[hpsdr] Image rejection

[Bill Tracey]

Surprised this did not get answered on the Flex reflector ... though it 
seems to be having some mail delivery issues of late so perhaps your query 
did not make it out to all the subscribers.

I think it's a known problem that the IQ correction factors vary by 
band.   There's been talk of having multiple calibration constants in the 
software so it could use the one most appropriate, or interpolate but it's 
never been implemented, probably because there's a better way to solve the 
problem.  The better way to solve the problem is to use the impulse 
generator on the RFE board of the SDR 1000 to measure the impulse response 
of the system and then to use that data to derive the correction factors 
that need to be applied.  I believe Bob McGweir (N4HY) has that one on his 
list of code to be done (among plenty of other things I'm sure)

As to the hump at DC -- it's normal - everyone sees it from one degree to 
another.  It's AC hum and various mixing product of that hum.    Better 
soundcards and grounding produce less of it, but I believe most people see it.

Regards,

Bill (kd5tfd)

At 10:48 PM 5/14/2006, David Woodhead wrote:
>***** High Performance Software Defined Radio Discussion List *****
>
>I hope this is not out of place here but I had little luck on the flex 
>reflector...
>
>I'm evaluating a homebrew SDR based on the SDR1K and not sure if the 
>performance I'm seeing is normal.
>
>I am finding that the image rejection is not consistent from band to band 
>but the software only supports one set of correction values. I can 
>optimize on each band with a different set of values. Is this normal? I 
>susspect it is as the QSD balance (gain and phase) is unlikely to be 
>consistent across the bands. It would be nice to be able to store seperate 
>correction values for each band.
>
>My offsets are:
>............... 60m....40m....20m.......15m.......10m
>phase.......-.89.....-1.11...-2.08.......0.02.......-5.44
>gain......... 6.8.....7.05.....5.14.......6.61.......19.33
>
>How much image rejection should I expect from this design? I'm adusting it 
>by tuning the sig gen to the image frequency and then nulling the received 
>image signal. I can acheive ~70dB null across 6KHz. However, this is not 
>optimum for the observed image on the panadapter screen when tuned to the 
>desired signal - the displayed image is only ~ 60dB down (unless I 
>optimize for it). Are my results consistent with others measurements?
>
>Also,
>
>I have a noise spike in the panadapter (corresponding to DC). Its level 
>bounces - it peaks at -80dBm to -115dBm depending on the preamp gain 
>setting. I susspect this might be AC hum getting into the IQ audio. Is 
>this normal or might I have a ground loop or something?
>
>David
>KM5TZ



 1147668415.0