[USRP-users] E310 RFNoC FFT Overrun Issue

Ramazan Çetin ramazan.cetin at gohm.com.tr
Sat Feb 9 17:59:01 EST 2019


Hi Marcus,

I have compiled proper image(fft, ddc, keep 1 to n) and did what you 
said. Now, it looks better, i am getting very little overruns sometimes.

I have another question. I am sorry if question is not related with this 
topic.

I changed my flowgraph to this:

RFNOC_radio --(56MHz)> RFNOC_FFT -> RFNOC_DDC --(2MHz)> complexToMag -> 
Log10 -> FileSink

I want to get wifi samples from E310 which has 56 MHz analog BW. So, i 
can get wifi samples from FPGA but not in CPU. I want to see wifi 
samples on frequency spectrum on CPU. So, i thought that i can get 56 
MHz signal on FPGA and get FFT. After FFT, downsample FFT results and 
push FFT samples to CPU as 1-5Ms/s. But when i looked my signal on CPU, 
its so distorted.

Q1. So, do you know any method to minimize signal distortion while 
downsampling in frequency domain? or How can i get wifi signal frequency 
spectrum in E310 ?

Q2. When i try compiling FPGA image with 4 blocks, it gives "Placer 
could not place all instances" error. But in guide, it says user can put 
up to 6 blocks. Why it give that error after 3 blocks?

Best regards.

On 9.02.2019 17:11, Marcus D. Leech wrote:
> On 02/09/2019 07:05 AM, Ramazan Çetin via USRP-users wrote:
>> Hello all,
>>
>> I want to get fft of 50Ms/s signal and after fft using DDC i will 
>> pass the samples with 1-5Ms/s to ARM processor.
>>
>> I have compiled an FPGA image with FFT, DDC and FIFO.
>>
>> |   |     _____________________________________________________
>> |   |    /
>> |   |   |       RFNoC blocks on this device:
>> |   |   |
>> |   |   |   * Radio_0
>> |   |   |   * DDC_0
>> |   |   |   * FFT_0
>> |   |   |   * FIFO_0
>>
>> I have a gnuradio flowgraph (It is attached);
>>
>> Radio -> strToVector -> FIFO -> FFT -> vectorToStr -> CompToMag -> 
>> NullSink
>>
>> When i used master clock rate 1M, there is no overruns. After 1M for 
>> example (1.5M) i am getting overruns.
>>
>> [INFO] [E300] Performing register loopback test...
>> [INFO] [E300] Register loopback test passed
>> [INFO] [0/Radio_0] Initializing block control (NOC ID: 
>> 0x12AD100000000000)
>> [INFO] [0/DDC_0] Initializing block control (NOC ID: 0xDDC0000000000000)
>> [WARNING] [RFNOC] Can't find a block controller for key FFT, using 
>> default block controller!
>> [INFO] [0/FFT_0] Initializing block control (NOC ID: 0xFF70000000000000)
>> [INFO] [0/FIFO_0] Initializing block control (NOC ID: 
>> 0xF1F0000000000000)
>> [WARNING] [RFNOC] Assuming max packet size for 0/FIFO_0
>> Press Enter to quit: overrun on chan 0O
>> overrun on chan 0
>> Ooverrun on chan 0
>> Ooverrun on chan 0
>> Ooverrun on chan 0
>> Ooverrun on chan 0
>> Ooverrun on chan O0
>> overrun on chan 0
>>
>> So, why is this happening? I thought i am handling data in FPGA so i 
>> can get FFT of 50Ms/s signal. I just need to reduce sampling rate 
>> when i pass the samples to ARM processor. What is wrong with my setup?
>>
>> Thank you. Best regards.
>>
>> Ramazan
>>
>>
> Notice how there aren't strToVector blocks in the list of RFNOC 
> blocks, nor complex-to-mag.   That means the samples are visiting the CPU
>   at high rate.
>
> The flow would be:
>
> Radio-->FFT--->IIR_FILTER-->KEEP_1_IN_N--->then do the complex-to-mag 
> on the host.
>
> However, you're missing RFNoC blocks for the IIR filter and KEEP_1_IN_N
>
>



More information about the USRP-users mailing list