Hi Wolf,

Well, it's not really avoiding the audio phase shift -- the 90 deg. shift
is still being done by your software, and the 180 deg. shift is being
done by taking the mux outputs differentially.

If you could provide in software, two controls, allowing the amplitude
and phase of the right channel to be adjusted slightly (relative to the left),
then it should be possible to suppress the unwanted sideband nearly completely.
That should work with any of the designs being discussed.  However, you'd
probably need to readjust then when making a big QSY, and you would need a
measurement tool (receiver) to guide the adjustment.

73,

Stewart KK7KA

To: <[email protected]>
Sent: Friday, January 14, 2005 12:09 PM
Subject: Re: Re: LF: Simple image-cancelling exciter for digimodes


Hi Stewart,

Thanks for the suggestion - sounds too good to be true !
(to build an image-cancelling "TX"-mixer without the need for a phase-split 
audio
if I understand you correctly).

I must admit I blindly followed the ARRL handbook which seemed to imply"you 
cannot
build an image-cancelling mixer without some phase shiftingon the low-frequency 
side"
(unless you use the "weaver" method withpilot tones).

I have some old CMOS 4052 somewhere (not 74HC...) so may try these.
When using std CMOS in my "conventional" prototype, thesideband-suppression was 
worse
than with the 74HCxxx because they weretoo slow, and the L.O. outputs were not 
symmetrically enough.


Will let you know about any progress with this.
At the moment  I'm quite happy with my old design which still waits to be 
connected to an
amplifier for an on-air test (but thatwon't be before next weekend).


Best regards,
Wolf DL4YHF .




--- original Nachricht Ende ----


Hi,

You may wish to consider a dual 4-input analog mux such as
the '4052.  For instance, see
http://www.semiconductors.philips.com/pip/74HC4052.html

You can connect the quadrature LO outputs directly
to the S1 and S0 inputs.  Then, because there are two
muxes, you can create a differential output without
needing to phase-split the incoming audio.

For example, assuming address sequence is 0-1-3-2-0-1-3-2
(S1 lags S0) connect:

1Y0=Left ch.
1Y1=Right ch.
1Y2=GND
1Y3=GND

2Y0=GND
2Y1=GND
2Y2=Right ch.
2Y3=Left ch.

Then, the voltage between 1Z and 2Z is the SSB output.

Perhaps, you can use a transformer-coupled output.
Put a cap across the primary to resonate at the
approximate output frequency, and connect it from 1Z to 2Z
through a suitable resistor.  No op-amps or precise
resistors needed, and it provides the first stage of
filtering.

Disclaimer: I have not tried this, and there may be
a stupid reason why it won't work.  Please don't
laugh too hard.

73,

Stewart KK7KA