The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> RF circuit Dc bias question
https://designers-guide.org/forum/YaBB.pl?num=1258474637

Message started by liletian on Nov 17^th, 2009, 8:17am

Title: RF circuit Dc bias question
Post by liletian on Nov 17^th, 2009, 8:17am

hi all
I have a question on the bias of the RF circuit. It is common to have a DC voltage connect through a 10K (R1 in the attachment)resistor to the gate of the mosfet as DC bias and AC signal couple with a big capacitor to correctly bias the circuit.
However, the DC bias resistor will be a noise source as KTR, the large the resistor, the higher the noise. Even we put a huge by-pass capacitor at the DC voltage, there still a fair amount of the noise coming from the bias resistor. Can anyone explain this?
Thank you

Title: Re: RF circuit Dc bias question
Post by pancho_hideboo on Nov 17^th, 2009, 8:35am

It's simple.
Consider noise equivalent circuit of R1 as noise current not noise voltage.

liletian wrote on Nov 17^th, 2009, 8:17am:

Even we put a huge by-pass capacitor at the DC voltage,

How do you put a huge by-pass capacitor ?
By-pass capacitor is located at node "Vg1".
You can not decrease noise from R1 by putting a huge by-pass capacitor without losing gain.

If you are still concerned about noise from R1, user RFC(Radid Frequency Choke) instead of R1.

Title: Re: RF circuit Dc bias question
Post by Mayank on Nov 17^th, 2009, 11:38pm

Hello,

@ pancho : I am still not able to comprehend how it doesnt affect noise.....Could you pls. explain a bit further ?

& at a MOS i/p why should we model noise as current source and not as a voltage source ??

mayank.

Title: Re: RF circuit Dc bias question
Post by pancho_hideboo on Nov 18^th, 2009, 1:05am

Mayank wrote on Nov 17^th, 2009, 11:38pm:

I am still not able to comprehend how it doesnt affect noise.....Could you pls. explain a bit further ?

I don't say no affect. It does affect.

But you have to consider degree of affect to output according to your various conditions.

If L1 is resistor, noise contribution to output will be greatly large for larger L1 resistor.

However if input impedance of MOSFET is high and source impedance is low, noise contribution of R1 to output will not be so large even for larger R1 resistor.
This is usual for this circuit configulation.

Which is more effective, series resistor or parallel resistor is dependent on input impedance of amplifier and source impedance of amplifier.

Mayank wrote on Nov 17^th, 2009, 11:38pm:

& at a MOS i/p why should we model noise as current source and not as a voltage source ??

If you calculate correcty, there is no difference between voltage noise and current noise.
But current noise add no extra node.

More one comment.
If R1 is small, gain of this stage will be small.
So total cascade system NF including this stage as top will be fairly worse.
This will come to nothing even if you make noise contribution from R1 small by decreasing R1.

Title: Re: RF circuit Dc bias question
Post by RFICDUDE on Nov 21^st, 2009, 1:03pm

As Pancho suggested, you need to think of the "shunt" resistor R1 as a thermal noise "current" source rather than a noise voltage.

The noise "current" is inversely proportional to the resistance, so actually the noise goes down the higher the resistance is (provided the bias side of the resistor is by-passed to ground to avoid any additional noise contributions from the bias circuit (not shown).

If the resistor were in series with the RF or voltage input, then the thermal noise voltage would be important and it would be proportional to the resistor value.

Title: Re: RF circuit Dc bias question
Post by loose-electron on Nov 23^rd, 2009, 8:31am

Couple of comments -

Voltage or Current model does not matter - last I checked, Mr. Norton and Mr. Thevien tend to convert between each other.

Take a look at the problem, small signal, what is the impedance at frequency here?

A 1 Meg resistor is a big noise generator, unless its in parallel with a 1 ohm resistor, in which case it looks pretty quiet, because small signal the net result looks a lot more like 1 ohm.

You get the idea.

Nothing divine here, that 10K can be a big noise source, I do agree.

I also agree that seeing it conceptually as either a current or voltage can be useful.

Look at the small signal equivalent.

Jerry

Title: Re: RF circuit Dc bias question
Post by RFICDUDE on Dec 15^th, 2009, 8:12pm

At high frequency separation of voltage and current dependencies does matter if the input impedance is relatively small (>1kohm).

I agree, of course, that Norton and Thévenin representations are equivalent for a thermal noise source, but the impact of the noise source on NF is more complicated if the thermal noise source is connected to a complex valued input impedance.

This is the reason that the procedure for noise modeling a circuit includes determining input referred noise voltage and current sources plus determination of the correlation between the two which is represented by a correlation admittance or impedance. If the input impedance is small compared to the noisy shunt resistor then the equivalent series noise source is very small. It is small not because of the Thévenin equivalent of the noise source but because the Thévenin equivalent of the large resistor in parallel with a smaller input impedance.

I have analyzed and designed a less than 1.5dB NF LNA which has a shunt resistor feeding the bias to the input. The shunt bias resistor doesn't show up anywhere in the major noise contributors because the input impedance is two orders of magnitude smaller than the shunt input impedance.

Also I have found that the dominant contributions of series and shunt elements can change dramatically over just 1-2 octaves when considering practical SMT inductor values for matching.

But the general trend is the same, minimize series resistance with input and maximize shunt resistance relative to the input impedance.