[Elecraft] MDS practical measuring?

[Don Wilhelm]

Brian,

The traditional way to measure MDS requires the use of an audio 
voltmeter (True RMS voltmeter) a low level very well shielded oscillator 
with a known and calibrated output level and a well shielded step 
attenuator.  Any oscillator leakage will produce erroneous results.

Turn the receiver AGC off, and using the attenuator, reduce the signal 
generator output to a barely perceptible level.
Then turn the generator off and measure the receiver noise.
Now turn the generator on and adjust the attenuator to produce a 
voltmeter reading 3 dB above the level that the noise alone produced.  
The input level from the generator and attenuator combination is the MDS.

If you do not have a low level signal generator and step attenuator with 
sufficient shielding to prevent leakage, a good approximation can be 
obtained with a signal generator such as the XG3 (or XG1 or XG2) with an 
output of -107 dBm - the calculation is in the XG3, XG1 and XG2 manuals.

While not an exact MDS measurement, I routinely check the sensitivity of 
all receivers I work on.  My process uses Spectrogram to view the audio 
output.  I set my HP8640 to -130 dBm and measure how many dB on the 
Spectrogram screen that signal is above the receiver noise floor.
Caution:  My HP8640 has all internal shield covers installed and has no 
detectable leakage at -130 dBm.  Some of those generators have had the 
right rear shield cover removed and discarded to make servicing easier, 
and those generators will have significant leakage.  To check a signal 
generator (and other test setup gear) for such leakage, start with a 
higher level, then add attenuation in 10 dB steps.  If the audio (as 
observed by Spectrogram) does not drop 10 dB with each step, there is 
leakage from something.

73,
Don W3FPR

On 5/4/2014 8:34 AM, Brian Alsop wrote:
> It was time to drag out the test gear and give my older K3 a going over.
>
> One thing of interest was Minimum Detectible Signal.
>
> There is a WIKI definition for it but the equation doesn't help me a bit.
>
> I just thought I'd attach a calibrated signal generator and keep 
> reducing the level (for a 500 Hz bandwidth) until I couldn't hear it.
> That doesn't seems subjective since the signal generator is a constant 
> tone and not information to decode.
>
> Then I tried looking at Spectrogram and defining a minimum S/N ratio 
> which would define MDS.  The MDS values derived this way were much 
> lower for a 6 dB S/N ratio.  In fact, I couldn't hear the signal at 
> this S/N ratio!  The integration time constant used for averaging 
> clearly was helping with detection.  So what time constant would be 
> appropriate for normal CW?
>
> How indeed is MDS measured quantitively?
>
> 73 de Brian/K3KO