Post subject: small signal amp./PA matching Posted: Tue Jul
18, 2006 5:12 pm
Joined: Tue Jul 18, 2006 9:52
when we match input/output for power
gain in small signal amp we match the input/output impedances conjugately
to 50 Ohm and the S11 and S22 are -15 dB or below or something like
that which indicates good matches. i thought I understood matching networks
farily well until i started reading on PA's. in a PA the output is matched
to an impedance that is found using loadpull analysis so the matching
network is not a conjugate match. however, in books and in many PA publications
the S22 is shown and it is usually pretty good -15 dB or something like
that. What does S22 mean in PA's? I thought since output is not conjugate
match the S22 shouldn't be that good.
if we assume that the S22
that is usually shown when talking about PA's is small signal S22 then
how come the match could be so good when the designer actually didn't
match for it---it was matched for loadpull! any comments/suggestions.
Post subject: Posted: Wed
Jul 19, 2006 10:51 am
Joined: Mon Jun
27, 2005 2:02 pm
PA matching is different than matching a small signal amplifier
for maximal output power. In PA many times you don't necessarily want
to get the maximal output power, but a good linearity or efficiency.
This results in different impedance contours. The best way to measure
it is with Loadpull. So with the Loadpull you can sweep the output impedance
to different locations on the Smith chart for different conditions (Like
best linearity or efficiency, which come on the expense of each other).
The best S22 (The lowest) doesn't necessarily means that your PA
will give you what you wanted to achieve from your design, because you
match for other conditions and not for lower S22.
Posted: Wed Jul 19, 2006 11:35 am
Jul 18, 2006 9:52 am
thanks for the reply
IR. but let's say i did loadpull for a class A (linear) PA and found
out that I need to match 50 Ohm to A+jX impedance, which is different
than the actual output impedance of the transistor (or set of transistor
cells). Hence, the match that I do will not be a conjugate match, right?
So, my question is that then why in books and in technical published
papers people talk about S22 and it is usually quite good and they highlight
it also? Even Cripps, on page 19 or 20 does so for a simple linear Class
A PA design.
as i said in the earlier post, it could be that
the S22 is shown for small signal but then why is it good because the
match wasn't done for small signal S22/conjugate match??
Post subject: Posted: Wed Jul 19, 2006 2:29
Joined: Mon Jun 27, 2005 2:02 pm
This is because in Class A design
you get a good linearity and bad efficiency, so you need to match for
conjugate in order to get maximal output power to improve your efficiency.
Besides Class A there are other classes like A-B or B, in which you
will get a degraded linearity but good efficiency. Therefore, in those
matching for conjugate is not required and you need to match for good
linearity, which will not necessarily yield a good S22. I was referring
in my previous post to those type of amplifiers -especially Class A-B,
which is most common in Base-Stations PA's.
Posted: Wed Jul 19, 2006 3:01 pm
Jul 18, 2006 9:52 am
hey IR, i don't know if i fully
agree---i am not sure. even in class A you don't do conjugate match
of you transistor out impedance; rather you transform 50 Ohms to some
impedance that you want your transistor to see that you get from loadpull.
let's say your transistor output impedance is 10-j3; in PA (eve for
class A) you don't match for 10+j3 rather some other value that you
find from loadpull (let's say it is 15+j3) and will provide optimal
output power but not necessarily maximum power gain. so you transform
your 50 Ohm to this 15+j3. so your small signal S22 shouldn't be quite
good because instead of having a conjugate match of 10+j3 you have something
else. but look at some PA papers and if they give out S22, in many cases,
it is as if it is conjugate matched to the transistor's actual output
impedance. may be i don't have a clear concept of S11/S22 or something
but it doesn't make sense.
Post subject: Posted: Thu Jul 27, 2006 3:10 pm
Joined: Mon Jun 27, 2005 2:02 pm
The transistor can be matched to any condition.
The conditions can be e.g.: NFmin, Specific Gain, Maximal Output Power
etc. For some of these conditions (NFmin, G) there are families of countours
that are plotted on a Smith chart, which specify the impedance the transistor
needs to be matched to for a given Gain and/or NF. You can match to
fulfill 2 conditions simultansously (An intersection of 2 countours).
These impedances will not necessarily will give you the best S11 or
S22 but they will provide you a maximal gain or minimal NF.
Attached is a link for an application note with some examples:
http://www.odyseus.nildram.co.uk/RFMicr ... torial.pdf
Post subject: Posted: Tue Jan 30, 2007
Joined: Tue Mar 15, 2005 11:43 pm
Just want to add notes
that the PA matching that has been mentioned about is large- signal
matching. As we understand that the output impedance at small-signal
conditions is different than large-signal conditions.
to Cripps method that large-signal output impedance is lower than small-signal
output impedance. (The large-signal impedance is 1/2 to 1/3 of small
I think it's one of the reason that large-signal
matching is not conjugate matching to the small-signal S-para (S22).
I hope it makes sense.