LNA (Low Noise Amplifier) Design Tips? - RF Cafe Forums
Post subject: LNA (low noise amplifier) design tips?
Unread postPosted: Tue Apr 26, 2005 3:23 pm
Joined: Sat Apr 23, 2005 2:09 pm
This is my first time designing a simple
LNA (low noise amplifier) project for a class. I was wondering if
anyone has any tips especially regarding biasing and feedback. I
think I have almost met the design specs, but I'm still have problems
getting my output return loss to meet the spec.
I am using
a microwave BJT transistor and have incorporated two feedback schemes
to get my circuit to be unconditionally stable at a certain frequency
range and also get my gain up.
Any good tips from the
LNA experts? Thanks!
Unread postPosted: Tue Apr 26, 2005 5:28 pm
What are exactly your problems with bias and feedback? Let
us know so we might be able to help you out.
Unread postPosted: Tue Apr 26,
2005 9:22 pm
Joined: Sat Apr 23, 2005
Location: Tampa, FL
One thing that I
am curious about pertains to having two types of feedback schemes
in the circuit. In reality, would this kind of design be appropriate?
Currently, in my design, I have a resistor feedback from collector
to base and also a reactive component attached to the emitter.
Aside from using an active feedback scheme, it seems like that
was the only option I had to get my circuit stable at a certain
range while getting a gain that meets my spec. Is there a resource
that can give me some sort of analytical view of the feedback system?
I kinda like seeing equations that I can follow around, so it makes
more sense to me.
I, somewhat, understand the resistive feedback
part using DC analysis, but as for the reactive element on my emitter,
I'm a bit clueless with a numerical analysis on how it stabilizes
For the bias part, I always thought the Q point
would ideally be the mid-point of the supply voltage to allow maximum
signal swing, but it seems like for the bias networks I've seen
in my book, they don't seem to care about this. They are more interested
in maintaining the collector current as temperature varies. Is this
always the case for LNAs in microwave frequencies?
if my questions seem off. I'm a bit new to this stuff.
postPosted: Wed Apr 27, 2005 12:14 am
questions seem a bit naitve, yet they are OK.
The bias point
should ideally be in the middle of the Ic vs. Vce curve. Yet, you
should check if the maximal outpt swing of the LNA per this given
operating bias point still gives you a complete swing without clipping.
Then if this is the case, then you can work at this bias point (that
still will allow a Class A linear operation).
resistor is a common way to provide a broadband stability. This
method is called uniliteralization, since it reduces the effect
of the reactive feedback element (the CB junction capacitance, which
is a cause for oscillation), making the transistor more unilateral
(reducing S12). This method reduces the gain of course, yet doesn't
hurt the NF or IP3 as placing a series resitor at the output or
at the input of the device.
Attached below is a link to a
tutorial that discusses bias schemes (Both active and passive) specific
for LNA design:
Hope this helps.
Should you need any
other help, please let me know.
Unread postPosted: Wed Apr 27,
2005 10:18 am
One very important thing people have tendency
to forget is that when you will adjust the amplifier on the bench
you have to stay very quiet.
Remember any source of noise
will add up the end result noise figure....
Post subject: LNA Design
Unread postPosted: Tue
May 10, 2005 12:15 am
What are the design specs, device
and frequency range you require? Techniques vary. Above a few hundred
MHz you should use "S" parameters. Your design should start at the
input and you must provide the match necessary to get the noise
figure you desire. With this known you can claculate the load required
to get the desired gain and be stable. Using "S" parameters takes
all this into account. By the way an amplifier is never matched.
Generally you take a 50 source and add a matching network to present
a desired driving impedance to the amp. On the output you take a
50 ohm load and a matching network to present to the amplifier the
load necessary to get the desired gain and stablility. This is not
matching in the sense of a conjugate match. It is simply a network
that provides a desired impedance when attached to 50 ohms. I have
a nice little "S" parameter design application on my web page if
you need it. It requires that you know something about design so
you may want to look at the book by G. Gonzalez, "Microwave Transistor
My web page is at http://members.cox.net/thse-3.14159/download.htm
Tue May 10, 2005 11:34 am
I think that for LNA the bias
point is lower than 50% IDSS.
The best rule for this is to
follow the manufacturers recomendations....
Post subject: LNA
Unread postPosted: Thu May
12, 2005 12:09 am
The bias point for an LNA is driven by
the noise performance you require. Look at the device datasheet
and the manufacturer will tell you what current and voltage is required
for lowest noise. They will also give you noise parameters for thes
conditions. In general less current means lower noise. More current
means more noise but more device gain. Front signal handling and
intermod are usually important. A good design has to trade off low
noise performance against high level signal handling by finding
a compromise in bias point that works for most expected conditions.