Hello all,
I want to use X15 with usrp b210,b205 and 2932. My question is if I connect usb series usrp with high speed 3.0 port of X15. then Will it support the highest sampling rate on USRP?. What is the maximum I can acheive?. Even with core i7 high speed ports it gives 40 MHZ hardly and sometimes overflows and underruns.
Another question. What is the maximum USB 3.0, Esata, main Data Bus (through which ports and emmc is connected) and eMMC read and write speed on X15 board?.
I want to use it as a single board computer for my mini usrp b205. I want to buy it soon. Need suggestions urgently. Whether to buy this thing for USRP or not.
Thanks,
Hashir
Hi.
I wish I knew the answers off hand. But am just starting to work through gnuradio, and the
https://www.ettus.com/product/category/USRP-Bus-Series LOOKS like it may be a good platform for hobbying around with gnuradio, and base band/bandpass processing blocks! Albeit, it’s $$$. Been hoping to work on an SDR project involving the X15, but that hasn’t really gotten off the ground yet. But am making progress towards it…
Maybe if you can provide more insight into what the highest sampling rate on the USRP is, what information is being exchanged over USB 3.0, and what you’re wanting the X15 to do with the USRP data, specifically, what portion of the application will run on the X15’s A15 cores, C66 DSP’s, PRU’s, MPU’s, etc, then that may spawn some thought on this forum.
Also, TI E2E2 maybe able to answer some of the questions as the TI 572X EVM is very similar to the BB-X15.
I want to use it to capture IQ data.
So then data is coming from the IQ demodulator at 40 MHz on the USB 3.0 bus from the b210, and you 'd like to have it fill up a buffer via DMA, and then generate an interrupt for one or both C66 DSP’s to demodulate, filter, downsample, filter some more, perform error correction coding, until you end up with a data stream??
I am not sure about dma and buffers. I dont have good knowledge of it. I just want to use normal gnu radio block diagram on X15. Even on my high speed computer (not so much high but good atleast). If I increase the sampling rate more than 40MHz. Gnu radio shows overflow and underrun error. This is because the system can not process it before the reception of new data. You are right that we can use buffers. I have tried it but it was not working on gnu radio. I still got the errors of underrun and overflow. Can this board give me 40MHz sampling rate with usrp ?
Thing is USRP can transmit data through USB 3.0 at max speed it can. But Can this board process it ?. It should because it has USB 3.0 port. I am not sure about the Bus speed of the board and other things
Well, is the usrp able to give you baseband data?
I did some very very rough benchmarks for each of the 2 C66DSP’s, several months ago showing
Each C66DSP:
Fs = 100kHz.
60, 152 tap FIR filters
Close to 100% utilization for each C66 core.
Assuming C66’s are clocked at 750 MHz.
There’s also a benchmark document from TI for the C66 (Think it is one of the technical documents you can view associated with the am5728 processor on TI’s sites). That document showed:
256 Point FFT
1782, c66x cycles
Execution time = 2.373 uS at 750 MHz clock speed.
If you’re IQ samples are at Fs = 40 MHz, depending on what you want the C66’s to do with them then this MAYBE an issue for the C66 DSP’s,…
Yes I thik it gives the baseband data after passing through DAC/ADC . The thing which you are explaining is the processing which we can do locally. But the USRP gives the data after sampling. I do not want to make things complicated. I am pretty confused to buy it or not for this operation. because its 280$ board. I can go with Intel NUC too
Hey Hashir,
I’ve been exploring similar setups and faced some of the same questions regarding USB 3.0 throughput and sampling rates with the USRP B210. From what I’ve seen, even on decent desktop systems, hitting reliable 40 MHz without overruns can be tricky, especially if the system isn’t optimized for high sustained I/O.
Regarding the BeagleBoard X15, it does have a USB 3.0 port, but the sustained throughput and overall system architecture (e.g., how fast it can move data from USB to memory and into DSPs or processors) may not keep up with what the B210 can output at its max rates. I haven’t benchmarked it personally, but others have noted similar performance ceilings when pushing above 30–40 MSPS.
If you’re mainly looking to capture IQ data without a lot of real-time processing, and just need a stable host for the USRP, you might also want to look into more compact x86 systems. Some users I know have had better luck with small-form-factor PCs like Intel NUCs or fanless mini-PCs they seem to handle USB 3.0 traffic a bit better in some cases.
For reference, I’ve been comparing different B210 variants recently, including models like this Luowave USRP LW-B210 and the standard Ettus-style USRP B210, and they all face the same host bandwidth bottlenecks when pushed hard. The differences mostly come down to firmware and enclosure, not core performance.
Hope this helps a bit while you decide.