Pre Amps....some Measurement thoughts.

Nearly every VHF/UHF/SHF meeting now includes the infamous and good fun 'Noise Figure Measuring Contest'.

I will attempt to explain and illustrate that a simple list of results in a table showing only gain and NF of the devices under test (DUT) should not be considered as a 'definitive result'.

I believe the majority of participants just want to know "where they are" with respect to what they have just built and/or are using. Most people simply do not have frequent access to good test equipment, which is prohibitively expensive to be dedicated to 'amateur' applications/situations.

In recent years, where I have been involved in the measuring processes, I have attempted to try and broaden the view from that of a 'contest' into one of information exchange and education. Rather than simply grabbing a device out of the collection box, measuring it, scribbling a number on it, and then chucking it back in the box. It is much more productive for everyone present to look inside, fiddle with it, poke it, look at the techniques used ( with the owners permission ) and share the collective groups experience to suggest possible improvements that could be adopted.

The simple measuring of a noise figure and the associated tables constructed do contribute positively to the ongoing struggle/race to improve peoples receivers. But it is clear that the process of measurement, the results obtained, and the accuracy of the results are eithe r; not clearly understood, or, at least not acknowledged.

I have repeatedly stated in various discussions at these events I have attended, and in emails ;

"if the measurment indicates that you broke the laws of either physics or common sense then it's the measurement that's broke and not the laws"

This is an over simplified statement and could be improved to; "if the measurement indicates that you broke the laws of physics or common sense then you should try to quantify the possible range of accuracy/uncertainty of the measurement and then go on to try and investigate the apparent anomoly"

I am repeatedly seeing NF. claims of 0.16 dB and 0.18 dB at UHF/SHF. Manufacturers data for the devices used and even a primitive session in any decent simulation software shows that even a genuine 0.3 dB is highly improbable at bands higher than 144 Mhz, using room temperature techniques, so perhaps it's time for a reality check.

I'm also repeatedly told; "ok we cannot say that these measurements are absolute, we accept that, but we can say that all DUTs measured on our test setup can be compared to each other equally, so this is a good relative comparison result"

.......well, it can be......

If the data provided by test equipment manufacturers is considered properly, it is readily seen that the noise figures we are now attempting to measure with modern devices are, in many cases, lower than the margin of error, or "uncertainty" , in the equipment being used + uncertainties introduced by parameters of the DUT that are often not being measured and completely discounted/neglected.

Further analysis of the literature available on the web (go to Agilent's site and take the various application notes available and the uncertainty calculation tools) clearly highlights where the largest sources of possible measurement error/uncertainty occur.

It is not my intention here, nor is it necessary with all this excellent professionally prepared & written material available, to go into details of all the equipment contributions and DUT contributions to the overall uncertainty. Instead I will focus on one parameter only, the one that makes the largest contribution to amateur, noise figure optimised, measuring uncertainty.

Let us take two devices to be tested out of the collection box........

The top pre amp (DUT) uses a piece of LDF550 as a tuned input and I will call this DUT 1.

this is most often referred to as a cavity input design

The lower pre amp is a broad banded input design and I will call this DUT 2.

It is very clear which of these devices will offer the best bandwith response, the cavity input will. The broadband input design does have a high pass filter at ouput For the moment I am not examining these parameters as their influence, in this case, on the noise figure measurement/assesment is minimal..

Before I start to analyse & measure some parameters of these designs let me be clear. I am not going to be trying to prove or say that one is 'better' or 'worse' than the other overall. I am not going to be proving that the measurement of one is bad and the measurement of the other is good. My sole objective is to quantify the uncertainty contained in the two measurments, so that this can be taken into account in any judgemental decision taken.

I am going to put them both, in turn, on a Network Analyser and look at the match they provide to the test equipment (Even this will change when we move the DUT from the Network Analyser to the Noise Analyser, but we have to start somewhere), why we might wish to investigate this parameter will become clear later.

You can express such a match in several different ways, 'amateurs' tend to be more familiar with "SWR" expressed as a ratio like "1.5 to 1", and professionals tend to be more comfortable with "Return Loss" expressed in dB like " -10dB".

You can express the port (entry or exit) of the DUT in two ways also, either referenced as the familiar "Input" or "Output", or using the more flexible "S" terminology.

In this case Port 1 will be the input to the preamp and what we measure will be a reflection from that port, so this is know as S11 and will be expressed in dBs, generally negative, but not always in the case of noise optimised pre amplifiers. In the following displays the gain of the DUT is expressed as a measurement of S21 (port 1=input, port 2=ouput)

It is not my intention here to go deeply into S parameters or the measuring of them, but I do need to explain at least this particular parameter of S11, or "goodness of device under test input match", or, if you prefer, "the input SWR of the pre amp."

Here is the analyser screen for DUT number 1. (sorry for quality it had to be snapped off screen with my camera)

The lower plot is parameter S11(input match), the higher plot is parameter S21 (gain)