, it's obained at Bs=Bopt.

For circuit issue I strongly suggest to read some specific book. One (good) example can be:
Kluwer CMOS cellular receiver front-ends-Janssen Steyaert that can be found in EDAboard.
You'll see that it's not so easy to design CMOS LNAs.
Good luck.
Mazz

There's a very old paper talking about noise parameters, i did not remember the title anymore. But it states that, if the matching compenent is lossless, the NFmin is unchanged.

So NFmin is unrelated to lossless mathcing network.

thaks a lot!
but what will affect the NFmin?
whether the circuit is matched or not, the NFmin is constant?

what role does the NFmin play in the process of designing LNA?

when i seek to the optimum bias condition or device width, i should watch the NFmin, am i right?

tiger_ads
for first two question I recommend once again to read a good book: the argument is too large to discuss in a forum.
For last two question, you're right. NFmin gives you an indication that, using the external matching components you cannot do better in term of noise.
But...
it can easily happen that:
1) first LNA design: NFmin=3 dB and NF=3.05 dB
2) second LNA design: NFmin=0.8 dB and NF=1.2 dB
with CMOS you can have it...with similar bias current!
Be aware and read books.
I hope it can help.
Mazz

Mazz
thanks a lot!

but could you explain what mean "first LNA design" and "second LNA design" ?

BTW, i have tried to download the book you have suggested, but always there is something wrong in the file, i couldn't downloat it. is there other way i can get it?