Designing tuned amplfiers

[[email protected]]

Hi.
I'm studying the design of simple tuned BJT RF amplifiers and I have
some questions.
Should I design my amplifiers to be unilateral? Otherwise, how do I
deal with the problem of mis-alignment?
This has been my approach so far to avoid misalignment between the
input and output tuned circuits (due to y21), but I can only realize
low power gains with good stability.

(I don't have a vector analyzer and the transistors I'm using ain't
that great, so I cant accurately determine transistor capacitances)

Regards,
Roger

 

[John Woodgate]

I'm studying the design of simple tuned BJT RF amplifiers and I have
some questions. Should I design my amplifiers to be unilateral?
Otherwise, how do I deal with the problem of mis-alignment? This has
been my approach so far to avoid misalignment between the input and
output tuned circuits (due to y21), but I can only realize low power
gains with good stability.

(I don't have a vector analyzer and the transistors I'm using ain't
that great, so I cant accurately determine transistor capacitances)

There is a book that goes into this in great detail:

Theory of transistor bandpass amplifiers, W. Th. H. Hetterscheidt,
Philips Technical Library 1964. No ISBN at that time.

When it was written, no transistors were 'all that great', so it should
be very helpful. These days, people use transistors that are 'great'
enough not to worry about neutralization, still less unilateralization.

With regard to measuring transistor capacitances, there used to be a few
pieces of test gear that would do this, a low-excitation voltage RF
bridge with control over DC collector voltage and current, but I suspect
there isn't much available now. You may have to build something.

 

[Ken Smith]

Hi.
I'm studying the design of simple tuned BJT RF amplifiers and I have
some questions.
Should I design my amplifiers to be unilateral? Otherwise, how do I
deal with the problem of mis-alignment?

You only need to neutralize to prevent oscillations. If you use
adjustable inductors, you can align the oscillator by trimming.

The method is to trim the base side close to right, then do the collector
and then go back to the base and get it exact. Trimming the collector
will effect the base but trimming the base won't effect the collector
(much).

This has been my approach so far to avoid misalignment between the
input and output tuned circuits (due to y21), but I can only realize
low power gains with good stability.

(I don't have a vector analyzer and the transistors I'm using ain't
that great, so I cant accurately determine transistor capacitances)

With "not so great" transistors, low power gains may be all you can have.

Also remember that the impedances of a transistor change greatly if you
run it class B or C.

 

[[email protected]]

So basically you recommend tuning out the parasitic capacitances with
inductors?
Another approach I considered was using transformers at the input and
output to reduce the effective capacitance by n^2 and make then
negligible. Does that sound feasible?

Another question I have is how can I accurately impedance match to the
input of a common emitter stage? It seems easier for a common base
stage since the input impedance is more predictable (1/gm vs B/gm for a
CE stage).

Regards

 

[Ken Smith]

So basically you recommend tuning out the parasitic capacitances with
inductors?

Yes, in the narrow band case, this works very well. You still have to
make sure there isn't too much real component in the collector to base
feedback. You don't want an oscillator.

Another approach I considered was using transformers at the input and
output to reduce the effective capacitance by n^2 and make then
negligible. Does that sound feasible?

Yes, but you give up a lot of power gain to do that. You will be driving
the base with an impedance much lower than a matching impedance.

It can also make the problems on the previous stage worse. The loading of
the collector of one stage bu the base of the next tends to reduce the
chances of an oscillator.

Another question I have is how can I accurately impedance match to the
input of a common emitter stage? It seems easier for a common base
stage since the input impedance is more predictable (1/gm vs B/gm for a
CE stage).

At RF frequencies, the input impedance of a transistor is a complex
matter. The "little Re" still has 25mV on it but you have lots of other
stuff in the circuit.

For RF transistors, you can uses the maker's datasheet to find the input
impedance in the intended operating mode. The value they give is just
about the middle of the distribution.

If you want to find it out for yourself and are'nt working at too high of
frequencies, you can make a circuit that allows you to adjust the
impedance at the base. The simplest is a low valued POT from a signal
generator.

Regards

 

[John Woodgate]

So basically you recommend tuning out the parasitic capacitances with
inductors?

With 'not so great' transistors, you are liable to end up 'chasing your
tail'; the thing will obstinately refuse to align and the bandwidth you
get will be anyone's guess. You get 'throw-ups', too, at the pass-band
edges, 10 dB or more, where the y12 feedback is nearly causing
oscillation.

Another approach I considered was using transformers at the input and
output to reduce the effective capacitance by n^2 and make then
negligible. Does that sound feasible?

It's normal to use transformers, or tapped inductors or, less common but
with notable advantages, capacitive tapping (you can vary the tapping by
changing C values; you don't have to make new inductors). This is so
that your tuned circuits have more practicable component values.

Another question I have is how can I accurately impedance match to the
input of a common emitter stage? It seems easier for a common base
stage since the input impedance is more predictable (1/gm vs B/gm for a
CE stage).

The input admittance of your CE stage is y11, and if you have values of
it for your devices, you'll see that beta/gm is a very rough
approximation. Because y11 is very much smaller than gm, you don't need
such an extreme tap-down on your tuned circuit, so it's actually easier
to match than a CB stage.

But the usual way to get stability with 'not so great' transistors is to
deliberately mismatch at both input and output, so that the low source
impedance and low load impedance keep the transistor from oscillating.

This is all explained in very great detail in the Hetterscheid book I
pointed you at.