Smith Chart is giving me an answer I didn't expect - Am I using it right?

Question

I'm trying to track down a VSWR issue with a RF power amplifier (25 W, 150 MHz). The radio we're using to drive it has a VSWR alarm, which triggers during transmit. Older manufacture runs of the PA don't trigger the VSWR alarm, and I'm not certain that the bill of materials used to build the current manufacture run is correct.

I've entered the power amplifier input circuit to an online Smith chart tool (https://www.will-kelsey.com/smith_chart/) and it shows me the input impedance is nothing like 50 Ω. Given my inexperience with this type of tool, and the experience of the engineer who designed the amplifier, it's more likely an error on my part than on the part of the original designer. I want to confirm that I'm using the tool correctly before I dive in and make wholesale changes.

Here is the input circuit. During transmit TR5 switches current through L3-D9-L11-D11, so I've modelled D9 & D11 as shorts. The input impedance to IC4 (RA30H131M1) is 50 Ω according to the datasheet. Receive signal comes through L12 / C51-52.

I modelled the TX detect circuit (D13-TR5) as a 100 pF capacitor in series with 220 Ω to ground, but tried values between 50 Ω and 10 kΩ and didn't see much difference. I left out the 2.2 µH and 1 nF parts as these don't affect the RF performance. This modelled as in an input impedance of 22+16j Ω. I noted that DP5 is close to 50 Ω, If my modelling is correct, then I guess there is a component issue between the input and C34.

So, I have three questions:

1. Am I making valid assumptions about the operation of the PIN diodes and large value parts?
2. Am I entering the data correctly into the modelling tool (e.g. correct order left-to-right)?
3. What is a suitable model for the TX detect Circuit - This has significant effect on the impedance (but I don't think this is the main cause of the mismatch I'm seeing).

As always, Any help is appreciated.

Answer 1

Your problem, essentially is that the VSWR is too high and it triggers an error. Any assumption you will take in those modeling simulation won't tell you were is the issue since it will be hidden in the assumption you take. For instance, if you doubt your BOM, maybe they used passive component with different frequency response, your model, won't show you that because you are assuming the component. Therefore, I'd say that this model is a bit useless. To give a parallel, it's like if you add a board that used to work then on the next prod, you used a spice simulation to understand what doesn't work. You won't be able to conclude.

Q1

At those frequencies, I'd say, most likely they don't behave like a short. The impedance they will show will also be power dependent. So I think it's a too broad approximation.

Q2

Yes and no, it really depends what you want to know. Both can be viable. Right now, you are measuring the impedance from your port. I doubt that this is what you want to measure. Your objective is, probably, to match the amplifier to thee rest of your circuit. If so, the impedance of your amplifier should be the conjugate impedance of the rest of your circuit to minimize SWR and reflections.

Q3

As I mentioned in introduction, I doubt that modeling will lead you anywhere (unless I don't understand your issue). So, I wouldn't recommend you to model either. My recommendation would be to use a network analyzer (VNA) to measure the impedance on both direction. You will see instantly which part of your circuit has a mismatch. If you don't have access to a VNA, you might be able to measure SWR with other mean like with an oscilloscope.

Answer 2

I took some measurements on a VNA, and neither the old or the new amplifier looked anything like the output from the online tool. Nevertheless, the old one looks good, but the new one is terrible, so I think I'm going to be swapping parts one-by-one until I find the error.

Original

New