[Amps] bipolar input and output impedance

[Manfred Mornhinweg]

Hi Dan,

> What are the bipolar HF power transistor specs that influence or 
> determine the input and output impedance? 

The output impedance is a rather simple matter. If you like simplified 
formulae, take the output impedance as the square of the supply voltage 
divided by twice the power. And for the impedance between the collectors 
of a push-pull pair, simply use the same formula, but doubling the 
supply voltage.

An example, for a typical 100 watt push pull stage fed by 13.8V:

13.8 times 2 (for the push pull) is 27.6. Squared, it is about 762. 
Divided by 200 (twice the power output), is 3.8. That would be the 
approximate collector-to-collector impedance.

Now, do you want it more precisely? It's not all that difficult, and 
it's no black magic. When you feed an amp with 13.8V, the peak RF 
voltage the transistor can produce is a little less than 13.8V, because 
of its saturation voltage. How much, depends on its characteristics, and 
how you are using it. At higher frequency, the saturation voltage is 
higher too. But let's take 1 Volt, which is typical. So the peak RF 
voltage is 12.8, and between the collectors of a push-pull stage, it's 
twice that, or 25.6V. And since the RMS voltage is 0.707 of the peak, we 
have 18.1V RMS between collectors.

You want 100 watts. Since W = V * A, we need 100/18.1 = 5.52 amperes of 
RF through the transformer primary. And by Ohm's law, 18.1 V divided by 
5.52 A is 3.28 Ohm. That's the real output impedance of this stage. And 
the discrepancy with the 3.8 Ohm given by the simplified formula is 
simply because that formula assumes the transistors are perfect, 
saturating at zero Volt! Plug in 12.8V instead of 13.8V in that simple 
formula, and you will get the correct result, as long as your 
transistors really saturate at 1V.

All this is for linear service. If you are willing to distort the 
envelope of teh signal, then you will be driving the transistors into 
saturation, resulting in the RMS voltage being more than 0.707 times the 
peak. Up to 0.9 would be typical, at lower frequencies. This will make 
the output impedance higher, for a given power and voltage. Or expressed 
the other way around, a 100 watt, 13.8V amplifier like those used in 
almost all our modern HF radios can deliver only those 100 watts in 
linear service, but if overdriven will produce up to about 160 watts on 
the lower bands, at the same impedance levels, but with severe 
distortion (the splatter will make you famous if you do this).

Input impedance cannot be calculated as simply. A good practical 
assumption is that it's about the same as the collector impedance! That 
is, as long as the transistors don't have internal impedance matching. 
VHF transistors often do, and then the input impedance is higher. And 
the input impedance of non-ballasted transistors is lower, but 
non-ballasted RF power transistors are pretty much an extinct species 
today. They weren't fit for survival in the wild.

Anyway, this gives just the magnitude of the input impedance. Its phase 
angle can easily be off by as much as 60 degrees, to either side. A 
typical transistor will have inductive input impedance at some 
frequencies, and capacitive at others. That's why most designers choose 
to swamp the bases with low value resistors. That gives a more 
predictable impedance into which to design the drive transformer and the 
feedback network, and also adds a lot of stability, at the cost of some 
gain.

By the way: Often it's a good technique to design the input transformer 
to work basically into the reflected impedance of the feedback network, 
and then add the transistors almost as an afterthought! If the feedback 
network impedance is low enough, this works like a charm.

The output impedance also has reactive components, but they tend to 
become really important only when the transistor is operated in the 
higher part of its frequency range.

> I have an ENI circuit using 
> TH430 transistors with 4:1 input and output transformers, the EB-63 
> circuit based on MRF 454s uses 16:1 transformers. 

If the two are of roughly the same power level and supply voltage, then 
of course you got something wrong there. Perhaps one is talking about 
turns ratio, and the other about impedance ratio? A turns ratio of 4:1 
yields an impedance ratio of 16:1. And that's indeed the correct ratio 
to use for a 100 watt push-pull stage running from 13.8 volt. It will 
transform 50 ohms into 3.125 ohms, very close to the output impedance of 
such a transistor pair, and the base-to-base impedance will also be 
close enough to that.

 > What transistor parameters influence transformer design?

All of them! ;-)

Kidding aside: For an HF broadband stage, you can pretty much do a 
design based on the power level and supply voltage, and then pick some 
transistors and plug them in. At VHF and higher, and generally for 
resonant designs, it's important to pay closer attention to the specific 
transistor's characteristics. The manufacturers usually give at least 
some information about the impedances of their devices over the intended 
operational range.

Manfred.

========================
Visit my hobby homepage!
http://ludens.cl
========================