S Parameters : Output Impedance and S22

[G0HZU]

It's not obvious what bias point you have gone for but it is probably less than 3mA. Either way, your simulation shows a reflection coefficient greater than 1 at about 70MHz and this is a bit unusual for a 2N3904 in your circuit. This indicates negative resistance is present at the input and this is bad for stability.

Can you plot the K factor for just your 2N3904 using your simulator?

In my case, I plotted K as in the plot below and it showed K was above 1 just above 70MHz for the BJT on its own. So I designed an amp at 70MHz. See the K factor plot below that shows the BJT on its own (the second image below).

At lower frequencies than 70MHz K is less than 1 and the circuit is only conditionally stable. At these lower frequencies, having an inductive load at the collector will cause negative resistance at the base. This makes this type of amplifier very prone to instability if this is the case. That's why I chose 70MHz as K was above 1 here. So I could add a tuned load without generating negative resistance at the base at 70MHz.

See the second image below for the K plot for my 2N3904. A typical common emitter BJT will show K below 1 at lower frequencies. Up at VHF this will eventually rise above 1 and stay there until some point up at UHF where it will often dip below 1 again.

I'm not sure how accurate your Spice model is for a 2N3904. Try plotting K factor at 10V and 5mA Ic and compare it to my real measurement below.

If I had tried to design an amplifier down at (say) 20MHz where K is less than 1 I would have to add some damping resistance somewhere to prevent negative resistance at the base at 20MHz after I add a tuned load to the output tuned to 20MHz. The negative resistance usually occurs just below the resonant frequency of the output network. This is the region where the output load at the collector looks to be net inductive.

It's generally a good idea to add some damping resistance anyway as this can then achieve unconditional stability at all frequencies. I usually plot K and B1 factors when looking at this stuff.

Adding the 680R shunt resistor R1 at the output of the tuned amplifier gets K above 1 everywhere. This causes a small reduction in gain but this is worthwhile as the amplifier is then unconditionally stable at all frequencies. See the first plot below.

[jamfletch]

here's what I've got when Ic is biased to 5mA at 10V. My stability factor is <1 beyond 2MHz, I've also tried another SPICE model for the 2N3904, and I get different results, but I'm wondering if this is an issue with me importing the file into QUCS Studio,

What Bias circuitry did you use for the results you measured on the VNA? Was it with a grounded emitter, or degenerated?

Would you be able to send your .s2p file too?

Also, out of curiosity, what software are you using for your simulations? It looks like uW Office?

Many thanks

[G0HZU]

R4 is wrong at 2k as you only have 1V at the collector. Change it to 100R or less and then adjust the potential divider to get 5mA collector current.
I used very basic biasing but the emitter was ac grounded. The SW I’m using is Eagleware Genesys. It’s 20 years old, but is still very good.
The input cap value is quite low at just 24pF.

[Joel_Dunsmore]

2. Measuring S parameters requires that all ports be terminated with the system impedance Z0. This is actually fundamental to the definition of S parameters. Because the ports are terminated with Z0, any waves exiting the DUT will be totally absorbed at the ports, nothing is reflected back towards the DUT.

Small nit to pick: Measuring S-parameters can be done in any impedance.  The impedance of the VNA (expressed as a reflection coefficient) is a_n/b_n.  And so we have a relationship between a's and b's at each port (a and b waves being a kind of sqrt of power wave, quite related to a voltage wave) .  So S-parameters can be measured into any impedance and use eqn 1.21 of my book to solve for the s-parameters.

BUT: this is only true, of course, when the DUT is operating as a linear device.  All this math goes away when the DUT is operating non-linearly.

[jamfletch]

Just a quick update. I've made a 2N3904 common emitter amplifier with an output matching circuit, used with components I have lying around. The output match and gain at 100MHz are respectable. The gain is lower than I simulated though, possibly due to inductor losses?

[jamfletch]

Output match measurement attached