[Premium-Rx] Noise Power Ratio Testing of Radio Receivers

[Adam Farson]

Dear Michael,

Many thanks for your comments. I have read the Sosin article in question.
Given that Dingley was experimenting with NPR testing at Racal, I wonder
whether Sosin's "failure factor" included a number derived from a
notched-noise test.

The ARRL Lab uses a spectrum analyser with 3 Hz RBW to measure the IMD3
product amplitude. This method subtracts out the reciprocal-mixing
contribution, and is endorsed by ITU-R Recommendation SM.1837. A baseband
spectrum-analysis program running in a PC can be used for this measurement.

I certainly agree that the incident power density involved in a classical
2-signal test will generally be well below the threshold for excitation of
passive IMD in filters or of IMD in switching (steering) diodes. This is
where the NPR test comes into its own, especially at high noise loading
levels.

73, Adam VA7OJ/AB4OJ

----- Original Message ----- 
From: "Michael O'Beirne" <michaelob666 at ntlworld.com>
To: "PREMIUM-RX" <premium-rx at mailman.qth.net>
Sent: Monday, December 03, 2012 11:05 PM
Subject: Re: [Premium-Rx] Noise Power Ratio Testing of Radio Receivers

>
> Dear Adam
>
> Many thanks for all that.
>
> Perhaps it is also time to re-visit the famous embargoed article by Sosin,

> Marconi's Chief Scientist in the Marconi house mag, Point-to-Point 
> Communication in or about 1972 in which he tried to devise a method of 
> testing hi-grade receivers using an all-embracing formula (which frankly I

> never understood) that took in IP3, front end preselection, reciprocal 
> mixing effects and much more to try to establish a "failure factor" - ie 
> the factor by which the receiver fell short of the perfect receiver.
>
> As I recall, mathematically he appplied a barrage of frequencies and input

> levels to the receiver to simulate the live mass of signals that appear on

> a profesional aerial and derived how the receiver would react to such an 
> environment rather than the  somewhat unreal standard two-signal IP2 and 
> IP3 test.  His analysis demonstrated the continued value of traditional LC

> preselectors, albeit that in his time, and still now, tracking 
> preselectors werw and are a very costly way of doing things.
>
> In a related sort of way there is also a debate as to what input levels 
> should be applied to an IP3 test to establish a valid test.
>
> I personally distrust searching for an IP3 product at the miniscule MDS 
> (3dB s+n/n).  At best one should be measuring this with a true RMS 
> voltmeter, but how many of us have one?  In contrast, going for a 10dB 
> s+n/n ratio should be more accurate since most average reading voltmeters 
> will be ok.
>
> But even then, the levels are unlikely to be "teasing" the switching 
> diodes in the half-octave filters and elsewhere.  And the levels can be 
> high.  Even on my small Wellbrook Loop feeding a spectrum analyser, levels

> well in excess of 10mV are to be expected from the usual broadcasters and 
> they are miles and miles away from me. I dread to think what would happen 
> if I had a kW down the road.
>
> 73s
> Michael
> G8MOB