On Mon, Sep 10, 2018 at 10:40 AM Rob Kossler via USRP-users < [email protected]> wrote:
> Hi, > I just built an X310 RFNoC image where I included two DDCs, two DUCs, and > the replay block. I have a couple of questions about the NOC block XML > files: > > - I am getting a warning that the block ID for the replay block is not > recognized (see console output below). I checked the <install > prefix>/share/uhd/rfnoc/blocks/ folder and I found 3 replay blocks > (depending on how many channels in the block). One of them has the correct > NOC ID. So, I don't understand why I'm getting the warning. My > understanding is that as long as one of the blocks in this folder has the > correct NOC ID, it should be found. Is that incorrect? > > I can't answer all your questions, but I think maybe this commit answers your NOC ID issue: https://github.com/EttusResearch/fpga/commit/fcb865e7a6e006141f81551cdcba419df0bc3000 There was one missing 0 so the NOC ID looks very similar but is off by 4 bits. Try changing that. > > - When I built the image, I used a yml file because that was the only > way I knew for obtaining 2 channel DDCs and DUCs. However, I did not > realize that I should have changed the NOC ID to have "2" as the last digit > for DUC. My question is: does it matter? If I'm not using gnuradio, is > there any issue with the XML file indicating that there is only 1 channel > in the DUC when there are actually 2? Will multi_usrp work differently > than it would if this were correct? > - Regarding the previous question, why do the NOC IDs for DDC and DUC > differ such that "ddc.xml" is for a 2-channel implementation but has a NOC > ID ending in "0", whereas "duc.xml" (also 2-channel) has a NOC ID ending in > "2"? Furthermore, "ddc_single.xml" is for a 1-channel implementation and > has a NOC ID ending in "1" while "duc_single.xml" (also 1-channel) has a > NOC ID ending in "0". > > From my understanding of the situation, unless you're using TwinRX cards, you can only use the single channel DDC. Each block can't really get more than 200MHz worth of data to it, so having the 2 channels is really for 2 real IF's which are being downconverted. Anything that samples IQ would only really use one anyway, and the other path is not utilized. As for the 0, 1, 2 endings, I don't think RFNoC is smart enough to understand how many outputs/inputs are on things inherently. So if you had a block with a generic setting the number of stages, you also need a different NOC ID to identify it as the one with that number of inputs/outputs. I hope this was helpful. Someone from Ettus should probably still respond, though. Brian
_______________________________________________ USRP-users mailing list [email protected] http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
