Sunday, March 5, 2023

The Ham Radio Series 36 - What Are S-Parameters and Why Should I Care?

I mentioned in February of ‘22 that I had bought a NanoVNA.  If you’re unfamiliar with the term, VNA is short for a Vector Network Analyzer, a test instrument that can measure the complex impedance of something, and a handful of different NanoVNA models have hit the amateur market in the last few years.  In particular, I bought the H4 model for a couple of reasons; the fact that it has a larger LCD than most of the other models and that I’ve heard good things about its ability to work to 1.5 GHz out of the box.  

It’s hard to convey just how big a deal these things are to people who are new to the game, but through some clever tricks in digital design, they’ve absolutely crushed the price of test equipment.  The first 1.5 GHz vector network analyzer I ever touched, back in about ‘84, was an exquisite piece of lab equipment from HP – who was just the test equipment giant in those days – that cost over $100,000.  The NanoVNA H4 today is under $100, or right around there.  That’s far less than 1/1000 of the cost of that HP, partly because of the constant depreciation of the dollar by the Federal Reserve.  Sure, the $100 Chinesium (almost always) NanoVNA doesn’t have as good specifications as the HP did, but a close approximation to the value is better than a guess or assuming “it should be good.” 

The default display that my NanoVNA starts up with is might be unique to this software, but from reading a NanoVNA users’ group for a few years, I think it’s pretty common.  It overlays four channels of information: three different rectangular plots of one S-parameter per plot and a Smith chart of one those S-parameters.  Since I’ve been working almost exclusively with S-parameters since the early 1980s, I’m at home with them, but questions about what those things mean are rather common. 

You can see a four trace display on this sample NanoVNA; I couldn't begin to guess what kind of thing it's sweeping from 50 kHz to 900 MHz.  (Image clearly from DX Zone; they don't sell these (or anything that I know of) but are an information aggregator.)

So what’s an S-parameter?  The S stands for scattering, but what does that mean?  Ever seen the “one way glass” that’s used for security (or used to be)?  The light is scattered in the favored direction to get it to work the way they want.  If you look at the glass from the outside, you see your reflection; the security guy inside sees you plainly but doesn’t see his reflection as clearly as you see yours.  If there’s a poorly placed light on his side of the glass, you may be able to see him as well.

In RF circuits, when you input a signal into an amplifier, filter or really anything, a (hopefully) tiny amount is reflected back.  Most goes through the network, exactly how the input changes to the output is what’s being measured.  Some of the output maybe reflected back to the input because the Device Under Test’s output impedance doesn’t match the circuit it’s connected to. 

While S-parameters can be measured and calculated for three, four, or any number of ports, the most widely used are for two port networks.  The four parameters are S11, S21, S12 and S22; you’ll see those written that way or as subscripts: S11, S21, S12 and S22.  If you want to be published, subscripts are probably the way to go; since it’s a bunch of extra work to use subscripts in Blogger’s editor, please forgive me for using plain script.  In real life, you’ll see both formats used.  In all cases the format is S (port measured at, port measured with respect to); in other words, S11 is the signal measured at port 1 with respect to port 1; S21 is the signal at port 2 with respect to port 1.  

Direction is all important;  S21 is the signal at port 2 with respect to the input at port 1; if it’s bigger than the input, the thing being tested has gain.  S12 is the signal at port 1 with respect to port 2; usually called reverse isolation.  While a passive network, like an attenuator, will have S12 = -S21, an amplifier won’t resemble the inverse of the gain (unless it was designed to be, amplifiers don’t work like that).  For example, the typical one or two transistor amplifier ICs you’ll buy might have an S21 of 10 or 15 db, but an S12 of 6 dB.  

In the case of S11, the input impedance, a signal from the VNA is applied to port 1 and the amount reflected to port 1 is used to calculate Return Loss – a single number for the input impedance in dB; the direction separates the signal going to the input and the (expected to be) much smaller signal returning to the VNA port 1 determines the value of S11.  It’s the same basic description for the output Return Loss; the VNA measures the signals going both directions on the output pin.  Return loss is expressed in decibels by the VNA, but since it’s a measurement of reflected signal, it’s conceptually the same as the Standing Wave Ratio, or Voltage Standing Wave Ratio (SWR or VSWR) you’ve seen before.  

When measuring S11, port 2 is grounded and for S22, port 1 is grounded.  Mathematically:

The value the VNA displays it 20*log of these calculated values. 

So what are these S-parameters and the NanoVNAs good for?  To begin with, they’re entirely a tool for people who build things, whether antennas, passive circuits like filters or RF switching networks, and active circuits like amplifiers. They’ll test a length of coax for loss and other things you’ll want to know.  If you’re not a “home brewer” no test equipment is useful.  Exactly how useful depends on what you’re measuring.  If you’re working on something like a lowpass or bandpass filter, they’ll tell you everything there is to know.  If you’re working on an amplifier, they’ll tell you gain, and input and output matches.  Those are the big characteristics for an RF amplifier but they won’t tell you noise figure, linearity, intercept points or other things you need to know – depending on the application for the amplifier.

One thing that I’ve used the NanoVNA for several times, as well as my first, one port, VNA, my antenna analyzer, was to export a file of S-parameters of an antenna scan into a Smith chart to design a matching network.  Coming soon will be an introduction to the Smith Chart. 



15 comments:

  1. Great primer, SiG! I'm about from "Your Era", which is why I have a preference for HP equipment. Did the HP VNA you used have the funky "mecahnical arm" with APC-7 connectors on them? IIRC the one at Hughes where I worked had them.
    I bought a NanoVNA shortly after you published one of your articles on it, and I'm impressed. Been using it to weed out crappy connectors, bad cables, and other things besides just playing with it. Haven't used it yet with a filter or amp, but the amount of power these things pack is astounding, and all for a measly "hunnert bucks".

    ReplyDelete
    Replies
    1. ISTRC it had APC-7 connectors on the front panel of the bottom box, not a mechanical arm. I should try to look up VNAs from that era and see if I can find the model.

      It was in a 19" rack and several boxes tall. I was there when I started working there. I'd have to try to dig up some details.

      Delete
    2. "It was there when I started working there", not "I was there..."

      Delete
    3. I would bet that BOTH versions are an accurate description.

      Delete
  2. When measuring S11, port 2 is not grounded, it's terminated, etc. Just a minor corrections, nothing to see here!

    ReplyDelete
    Replies
    1. I was going by the equations in that graphic showing them. Top left of the bottom. It says a2=0 as the condition. Notice that a and b are swapped top to bottom as you go left to right. Also notice the box represents the network (Device Under Test), not the VNA.

      Delete
    2. Looking at p. 2 of HP-AN-95-1, you will see S11 = B1/a1 with a2 = 0 = Input reflection coefficient with the output port terminated by a matched load ....

      Delete
  3. I started my microwave career on one of these. I was the young kid who could spell RF and write code. https://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1970-02.pdf

    ReplyDelete
    Replies
    1. And this is probably still the best tutorial on s-parameters you can find. There was a myriad of good technical information in the HP app notes from that era. http://hparchive.com/Application_Notes/HP-AN-95-1.pdf

      Delete
    2. D'oh! I should have remembered that one.

      I would have sworn I had a copy on this machine, but nope.

      Delete
    3. Thanks for the link, John. Not sure if I've seen that before, HP is an absolute gold mine of technical documentation. They wanted people to be able to get the most out of their purchase, and had excellent support. HP Manuals are like Heathkit manuals...The Gold Standard of "How To Do It Right".

      Delete
    4. The unit in the left rack, just above the desk top, has the "Mechanical Arms" I mentioned, in case you've never seen them. Amazing things, with swivel joints, airline "coax", and very well made.
      The young guy sitting there could have been me! Geek glasses, respectably longish hair, and you just know he's carrying a loaded pocket protector.....

      Delete
    5. I remember the swivel arms well. I also remember that they weren't quite phase stable and match invariant. I had to repeatedly explain to design engineers that you could not measure high Q devices using those arms because the calibration would quickly become invalid.

      Delete
  4. Thanks for that. Seems funny that in neither of the two ham presentations I attended, on the NanoVNA, nobody explained the S11, S21 labeling, with the arrow graphic.

    ReplyDelete
  5. No, No, No, I was a Satcom tech in the USAF before I retired and then did it for 10 years as a military contactor. I want nothing to do with RF technology. I drive a CDL Hot Shot truck now as a retirement job.

    ReplyDelete