# cascode resonance frequency

Discussion in 'Electronic Design' started by Matteo, May 25, 2007.

1. ### MatteoGuest

hi all
I'm trying to design a cascode amplifier in resonance, here is the
schematic:
http://www.thebags.it/listing/uni/cascode_sc.png

my project target are:
gain 20dB
resonance frequency 850Mhz
BW 100Mhz

but I've a problem ..I am not able to set the resonance frequency at
850Mhz; every combination of C and L I try I always get these
resonance frequency: 629Mhz, 792Mhz, 997Mhz, 1256Mhz.

here is a little parametric analysis:
http://www.thebags.it/listing/uni/plot.png

how is it possible? I tried to edit all the circuit parameter in order
to understand the reason but I always get these values

any help is appreciated!

thanks

Matteo

2. ### Andrew HolmeGuest

It looks like your parametric analysis is only plotting data points at
those ordinals. The data points do not fall on the true peaks. You
need to increase the plotting resolution.

3. ### MatteoGuest

I was plotting only 10 points per decade! ..I set it to automatic and
it runs perfectly!
thank you andrew ..it was driving me mad
matteo

4. ### MatteoGuest

I was plotting only 10 points per decade! ..I set it to automatic and
it runs perfectly!
thank you andrew ..it was driving me mad
matteo

5. ### Tam/WB2TTGuest

You really don't want to do it this way, unless you absolutely don't care
about noise figure, and have a signal that has previously been band limited.

On the input, you need an LC matching network to match the source to the
gate input impedance. No physical resistors. Let the resistive part of the
gate input impedance control the loaded Q.

On the output, same thing; let the input impedance of the next stage be the
R.

Getting 100 MHz bandwidth is going to be difficult with simple 1L/ 1C
networks. You would need an effective Q of about 8 for the two stages
combined. Without doing all the math, a Q of 7 on the input and a Q of 4 on
the output might be close.

Tam

6. ### Robert BaerGuest

I believe the term you are looking for is "complex conjugate" matching.
Something else to look into is "staggered tuning".
One nice trick is to use an over-coupled transformer; that
over-coupling widens the (tuned) bandwidth if primary and secondary are
tuned the same (as i remember from~30 years ago).

What initially threw me for a loop was C0 and C1 in *Farads*, which
is insane.
The second insane thing was that C0 was specified to SIX significant
digits!!!