[R-390] Receiver Sensitivity Measurements

[Roy Morgan]

R-390 folks,

The below rant was instigated by a question from Randy Stout  I hope it is 
at least interesting to folks on the list:

Randy asked:
>  Could you please give me a simple explanation of impedance matching for 
> connecting a 50 ohm signal generator (8640B) to the balanced connector on 
> the R-390A.

I can give you an outline of the factors involved.
I've been wanting to write a rant on this topic for some time, so here goes:

Signal Generator Termination:

Most modern signal generators will deliver the indicated output voltage 
only if the output, including any connecting cable, is terminated with 50 
ohms.  At low frequencies, this can be done at the output connector but at 
higher frequencies it must be at the end of a 50 ohm cable.  "Low 
frequencies" means where the cable is, say, less than 5 percent of a 
wavelength.  At 30 mc, the  effect of SWR on the cable may change the 
voltage at the output end if the cable is of some length and is mis-matched.

Recommendations:

  1) Carefully read the manual for your generator to understand what 
conditions are needed for correct output voltage indications.

  2) Provide the proper cable (low leakage coax) of the correct impedance 
(50 ohms) with correct connectors.

  3) Use a flat load or attenuator of appropriate impedance at the 
termination end of the cable.

Notes:

  1) The GR 1001A has fairly unique output termination requirements. The 
source impedance is not the same for all attenuator settings.  Read the 
manual if you are using one.  As far as I know, the HP 606 and later 
generators all require a 50 ohm termination to indicate output voltage 
correctly.

  2) The URM-25's all need 50 ohm terminations also.

Signal Generator Leakage
All signal generators leak to some degree. This means that when you set the 
output attenuator and level adjustments for very low output levels, say 10 
microvolts or less, signals that leak out of the generator by paths other 
than the output connector can be significant sources of error.  One fellow 
reported that he measured his R-390A receiver sensitivity at about 5 
microvolts.  Trouble is that the Heath signal generator he was using was 
across the room and not connected to the receiver at all!

Recommendations:

1) Do not use Heath, Eico, or similar signal generators and expect to get 
meaningful results at very low signal levels under any circumstances.

2) The GR signal generators and any by HP or other professional grade 
makers will likely do well if properly used.

3) TEST for generator leakage. Put a modest antenna on the radio, a good 
load on the generator at the output connector, set the generator for one 
microvolt, and see what you can hear.  A more severe test is to use a 
little loop or hank of wire at the end of a coax as a moveable antenna and 
"sniff'" around your signal generator with the receiver RF gain wide 
open.  Then put the load at the end of the connecting cable you intend to 
use and sniff some more.  Running a terminated cable at a generator output 
of a volt or so will let you estimate the shielding performance of the 
cable.  You may need to clean and reseat grounding and shielding features 
of your generator to reduce any leakage you find.

4) Use a matched attenuator at the end of very good coax for testing.  This 
minimizes the exposure to any existing leakage at the input to the 
receiver.  One path for signal leakage is: radiated signal from the 
generator enters a poorly shielded connecting cable and adds to the signal 
from the attenuator in the generator.  For the R-390A, this might mean 
building an attenuator in a metal box that has a twinax connector on 
it.  Do not trust any coax supplied by a popular company that used to 
supply electronic parts and pretty much only sells tv's and cell phones now.

Load Presented by the Device Under Test

The load that the device under test (the rf input terminal of your R-390A) 
puts on the generator load or attenuator matters a lot if you want to know 
the actual input voltage.  If the receiver is actually 50 ohms (very 
unlikely in the R-390A) then it can be the proper termination for the 
generator cable.  Since it is both likely NOT 50 ohms, and you are unlikely 
to know what it really is, other methods must be used if you want to be 
reasonably assured of the actual voltage at the input terminals.

Recommendations:

  1) Understand the input impedance of the device under test, at least to 
the extent that you don't assume what it is.

  2) Use methods to reduce the effects of variable and unknown input 
impedance.  Specifically, use a low output impedance voltage divider at the 
input to the receiver.  A little shielded box, two resistors and two 
connectors will do the job nicely (at HF frequencies.)

>  I want to do reliable and comparable SN ratio measurements, so I can 
> track how my receiver behaves following alignment and other tweaks, such 
> as tube replacement, IF gain adjustment, etc.

Reliable, can mean a couple of things:

1) If reliable means that you can to be pretty sure of the actual receiver 
input voltage, you need to use a low output impedance voltage divider at 
the receiver input, and test for generator leakage.

2) Reliable can mean repeatable. To get repeatability, use good equipment, 
careful methods, and keep good notes (and never transmit into the output of 
your signal generator with a transceiver!)

Comparability can mean a couple of things also.

1) It can mean consistent results from time to time.  If you are careful in 
your setup and methods, you should be assured of consistency of results 
over time at your place.

2) It can mean assurance that you can compare measurements made with your 
set up with those made with someone else's. In order to compare your 
results with those of other people, you may want to use pretty much the 
same methods they use.  In the case of the R-390A, using military manual 
procedures, this means grounding one side of the balanced input, using a 
50-ohm load at the receiver end of the cable connecting the generator to 
the receiver, and accepting the errors that the variable input impedance 
and mismatch are introducing.

Tracking changes in receiver performance due to alignment, tweaks and tube 
changes can be as simple as just using the same setup for RF signal input 
every time.  This means carefully using the same equipment in the same way, 
based on notes and records.  You may choose to use the manual procedure, or 
its equivalent with the gear you have, or you may choose to make a more 
careful setup to be more assured of the actual receiver input voltage.

Recommendation: choose one of the below:

1) Use the manual procedures. If you don't have a URM-25 in good shape, use 
any more modern signal generator with an understanding of its termination 
requirements.

2) Use a reliable generator but use an attenuator with a low ouput 
impedance to establish a better-known receiver input voltage.

3) Do both, and compare results.

>I also would like to generate numbers that can be compared to other 
>peoples, using best principles of measurement.

Unfortunately most folks who have R-390A's don't use the best principles of 
measurement, depending on what we mean by "best".  They use the methods 
specified in the manual(s).  Those methods are just fine for military 
situations with the goals of performance checking to determine suitability 
for service, and are very good if we want to compare numbers from two or 
more different people/places.  They leave something to be desired if what 
you want to do is accurately measure such things as receiver sensitivity.

Recommendation:

1) Carry out the methods in the military manuals as closely as you 
can.  See if leakage is contributing to your sensitivity measurements.

2) If you want to compare your measurements with those of other folks, find 
out as exactly as you can what setup they used. Duplicate their methods as 
closely as possible, including any sources of error.

Miscellaneous observations:

  1) Measuring rf voltage accurately is far more difficult than most folks 
realize. An accuracy of 5 percent is achievable with good equipment and 
very careful procedures. Accuracies of 3 percent or so are questionable 
under almost any circumstances. Accuracies of one percent are almost 
impossible anywhere.  Careful reading of the specs of modern signal 
generators will give you an idea of the voltage accuracy they 
offer.  Accuracy of rf voltage measurements is not normally an issue in 
receiver sensitivity.  The difference between 1 microvolt sensitivity and 
1.02 microvolts is seldom important.  Whether we have one micovolt at the 
antenna terminal or one tenth of a microvolt is much more important, and in 
my opinion deserves looking into.  Most signal generators depend on the 
following three assumptions (at least) for proper indication of output 
voltage level: a) the rf voltage is reliably measured or set at some high 
level, such as 2 volts b) the attenuator is working as it is supposed 
to:  c) the attenuator and connecting cable are properly terminated.  And 
at low output levels, the assumption is that leakage is minimized.

2) Measuring RF voltages directly at very low levels is nearly impossible. 
Calculating what we expect them to be is much easier, and is about all we 
can do.

3) Noise and interference from the environment may affect sensitivity 
measurements.  Almost none of us have access to an RF shielded room to make 
measurements in.

4) There have been reports that re-soldering the RF path connections in an 
R-390A with silver bearing solder improved the sensitivity.  I for one 
would like to hear of carefully documented experience with this.

5) If I remember correctly the R-390A receiver sensitivity measurements are 
made with the RF gain at maximum (and AVC off.)  Thus, the setting of the 
IF gain adjustment may make a difference.  Presumably, receiver sensitivity 
measurements should be made after the IF gain is properly set.  If the IF 
gain has been cranked up to the point where the IF sections are 
contributing significant noise, measurements of receiver sensitivity are 
questionable.

6) In receivers such as the R-390A, it is the first, and possibly the 
second, RF amplifier that completely determines the sensitivity of a 
receiver if it is set up according to normal design intentions.  That is, 
the IF section and mixers contribute very little to the noise against which 
the RF input is being compared.

7) One way to detect leakage in a setup is to move your hand or a metal 
sheet, possibly grounded, around the generator and receiver. Also, move the 
connecting cable(s).  Any variations in measured receiver output indicate 
that RF is getting into the receiver by some path other than through the 
path from generator attenuator output to receiver input.  (This method is 
especially useful in UHF and microwave situations.)

8) The input impedance of the R-390A is specified to fall within a 
range.  Mil-R-13947B (which specifies the R-390 "non-A") states:

"3.13.3 Antenna input impedance.- The rated input impedance for the 
balanced input circuit shall be 125 ohms. In the range from 500 kc to 16 
mc, the measured input impedance shall not be less than 50 ohms nor greater 
than 375 ohms; for the range from 16 mc to 32 mc the measured input 
impedance shall be not less than 100 ohms nor greater than 700 ohms."

(I don't have corresponding specs for the R-390A/URR.)

9) With the URM-25, it's very easy to get leakage. Simply connect your 
plastic cased counter to the "RF OUTPUT X200K" connector.  With any 
generator made by Heath, Eico or of similar design, you get lots of leakage 
just the way the thing is.

10) The use of a low-output-impedance voltage divider at the receiver 
reduces the impact of odd receiver input impedances.  Even if the receiver 
input impedance varies from 50 ohms to 700 ohms, a 10:1 voltage divider 
with an output impedance of 5 ohms will deliver an output voltage very near 
one tenth of the input.  The 100:1 divider with an output impedance of 0.5 
ohms will do even better.

Possible conditions with the R-390A:
Lets take a look at some possible scenarios when measuring the receiver 
sensitivity of an R-390A.

A very bad case:
- The signal generator in use has some leakage and the coax  being used to 
connect the generator to the receiver is has poor shielding (or worse yet 
unshielded wires are used).  The connectors in use are not all that good 
and allow for the leakage signal to enter the cable. This combination 
contributes a third of a microvolt of RF at the receiver end of the cable.
- The cable is connected to the receiver with a UHF-to-twinax adapter, with 
no terminating load on the cable other than the input  impedance of the radio.
- The input impedance of the radio happens to be 200 ohms at the test 
frequency.
- The receiver is in quite good alignment and actually has a sensitivity of 
about six tenths of a microvolt actual rf signal voltage at the antenna input.

Here is what is going on in this situation.  The signal generator's output 
impedance is 50 ohms.  The generator is creating twice the indicated output 
voltage, with the assumption that the output is terminated in 50 ohms.  The 
actual load of 200 ohms forms the expected voltage divider to deliver four 
fifths (0.8) of the voltage at the termination, not one half (0.5),  thus 
an error of 30%.  The actual voltage is 30 percent higher than indicated.

The operator makes the measurement and arrives at a signal generator 
setting that produces 10 db signal plus noise to noise ratio.  What might 
the indication be?  About point one nine microvolt.  The leakage is 
contributing three tenths of a microvolt. The actual attenuator 
contribution to the signal is three tenths of a microvolt, but the 
generator indicates 0.19.

So the proud owner reports 0.19 microvolt sensitivity when it is actually 
0.6.  If the actual sensitivity was a little better (say 0.4 microvolt), 
and the generator leakage a little higher, the proud owner might report 
sensitivity way below that.

Another contribution to error (that may or may not have ever happened) is 
speaker impedance variation if the output voltage is measured across a 
speaker load.  Many speakers have a resonance somewhere in the low audio 
range, and it could be close enough to the 400 cycle or 1000 cycle 
modulation frequency to cause trouble.  Even without pronounced resonances, 
actual speaker impedance can vary over a range from one third to four times 
the nominal impedance.  Increased impedance at the test audio frequency, 
combined with the moderate output impedance of the receiver audio output 
would increase the measured audio voltage at that frequency. The noise is 
broadband and spread over the audio spectrum and the test modulation 
frequency is enhanced by the resonance of the speaker load.

A bad case in the other direction:
- The signal generator has very low leakage and good coax and connectors 
are in use.
- The actual receiver input impedance is 25 ohms at the test frequency
- The coax in use is not terminated at the receiver and SWR on the coax is 
such that the voltage at the receiver is slightly less than would be expected.
- The audio load is purely resistive with no resonance errors.
- The signal generator actually modulates 25 percent not 30 percent as 
expected.

In this case, the signal generator will indicate a voltage higher than the 
actual sensitivity of the receiver.  The measurement might be about 0.9 
microvolts or more for a receiver that actually has 0.6 microvolt 
sensitivity.  The voltage divider formed by the load and the generator's 
attenuator output impedance will deliver two thirds of the indicated 
voltage.  RF leakage and speaker resonance contribute little to the 
measurement.  The weak modulation level makes a higher rf level necessary 
to generate the needed audio signal

A case of good conditions:
- The generator has very low leakage.
- The coax in use is moderately short and has very good shielding.   All 
connectors are good and a minimum of adapters is used.
- A 10:1 or 100:1 voltage divider is in use at the receiver antenna jack 
and properly terminates the cable.
- A resistive audio load is in use.
- Indicated modulation level is accurate.

In this situation, a number of good things are happening.  The generator is 
properly terminated and well shielded coax is in use.  The voltage divider 
at the receiver both reduces leakage into the receiver input terminals and 
divides any leakage into the cable by 10 or 100.  We can have good 
confidence that the numbers arrived at are pretty good.

Here are my suggestions for things that others can carry out and report on:

1) Build and use a termination/divider of 10:1 or 100:1 ratio, either 
inside a twinax connector or in a small metal box with a well-installed 
twinax connector. (A 50 ohm resistor and a 0.5 ohm resistor will get you 
very close to 100:1.) Use high quality coax either well fastened to the box 
or with very good connectors on ether end (Type N perhaps) and appropriate 
connector on the box.

2) Build such a termination with a toroid to match the 50 ohm cable to the 
nominal 125 ohm balanced input of the receiver.

3) Assemble a setup with a leaky generator, unshielded connecting wires, or 
poor coax, and compare results with a well designed setup.

4) Compare results at 30 mc with a long mis-matched cable and properly 
matched cable.  Note that a quarter wavelength gives maximum impedance 
transformation and this length is about 8 or 9 feet at 28 mc. for foam coax.

5) Find a speaker with pronounced resonance at 400 or 1000 cycles and see 
if it causes different results compared to a resistive load. Use of a 600 
ohm to voice coil transformer tuned by a capacitor at it's input might 
simulate this situation with little trouble.

6) Investigate the RF Input coil balance adjustment circuit. The input 
circuit of the R-390A has a fixed and a variable capacitor to establish the 
balance.  Intentionally ground the fixed capacitor side and vary the 
capacitor adjustment to see if it affects the RF sensitivity measurement.

7) Intentionally make a set up that includes:
- Generator leakage in the range of the receiver actual sensitivity. 
(Possibly with a URM-25 with a radiating wire attached to the high level rf 
output connector.)
- A well aligned and sensitive receiver.
- Unmatched generator load.
- Leaky or unshielded connection from generator to receiver.
- SWR effects in the generator to receiver coax or wire.
- Somewhat high modulation level
- Speaker load resonance.

It might be possible to measure a receiver sensitivity below 0.01 microvolt!

Roy


- Roy Morgan, K1LKY since 1959 - Keep 'em Glowing!
7130 Panorama Drive, Derwood MD 20855
Home: 301-330-8828 Work: Voice: 301-975-3254,  Fax: 301-948-6213
roy.morgan@nist.gov --