[hpsdr] QSD architecture and image suppression

Bob McGwier rwmcgwier at gmail.com
Thu Feb 21 08:24:59 PST 2008


Oleg Skydan wrote:
> ***** High Performance Software Defined Radio Discussion List *****
>
> Hi, Ahti,
>
>   
>> Now with the new PowerSDR versions and especially with the new
>> Flex-5000 hardware(firmware) the situation has changed. 
>>     
> It is not a big problem to implement "multipoint" calibration to
> compensate for the phase/amplitude I/Q channel variations with
> frequency. But I think that tempereture variations will have some
> influence to the calibration. So even if we will be able to achieve
> 90-100dB suppression after the calibration it probably will not
> survive several hours of the contest for instance, when the 
> temperature inside the transceiver will grow substantially...
>
> I see only one solution for the QSD/QSE architecture - the 
> calibration hardware/software should allow us to do the calibration 
> during the operation. In this case we will be able to compensate 
> for component variations due to the different factors (temperature,
> humidity, time and etc.)  continuously. 
>   

That is available in the Flex 5000 hardware and it will be available in 
the Flex 5000 software.  The Flex 5000 design was clearly to have a self 
calibrating radio.  It has all sorts of built in test equipment, 
including an all band signal generator for first reducing the image in 
the receiver to the noise floor and then using that receiver, with a tap 
at the output, nulling carrier and image on the TX.  It works.  The 
problem with the SDR-1000 is several fold.  It does not have built in 
test equipment.  So external signals will have to be used.  But there is 
so much temperature drift of the oscillator driving the DDS that the 
phase relationships and amplitude relationships drift sufficiently in 
time to reduce the null by at least 10 dB fairly quickly.   Continuous 
operation is possible but it will be necessary to use off the air 
signals almost continuously to tune the cancellation.  On a quiet band,  
this will be problematic.
> Oleg
> 73 de UR3IQO

73's
Bob


 1203611099.0


More information about the Hpsdr mailing list