Lucas,
I've confirmed there's some issues with the B200/AD9361 code. I've added
a ticket on our internal issue tracker; I have no timeline for a
resolution though.
-- M
On 03/19/2018 04:47 AM, Lucas Val Terrón via USRP-users wrote:
> Hi all,
>
> Related to old questions I have previously asked in this mail list, I
> think I have found two possible bugs (or at least I think so) inside the
> UHD code. (IMPORTANT) Following explanations were proved using the uhd
> example tx_waveforms.cpp and an ettus B210.
>
> ./tx_waveforms --rate 61440000 --freq 2440000000 --wave-type SINE
> --wave-freq 100000 --gain 60 --bw 50000000
>
>
> First, I have realized that once you set a bandwith greater than 40MHz,
> the follwing warning appears:
>
> [WARNING] [AD936X] Selected Tx bandwidth (61.44 MHz) exceeds analog
> frontend filter bandwidth (56 MHz).
>
> From this warning, its clear, that somewhere inside the code is
> interpreting the sample rate (61.44 MHz) as the bandwith (50MHz).
> Looking in the code, I have found the piece of code that produces this
> warning:
>
> /home/com/uhd/host/lib/usrp/common/ad936x_manager.cpp (line 193)
> bool check_bandwidth(double rate, const std::string dir)
> {
> if (rate > _codec_ctrl->get_bw_filter_range(dir).stop()) {
> UHD_LOGGER_WARNING("AD936X")
> << "Selected " << dir << " bandwidth (" << (rate/1e6) <<
> " MHz) exceeds\n"
> << "analog frontend filter bandwidth (" <<
> (_codec_ctrl->get_bw_filter_range(dir).stop()/1e6) << " MHz)."
> ;
> return false;
> }
> return true;
> }
>
> Here, we can see clearly that the code is checking if the sample rate
> (up to 61.44MHz) is inside the bandwith range limits (up to 56 MHz).
> From my point of view this check has no sense, because instead of the
> bandwidth of the signal the function is provided with the sample rate...
>
>
> Secondly, there is a limit of 40 MHz when the bandwith of the signal is
> set (It must be remembered that the maximum bandwith of the B210 is 56
> MHz). This "bug" comes from
> /home/com/uhd/host/lib/usrp/common/ad9361_driver/ad9361_device.cpp (line
> 385).
>
> double ad9361_device_t::_calibrate_baseband_tx_analog_filter(double
> req_rfbw)
> {
> double bbbw = req_rfbw / 2.0;
>
> if(bbbw > _baseband_bw / 2.0)
> {
> UHD_LOGGER_DEBUG("AD936X")<< "baseband bandwidth too large for
> current sample rate. Setting bandwidth to: "<<_baseband_bw;
> bbbw = _baseband_bw / 2.0;
> }
>
> /* Baseband BW must be between 20e6 and 0.391e6.
> * Max filter BW is 32 MHz. 32 / 1.6 = 20
> * Min filter BW is 625 kHz. 625 / 1.6 = 391 */
> if (bbbw > 20e6) {
> bbbw = 20e6;
> } else if (bbbw < 0.391e6) {
> bbbw = 0.391e6;
> }
>
> double txtune_clk = ((1.6 * bbbw * 2 * M_PI) / M_LN2);
> uint16_t txbbfdiv = std::min<uint16_t>(511,
> uint16_t(std::ceil(_bbpll_freq / txtune_clk)));
> _regs.bbftune_mode = (_regs.bbftune_mode & 0xFE)
> | ((txbbfdiv >> 8) & 0x0001);
>
> /* Program the divider values. */
> _io_iface->poke8(0x0d6, (txbbfdiv & 0x00FF));
> _io_iface->poke8(0x0d7, _regs.bbftune_mode);
>
> /* Enable the filter tuner. */
> _io_iface->poke8(0x0ca, 0x22);
>
> /* Calibrate! */
> size_t count = 0;
> _io_iface->poke8(0x016, 0x40);
> while (_io_iface->peek8(0x016) & 0x40) {
> if (count > 100) {
> throw uhd::runtime_error("[ad9361_device_t] TX baseband
> filter cal FAILURE");
> break;
> }
>
> count++;
> boost::this_thread::sleep(boost::posix_time::milliseconds(1));
> }
>
> /* Disable the filter tuner. */
> _io_iface->poke8(0x0ca, 0x26);
>
> return bbbw;
> }
>
> Here, you can see that the function sets to 20MHz (in baseband, 40MHz in
> RF), bandwidths greater that 20MHz (in baseband, 40 MHz). That is why
> the limit appears. Taking into account that B210 max bandwith is 56 MHz,
> this check must be set to 28 MHz (56/2 MHz in baseband).
>
> Please, confirm me if my assumptions are valid.
>
> Thanks in advantage,
> Lucas
>
>
>
> _______________________________________________
> USRP-users mailing list
> [email protected]
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>
_______________________________________________
USRP-users mailing list
[email protected]
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
