[Elecraft] KX3 AM-breakthrough

[Rich Heineck]

Hello Klaus,

On 07/30/2013 02:20 AM, Klaus Dittrich wrote:
> ...
> I wonder why this isolation amplifier is off by default,
> but assume the reason is to save battery power which
> does not matter in case of use as home station.
>
> So I would like to know if there are any degradations
> in IP3 or more probable in opposite sideband suppression
> if this amplifier is permanent in the signal path because
> of additional phase shifts.
> As far as I remember this case is not included
> in the OSB adjustment procedure.

It consumes about 10 to 15 mA, and when used at the 0 dB gain level 
(strictly for isolation) it degrades MDS slightly.  If used at the 10 dB 
gain setting (10 dB preamp), it will improve MDS slightly. This circuit 
has a very high IP3 so it's affect on IMD3 dynamic range is mostly due 
to the change in MDS and signal gain to downstream circuits.  This stage 
is before the signal is split into quadrature components, so it doesn't 
directly affect opposite sideband rejection.


> I was astonished to see during OSB adjustment how
> sensitve the procedure is to frequeny dependend phase
> shifts which I never thought they could be of importance.
>
> Every 10th mm on the circuit board and every ns in software
> seems to matter.
>
> Wayne, could you please drop a few words about this?

Getting high levels of OSB rejection requires very precise phase and 
gain matching between the quadrature signal paths.  40 dB requires gain 
matching to be about 0.1 dB and phase matching to be about 1 degree.  
With precision components, 40 to 50 dB is about the practical limit of 
what can be done in hardware.  The rest is done in DSP.  Every 20 dB 
improvement requires about a 10x decrease in both phase and gain 
mismatch, so 80 dB of OSB rejection requires the phase balance to be 
about 0.01 degrees and the gain balance to be about 0.001 dB.   It's 
possible to get a deep null during OSB calibration, but it takes very 
little change in the balance to move away from that point.

73,
Rich AC7MA