[TenTec] New Sherwood Presentation - Powerpoint & Audio

[Dr. Gerald N. Johnson]

On Tue, 2008-03-18 at 13:17 -0400, Zivney, Terry L. wrote:
> The direct conversion radios, such as the Flex 5000, make trade offs, too.
>  
> For example, the Flex 5000, and all other direct conversion radios have
> a problem with the audio image.  With image reject mixers (which is
> what the (I-Q mixer technique basically is), rejection of this image 
> depends upon perfect balance between the channels. Critical (i.e.,
> knowledgeable)  reviewers of these type of radios have noted that
> these images may be as little as 40-50 dB down, and in well-performing
> radios may be as much as 70 dB down from the desired weak signal. 

That has always been the nemesis of a phasing radio, transmitter or
receiver. The amplitude match has to be closer than the amount you are
willing to have the image rejection and the phase angle match also has
to be good. 90 degrees +/- 1 degree (at all audio frequencies and the LO
frequency) won't do better than 30 dB image rejection. Both these are
frequency sensitive which complicates their adjustments.

However since the "audio" phasing is done in software, and the sum and
difference is done in software it is possible to correct for RF front
end errors in software, though it probably runs into a Collins patent. I
know Collins has had a SDR on the market for several years that covers
DC to microwave and has an algorithm built in to correct for the LO and
mixer level and phase errors.
>  
> These images are much more of a problem than third-order intermod.
> Third-order intermod requires two interfering signals of great strength
> being in the proper (or unlucky, depending upon your viewpoint) frequency
> in relation to the desired weak signal.  The direct conversion image,
> on the other hand, requires only ONE strong signal to produce a
> problem.  Which do you think is more likely to happen in a crowded
> band, say 40m cw:  TWO signals EACH spaced at a specific 
> frequency and 90dB above your desired weak signal - or - ONE
> signal only 40, 50, or 70 dB (see above), spaced at the critical 
> frequency, often only 10 or 20 kHz away from the desired signal?
> My experience tells me that the second (Flex) is more likely to
> be a problem.There are many signals 40 to 70dB above the noise
> floor of my radio on 40m but there are relatively few precisely spaced 
> PAIRS of signals 90 dB above my noise floor, even with a full-size 
> 3 element yagi.
>  
> Roofing filters exist in part to reduce images.  The roofing filter 
> approach requires a second conversion in the radio.  In principle, 
> this shouldn't be any more difficult than providing  good first 
> converter, but years of experience and observation show that 
> is not child's play.

Neither is "child's play." The roofing filter limits some of the
receiver versatility, such as showing band scans or 100 KHz of waterfall
while receiving a single signal. With a GOOD roofing filter, I fail to
see the need for any more IF frequencies such as converting in analogue
down to a 14 KHz IF except to make use of a slow A/D. Today the fast A/D
exists to need no analogue conversion other than through the roofing
filter or (though expensive) direct A/D after the RF amplifier and a
band limiting filter.  Doing a single conversion with a sampling A/D
after the roofing filter (which might have a fairly wide bandwidth if
the A/D has enough dynamic range) should work as well or better than the
multiple conversion as used in TenTecs to this point. Mixers have always
been the limiting stages for receiver dynamic range ever since Armstrong
invented the superhet receiver, having high noise, low gain, and low
clipping points so any scheme that uses fewer mixers or protects those
mixers from extraneous signals (so direct conversion or a narrow roofing
filter) makes great strides in improved dynamic range.

The direct conversion receiver with super quality audio (100 KHz or so)
A/D with lots of bits (18 or 20 or more) can achieve good dynamic range
without needing a roofing filter and if the DSP code can include an
input for LO frequency and phase/amplitude errors of the LO system and
mixers, the DSP can achieve adequate or better image rejection.
>  
> Terry Zivney, N4TZ/9
> 
73, Jerry, K0CQ