I know that you've all been waiting for some closure on my hair-splitting exercise. (I don't really think that...)

Maybe it's just me or maybe it is really hard to measure receiver sensitivity correctly. At every turn I was confronted with another reason why the numbers were wrong. I have also come to a better appreciation of how much psychology is involved and how it can lead you astray.

Here are a few of the things I learned:

1. Regardless of whether you build or buy, your test equipment will mislead you at every opportunity. I built a 40 db attenuator several years ago in a cast metal box using the best RF practices that I could muster. I used my Boonton RF millivoltmeter and my Boonton 102C signal generator to verify that it was doing what I expected. Then I proceeeded to use it for measuring my receivers and the results were in line with what I expected based on other peoples brags about the same receivers. But something has changed or I made a mistake back then because my 40 db attenuator is actually more like 35 db when examined carefully.

2. My Anritsu ML422C Selective voltmeter has a bridging mode where the input impedance is 10K ohms. This is for making measurements in circuits that are already terminated. The specs also say that the unit goes from some very low frequency to 30 Mhz. However, closer reading of the manual reveals that by the time you get to 2MHz, the input impedance has dropped to 1K and at 30Mhz it's more like 50 ohms. Using this to meaure signals at the antenna terminals is a fine idea but it distorts the results above a few MHz.

3. I have re-learned this too many times to count and I will learn it again some day. When two impedances are connected in series across an AC generator, the voltage divides proportionately across the two impedances. True enough but this does NOT mean that the AMPLITUDES divide proportionately! My voltmeters just measure amplitude! When the receiver input impedance is not resistive, you will measure higher voltages than you would expect if it were resistive.

4. Test equipment generates noise. My ML422C is great for measuring small RF voltages but it radiates junk from it's 80's microprocessor. It is best to disconnect it before trying to look at the receiver noise level, one step of every sensitivity measurement.

5. If your receiver has internal spurs. You will always choose the spur frequency to make a sensitivity measurement at.


6. 3 feet of unterminated coax is not always negligible at 7 Mhz, particularly when you are measuring microvolts and fractions of microvolts.

7. Every BNC connector will be solid until it is time to actually measure something.

So, in the end I THINK I have actually measured the sensitivity of my R-390a and I think I believe the results. To do it I ended up building a MOSFET probe right into the DA-121/U box. This eliminated the loading effects of the ML422C and the antenna peaking control would peak at the center throughout the process. I also learned that FET voltage followers have a gain that can be quite a bit less than 1. Mine came out at about .76. Using the modulation on/off method, I used to get sensitivity numbers around .5 microvolts. Using the FET probe and the Selective Voltmeter to measure the actual voltage at the antenna terminals, I get something closer to 1 microvolt.

I think that things like reactance, impedance variations, equipment errors, and a tendency to not question results that look good (my 40 db attenuator for example) lead to most sensitivity measurements by non-professionals (includes me) being substantially off. There is still room for doubt with all that I've done so far. For example, my SLM has not been calibrated against a traceable source. I am still using 2 feet of coax to connect from the FET buffer to the SLM with the SLM input Z varying from 10K to 60 ohms. I did some experiments to see how much this might be affecting the numbers and they seemed to indicate a small effect. But who knows.

I have thoughts of making these measurements by the book and in many ways that people use and comparing them to see how much variety you can get with the same receiver. I don't know if I will do it (lots of time involved) but it would be interesting.

Welcome aboard. Once you get the hang of it, R.F. measurements can be fun or frustrating, usually not at the same time.

One thing to take into account, 1 meg input Z on scopes, meters etc. is always shunted by some capacitance that lowers the input impedance as the frequency is increased.

Nearly all meaningful R.F. level measurements are made at some common impedance like 50, 75 95, 125, 300, 450 and 600 ohms.

Using equipment that matches the impedance's of the associated equipment goes far in reducing the parasitic or stray capacitance errors. I never worry about finding a 10X probe, with bandwidth to match my scope. I just use 10X probes for troubleshooting.

Every now and then, when I find a free moment, I try to solve the problem of the first part of this post. I would like to use the "receiver under test", by monitoring the avc voltage, to maintain a consistent level at the antenna terminals, by adjusting the signal generator output and noting the generator's level meter reading as a 50 ohm load is applied/removed, across the receiver's antenna terminals, to calculate the receiver's contribution to the known load. Whew, I hope that makes sense.

I am sure a Mathmagician (not me for sure), could solve this problem, then the receiver could be used to measure its own antenna impedance.

We won't give up.

The ARRL has several noise bridge circuits, one which I like, that could, with a sensitive detector, could measure the input Z of the receiver, then you could wind a torrid matching transformer to match generator to receiver. (with the right core).

Copper plumbing pipe, end caps etc. can be used to make shielded R.F. attenuators, etc.

Sorry I can't be more active help at this time, I am overhauling several 8mm movie projectors, in order to show some 50 year old family movies to younger family members.

Good luck,

Charlie

Thinking of your problem this morning, during caffeine consumption, I thought of another way, not too practical, is to take non-inductive resistors, one 49 ohm and one 1 ohm and make a voltage divider, in a "test fixture" at the end of the coax, right at the receiver's input. A third non-inductive resistor, selected experimentally, connected in series, between the 1 ohm output of the divider and the receiver, which causes a 6db drop in signal should equal the receiver's input impedance. (Add the 1 ohm resistance of the divider to this value).

Lead length, finding a 1 ohm non-inductive resistor, making the test fixture, are the obstacles, to be overcome

A math person could probably be able to calculate the receiver's input impedance, using a single "third resistor" instead of having to select one for a 6db signal drop.

If you trust the signal generator's output meter to be more linear than the receiver, adjust the output (look for +6db) for a constant signal, (monitor avc buss voltage) at the receiver.

If a simple, few component, standard attenuator, termination, all designed for r.f. use, way, using a calculator, can be devised to determine a device's input impedance, than I think we have accomplished something.

I am waiting to see a post by someone saying: "hey this has already been done years ago, here's how".

Charlie

Oh how true and it generally ends in a fist fight over what that silver tarnish really is Hint: Its rarely conductive.
It is also a good excuse to convert to TNC as they are interchangable....just watch out for those 75 Ohm versions.

Having just about grown up in a screen room enviroment after leaving the active duty Navy in 1963 I learned early on what is required to convince the government inspector that he is signing off on what the taxpayer purchased and not some inflated number such as vehicle horsepower. My baptism was the AN/WRR-2A which a few here may be familiar with.

With quality equipment it is easy to get a leakage free system without impedance bumps; the problem is to maintain it that way by ignoring regular tests and PM.

Carl

Hewlett-Packard made some spectrum analyzers with special 75 ohm "BNCs". These connectors had different dimensions than standard BNCs and mistakenly interchanging cables with standard BNC cables could damage the BNC on the spectrum analyzer.

If you have a surplus h/p spec. analyzer, you may want to watch for that.

Charlie

HP had/has 75 Ohm options on lots of test equipment and using 75 Ohm N and BNC connectors. The CATV, satellite, and video industry are big on 75 Ohms.
Be sure to read the Option numbers if not sure as the 75 Ohm stuff is often much cheaper.

Carl

I've been through this

Resistive loading will correct the impedance, but it may also load the
RF tuned input and reduce the image rejection measurements.

A matching transformer preserves the dBm readings, but the voltage
at the antenna terminals will no longer correspond to the signal
generator reading.

The real kicker is that few receivers will exhibit an input impedance
that is close to what is specified. That begs the question as to how
the manufacturer the sensitivity specs, and whether they are
accurate to begin with.

Pete

Correct or not, when I checked the sensitivity of my SP-600, I used a matching transformer and read the power on the signal generator's output level meter and converted the power level into volts, using the receiver's (supposed) input impedance.

To use a R.F. bridge that I built, to measure the SP 600's input impedance, I would have to use (lug into the basement) another 70 lb receiver, as a null detector. Next time, maybe.

I have had a very busy week, but I have not forgotten this problem. It seems that there should be a math solution, using the "receiver-under-test" own AVC buss to maintain a constant level at the antenna input, while switching a known termination on and off of the antenna terminals, while reading the different signal levels, from the signal generator's meter.

Charlie