[Top] [All Lists]


To: <>
Subject: [AMPS] IMD
From: (Peter Chadwick)
Date: Wed, 19 Sep 2001 07:47:35 +0100
Tom says:

>Wide band noise does not test power supply dynamics.

That means that your wideband noise doesn't have a low frequency component, so
isn't the required 'noise'. An appreciable LF component is required.

The illusion on 40dB may be there that it is 'in channel' requirement for
audible distortion, but I didn't have that understanding from what was said. I
was thinking more on the lines that 40dB should be a design aim on 3rd order
products, with the assumption that higher order products fall off pro rata.
Unfortunately, this assumption is frequently invalid.

The concept of a 'measurement receiver' to measure what the digital cellular
people call 'ACPR' (Adjacent Channel Protection Ratio) as Rich suggests, has
advantages. It was certainly used a lot over here for measurements on VHF/UHF
Mobile radio equipment, and the concept is fairly well enshrined in ITU
Recommendations as the Out of Band Emissions. You have to watch the phase noise
though: assume the receiver has a 4kHz bandwidth as Rich suggests, and is tuned
such that the 4kHz window it represents starts 4kHz away from the 'carrier' of
the SSB signal, and on the other sideband. (e.g. signal 14200 USB, measurement
rx 14192 to 14196). Then for the phase noise to be insignificant (i.e. less than
1dB effect on measurement), when integrated over that 4kHz band it needs to be
46dBc/Hz  below the splatter level you are looking for.

This business of the wide band IMD products is not generally considered a worry
by many operators, since it doesn't affect their QSO! It certainly shows up with
strong signals and low noise floors, especially on 2m. Wide band noise is
ignored too - I did the sums some years back on a FT480 feeding a 9 ele Yagi,
and a guy 25 miles away with a 9 ele Yagi (both antennas pointing at each other)
and a 3dB rx noise factor would get a 4dB rise in rx noise floor 25kHz away
every time the FT480 operator hit the transmit switch. Admittedly, the FT480 was
exceptionally bad in this respect, but used to drive a linear and it would have
been horrendous - mind you, the IMD products were bad, too.

IMD measurements with spectrum analysers and sound cards need treating with some
care to avoid aliasing problems. Additionally, besides needing 1 bit of
resolution for 6dB of range, the IMD of the ADC and any signal processing prior
to it must be considered. Just because an ADC has an n bit range doesn't mean
that IMD products are necessarily n bits down! There can also be the analogue
bandwidth to consider -  been there done that, wasted a lot of time on it a few
years ago.

And Rich. I don't know about 75m in California, but over here, there's plenty of
incoherence on 80m SSB without any ethanol at all!


Peter G3RZP

FAQ on WWW:     
Administrative requests:

<Prev in Thread] Current Thread [Next in Thread>