|
|
|
|
Wideband Mosfet matching - RF Cafe Forums
|
Extreme
Post subject: Wideband Mosfet matching Posted: Wed Oct 31,
2007 7:07 am
Captain
Joined: Sun Feb 04, 2007 2:58
pm
Posts: 9
Hi people,
I would like to learn how to match
this mosfet to 50 ohm in wideband 88-108 MHz.
http://www.mitsubishichips.com/Global/c
... 06hvf1.pdf
I would like to know how to start,and what tools
can be used.
Not asking for a fully explanation,but maybe some steps
how to deal with this.
Thanks
Top
IR
Post subject: Posted: Wed Oct 31, 2007 2:18 pm
Site Admin
Joined: Mon Jun 27, 2005 2:02 pm
Posts: 373
Location:
Germany
Hello,
I had problems to view the data sheet you
provided.
Anyway, general rules for matching of RF power transistor:
1. Detemine the DC operating point of the device.
2. Measure
the S-Paramaters at this operating point, or use the
S-parmaters
provided by the manufacturer if applicable.
Note: Some manufacturers
provide a more comprehensive simulation model (Like Spice model), which
enables to perform also Harmonic-Balance simulation.
3. Design
the matching networks based on the the provided/measured
S-paramaters
for the required bandwidth of operation.
4. Ensure that the
following conditions are met:
- Stability for the entire operating
band and also out of it as much as possible.
- Achieving the required
gain.
Tools which are useful to perform such design are: ADS,
Microwave Office.
Hope this helps!
Top
Feng
Post subject: Posted: Wed Oct 31, 2007 2:35 pm
Captain
Joined: Sun Mar 19, 2006 5:25 pm
Posts: 21
Location: Burnaby
IR is right. The only S-parameter you can use is on the situation
of Vdd=12.5 and Id=500mA. There are smart components in ADS for wideband
matching, and you can find them at Impedance Matching. After you figure
out the initial parameters of matching circuit, you may simulate it
in SP or HB by using the s2p file created from the provide S-paramters.
As IR suggested, you may measure the S-paramters in order to
do sweep simuation and observe the operating states.
Top
Extreme
Post subject: Posted: Wed Oct 31, 2007 2:35 pm
Captain
Joined: Sun Feb 04, 2007 2:58 pm
Posts:
9
I realize now,i ask an impossible question.
Let me ask
you guys some partial question.
But i rediscovered RF-sim99
i have on my system.
It has an 2 port parameter function.
I
copyd the S-parameters from the datasheet and made a S2P file of it.
The curve was on the smith chart and @100MHz i can read some
data:
This is just output data.
Imp. 24.52R-j38.02R
Series
equivalent 24.52R/41.87pF
Parallel equivalent 83.45/29.56pF
With an automatch i get a sharp match with 100nH and 32pF.
First question:
Is it correct to say:
@100MHz the output
of the fet can be compared to real 24.52R with a 41.87pF cap in series.
Or a real 83.45R with a 29.56pF cap in parallel?
I think
i'll wait for an answer before asking more questions.
Top
IR
Post subject: Posted: Wed Oct 31, 2007 2:55 pm
Site Admin
Joined: Mon Jun 27, 2005 2:02 pm
Posts:
373
Location: Germany
Hello Extreme,
Yes what you wrote
is true.
Only one correction: For parallel equivalent the imaginary
part is inductive, as the lower part of the Smith Chart for admittances
is inductive.
Top
Feng
Post subject: Posted:
Wed Oct 31, 2007 2:57 pm
Captain
Joined: Sun Mar
19, 2006 5:25 pm
Posts: 21
Location: Burnaby
I thing you can
say that, but them vary with the frequency. Therefore, it is not a good
news for a wideband matching. By the way, what is the bandwidth of the
sharp match you got?
Top
Extreme
Post subject:
Posted: Wed Oct 31, 2007 3:32 pm
Captain
Joined:
Sun Feb 04, 2007 2:58 pm
Posts: 9
I have to say i am a absolute
beginner in calculation/simulation with these kind of things.
Hobby'ing
with RF for only 2 years now.
Mostly building designs or combine
schematics.
http://fets.arecool.net
But i like to know why i
build it this way,and learn designing myself.
But Feng,you bring
me to another question.
You asked how wide my matching is.
I
have a strong dip in the reflection line @100MHz
I think the wider
the match, the reflection is getting more
So there has to be a compromise
between them.
What would be an acceptable reflection coefficient
for 88-108.
25dB or higher?
Top
Feng
Post
subject: Posted: Wed Oct 31, 2007 4:34 pm
Captain
Joined: Sun Mar 19, 2006 5:25 pm
Posts: 21
Location: Burnaby
I visited your website and saw many amazing works although I don't
know the language on it.
The practical return loss depends on
the isolation between the transmitter and the antenna as well as how
many reflection power can be endured by the transmitter, although the
reflection power should be as less as possible. I really don't what
is better for your transmitter, but for a basestation of mobile communication,
the VSWR should be less than 1.3, the corresponding return loss is 17.69dB.
Top
Extreme
Post subject: Posted: Wed Oct 31, 2007
5:12 pm
Captain
Joined: Sun Feb 04, 2007 2:58 pm
Posts: 9
Ok now i can answer.
@18dB reflection and simulation
@100MHz i get a bandwitdh from 94 to 106 MHz.
Just thought of
someting:
If i would create a matching from coils and C's from
Fet to 25Ohm @92MHz.
and followed by a matching from 25ohm to 50ohm
@102MHz.
Could this be a way to increase bandwidth?
Thanks guys for your fast answers for so far.
Top
Feng
Post subject: Posted: Wed Oct 31, 2007 5:30 pm
Captain
Joined: Sun Mar 19, 2006 5:25 pm
Posts: 21
Location: Burnaby
It needs to calculate the serial connection matching
networks first, and then you will find out whether this method increase
the bandwidth or not. You may use simulation software to figure it out,
or transform to Z matrixes to get their product.
Posted 11/12/2012
|
|
|
|