Janne Pulkkinen wrote: > The 192kHz ASIO with my PCI E-MU 0404 don't seem to work. The software > does something unusual: No audio output and the S-meter drops kind of > "off". But it does great with 96kHz ASIO or 48KHz ASIO which I have > tested.
Hi Janne, EMU cards are a bit peculiar, as each one seems to work a bit differently from the others. I have recently bought on eBay a second hand EMU-1212M, but never had the chances to install and to test it, so the 1-to-4 sample delay added in the V1.30 was tested by another OM who has this particular card, and he reported back that now, inserting a 4 sample delay when sampling at 192 kHz, he is able to perfectly null the image. Frankly I don't have enough information on your EMU-0404 to formulate any hypothesis, apart from what said by Jeffrey, i.e. that probably when using the ASIO driver the channel numbering is non-standard. I plan, in one of the next versions, to give the user the possibility to dynamically alter that channel assignment. > > Also a question about the DLL support. I think I didn't see a > possibility of chancing the "channel skew calibration" -parameters? I > think there are many designs that have a changing IQ balance depending > on the frequency. I can have -60db image rejection anywhere through > 100kHz to 30MHz (thanks for your 0-4 sample delay!) but I have to do it > manually now. Is there anyway to add some kind of frequency/image > rejection mapping? It would also be great to have those parameters on > DLL support. Yes, I see your point. It would need two things : 1) The DLL is informed about a change of the Tune value on the Winrad screen. But this is already implemented, see the API TuneChanged in the document describing the DLL architecture. 2) The DLL has a mean to send back the parameters for the image rejection routine, relative to that frequency. This is not implemented presently, but it would not be much difficult to do. Of course the drawback is that, while the rejection would be almost perfect at the frequency of reception, it would become visually worse in all of the other points of the wideband spectrum/waterfall... My personal opinion is this. First of all, an automatic rejection algorithm could be coded, there are examples already implemented and that work fairly well. Secondly, the need for am image rejection routine will be disappearing in time. Such a routine is needed only for RX architectures where the generation of the I/Q pair is done in the analog realm, like all the RX based on the Tayloe mixer or variants thereof (H-mode, etc.). But the present trend is to go more digital, i.e. the ADC is moved nearer to the antenna, see the SDR-14, SDR-IQ, Perseus and the new QuickSilver (and Mercury, when ready). In all of those receivers, the generation of the I/Q pair is done when the signal is already in a digital status, and as such, the I and Q components are already perfectly balanced, and no compensation is needed. So personally I see this balancing a thing where to spend some time, but not more than a given amount. Things would be a bit different if TX also is contemplated, but then the problems are different, and so the solutions. Enjoy Winrad ! 73 Alberto I2PHD _______________________________________________ Winrad mailing list [email protected] http://winrad.org/mailman/listinfo/winrad_winrad.org
