Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711

E-mail: Jack.Smith@cliftonlaboratories.com
 

 

To search within the Clifton Laboratories site, enter your search term below.
 

 


Home
Up
Updates
Current Products
Prior Products - no longer available
Documents
Book
Software Updates
Softrock Lite 6.2
Adventures in Electronics and Radio

 

How to Measure Receiver Sensitivity

I recently explained how I go about measuring receiver sensitivity and it occurred to me that others might find the description useful.

In the case at hand, the receiver to be measured is an Elecraft K2 transceiver, although that is not an important part of the technique, as a similar approach is usable with any CW/SSB receiver.

One note first, however. The K2 is a transceiver, which means that it can transmit back into your (probably) expensive signal generator. A few signal generators have protective circuitry that will save you the expense of repair, but most do not. So, before you connect your transceiver to your signal generator, disconnect its microphone and key to avoid accidental transmission.
 

What we will Measure

We will make a very simple measurements--how much signal is required for a 20 dB signal-to-noise ratio at the headphone output. More precisely, we will measure the required signal for 20 dB (signal+noise)/noise ratio, but with a target of 20 dB, the difference between S/N and (S+N)/N isn't too great and we will call the result the 20 dB S/N sensitivity.
 

Required Equipment

The required equipment is:

  1. A signal generator with a calibrated, variable output covering the frequency range of interest. I used a Boonton 102D signal generator because it happened to be sitting within two feet of my K2 transceiver and my HP8640B was on the other side of the basement.
  2. A multimeter (analog or digital) capable of reading audio output of your receiver. I used a Fluke 189 digital voltmeter with dB scale, but in fact a good analog meter is preferable as you have to do some "eyeball averaging" when reading the noise level.
  3. Connecting cables.
Test Setup

The illustration below shows the test setup. The headphones let you listen to the receiver audio at the same time you measure the level.

Before you connect your K2 transceiver to the signal generator, please re-read the note at the top of the page about first rendering your transceiver unable to transmit. You will be irked, to say the least, if the attenuators in your signal generator are smoked by a brief burst of power out of your transceiver.

 
Test Methodology  
  1. Disable the automatic gain control in the receiver to be tested. In the K2, this is accomplished by simultaneously pressing the [PRE/ATT] and [AGC] switch buttons. If the AGC is turned off, the K2 main display will report "OFF" and the decimal point between the mode letter and the last displayed frequency digit will blink on/off.
  2. If your signal generator is not synthesized or phase locked (like the Boonton 102D or an HP8640B), let it warm up thoroughly.
  3. Make the connections in the test setup shown above.
  4. For each band you wish to measure (and each mode), repeat the following steps:
  • Set the K2 to the desired mode, preamplifier = OFF. Set the RF gain control to maximum.
  • Set the signal generator for a test frequency in the band and set the K2 to the desired mode. The signal generator should be in CW mode, i.e., no modulation. Assume we start in the 40 meter band. I set my signal generator for 7010.00 KHz.  The signal generator output level should be at approximately -100 dBm. (2 uV if your signal generator is calibrated in microvolts) Tune the K2 to the signal generator frequency and verify that you can hear it.
  • Tune the K2 away from the signal generator 10 or 15 KHz. You should not hear the signal generator in your receiver; to test this unplug the coax cable connecting the generator to your receiver. You should see no significant change in the voltmeter and you should hear no significant difference in audio output. (connect the cable back, of course.)
  • Keeping the K2's RF gain at maximum adjust the K2's AF gain to provide a low but comfortable listening level. Record the voltmeter reading. In my test, this level was -60.0 dB on a Fluke 189. If your voltmeter reads only in volts, record the voltage reading.
  • Tune the K2 to the generator frequency, adjusting the K2's frequency so that the audio output as read on the voltmeter is maximized. Do not adjust the K2's audio level!
  • If your voltmeter reads in dB, vary the signal generator's output level (should need to be reduced if you started at the recommended -100 dBm level) until the audio level is 20 dB above than the noise reading. In my case, the noise-only reading in my case was -60 dBm, so I adjusted the signal generator to yield a reading of -40 dBm. If your voltmeter reads only in voltage units, then recall that 20 dB change represents 10 X change in voltage. If the noise reading was 0.2 volts, then the +20 dB reading with signal will be 2.0 volts.
  • Record the signal generator output level.
  • Set the K2's preamplifier to ON and repeat the steps above, starting with tuning off frequency and adjusting the gain for a comfortable listening level.
  1. As a bonus measurement, you might wish to switch to CW mode, 200 Hz bandwidth and tune to the signal generator. Keep reducing the generator's output until you think that the signal represents the weakest CW signal you could copy. Record the signal generator level. This can be considered the "minimum discernable signal." Since there is some difference between the signal level that you consider readable and my assessment, there is more room for operator-to-operator variance here. (We could make this more quantifiable by saying that the MDS is 6 dB (S+N)/N but in truth, it is whatever your ears tell you it is.)
My results for LSB mode are:
 
  Signal Generator Output in dBm
Frequency Pre-Amp OFF Pre-Amp ON
7.0 MHz -106 -107.2
24.9 MHz -105 -108
28.0 MHz -102.3 -103.5

MDS: 28 MHz
     Pre-amp off: -125 dBm
     Pre-amp on: -133 dBm

I've recorded the values to the nearest 0.1 dB, but I would be surprised if the measurements are accurate to 1 dB, as I have not used other equipment to calibrate the signal generator output and because there is considerable room for error in estimating the average noise level as the meter readings bounce around.
 

What do these numbers mean?

First, my measured MDS levels are reasonably close to Elecraft's published specification of -130 and -135 dBm, for pre-amp off and on, respectively. I probably could have reduced the generator output a bit more with the preamp off and still had a copyable result.

Next, we note that although the preamplifier has 10 dB gain, it only changes the 20 dB sensitivity figure by at most 3 dB. Why is that? The reason is that the K2's 20 dB SNR sensitivity is determined by the receiver's noise figure. In simplistic terms, when the preamplifier is turned on, it increases the noise and signal both, as well as adding its own internal noise to the mix. Now, if we look at the results and work through the math behind the K2's front end design, we see that when the preamplifier is enabled, the composite noise figure improves several dB. This is because the composite noise figure of a multi-stage receiver is determined mostly by the first stage or two. With the preamplifier off, the SNR is determined chiefly by the losses in the K2's input filtering and the TUF-1 mixer loss and then the noise figure of the 2N5109 post-mixer amplifier. When the pre-amp is engaged, the noise figure is reduced to a value closer to the sum of the K2's input filtering and the pre-amp noise figure. (With only 10 dB pre-amp gain, later stages will still have some effect on the overall noise figure.)

With respect to the 20 dB S/N sensitivity, the measured parameters seem consistent and within range of a normal SSB receiver.

 
Other Results

Stan, W5EWA, has provided measurements he made on his K2 following the procedure described at this page. Stan's data is with the preamplifier off. Except for 28 MHz, our readings agree reasonably well. The 6 dB difference at 28 MHz is more than can be attributed to calibration error.
 

Band K8ZOA W5EWA      
3.5   -107.9      
5   -108.9      
7 -106.0 -107.0      
10   -110.1      
14   -106.2      
21   -108.4      
24 -105.0 -107.0      
28 -102.3 -108.4