Hi Leif! >The sampling capacitors should not cause gain unbalance at very low >audio frequencies. I suggest you check the DC blocking capacitors. >They form a high pass filter and maybe the capacitors differ, or >maybe the impedance the capacitors are loaded by differs. You should >be able to see easily with an oscilloscope.
I found the reason for the changing amplitude/frequency relation problem. The problem was found from the feedback capacitors of the opamps (10nF). I took them off and measured them and I had to realize that they both were from the opposite ends of the tolerance scale. There was almost 1nF difference and doing a short calculation how this effects different frequencies, proved that the problem was there. Obviously, I changed them to a matched pair (checked with my capacitor/inductor meter). Now, using Winrad, I can get solid -30..-35db image rejection. It changes only few db across the whole spectrum of screen. I'm quite happy about that, but this leads to the problem no 1. In Winrad, I still have to have the phase slider set to max value and the delay tag pressed to get even this image rejection. Is this the fault of the soundcard? What I also noticed, is that I could get better attenuation for the image with Powersdr, but only in a single spot. I dunno how these softwares differ, but Winrad seems to work better in this. Still have to study the linrad. >Oooh! I was referring to mirror image suppression and how it varies >with signal frequency. The boxes in the upper corners of the >waterfall can be used to thange the X-scale so you see the entire >frequency range. Ok, I have to do this. Do you prefer the IQ calibration first, or is the idea to see that what kind of image we have, without the calibration? 73 de Janne Pulkkinen, OH1GTF
