Official bladeRF forum

Hi ,all
I successfully connected my bladeRF x115 to the linux laptop(Ubuntu 13.04).After I run the apps osmocom_fft,I found there is a very high peak in the center frequency ,about -30 dB!,while the ground noise is -90 dB. I tried to change the gain ,but the peak is still 40~50 dB higher than the ground noise.Could anyone tell me what causes this ? Is there any proach to solve this problem? Thank you !

Center frequency in osmocom_fft corresponds to zero Hz at the inputs of the ADCs. This means that the peak comes from the DC voltage at the inputs. This happens quite often in systems where the RF signal is downconverted to 0Hz. Even slightest offset in bias voltage at the inputs of the ADC will show up as a peak at the center frequency on fft plot.
Most of SDR programs can filter out this peak. You can try gqrx and switch on DC removal.

j.

piranha32 wrote:Center frequency in osmocom_fft corresponds to zero Hz at the inputs of the ADCs. This means that the peak comes from the DC voltage at the inputs. This happens quite often in systems where the RF signal is downconverted to 0Hz. Even slightest offset in bias voltage at the inputs of the ADC will show up as a peak at the center frequency on fft plot.
Most of SDR programs can filter out this peak. You can try gqrx and switch on DC removal.

j.

Thank you very much!

You may have figured this out all ready but using gnuradio-companion you can add a DC blocker right after the osmocom source with a length of 1024 and that seems to remove the spike. Not sure if this is the correct way to do this but it worked more me. I then added the WX GUI FFT and Waterfall just like the osmocomm_fft app does.

o/ to everyone...

I'm seeing a similar problem with a different configuration. Ran the bladeRF CLI, loaded properly, updated to the latest firmware, etc., then launched SDR Console to explore the spectrum above 300 MHz. I find a 20-30 dB spike on my center frequency with the Rx SMA terminated. Looking at this topic makes me think I have a similar issue but haven't figured out how to resolve it (yet). Suggestions are always (well, nearly always) helpful.

Bob

Some relevant info is presented in this wiki page. However, it's using gr-osmosdr for examples, so that won't help much in windows.

The basic ideas regarding the use of libbladeRF's bladerf_set_correction() still apply, though; one could write something similar for Windows. On the TX side of thinkgs, you could use the bladeRF-CLI to transmit transmit a tone and manually dial in DC offset & IQ balance corrections, if you have a spectrum analyzer. The RX side of things would be a little more painful, given that only one process can have the bladeRF open at a time.

I know there had been some talk of a DC correction block going into the FPGA...not sure what the priority on that is at the moment.

drbob wrote:o/ to everyone...

I'm seeing a similar problem with a different configuration. Ran the bladeRF CLI, loaded properly, updated to the latest firmware, etc., then launched SDR Console to explore the spectrum above 300 MHz. I find a 20-30 dB spike on my center frequency with the Rx SMA terminated. Looking at this topic makes me think I have a similar issue but haven't figured out how to resolve it (yet). Suggestions are always (well, nearly always) helpful.

Bob

Bob,

SDR Console should be always correcting this spike. Without an antenna you don't have any real data though - with an antenna the spike should reduce over time.

In some weeks I'll have decent antennas up and will use the bladeRF more.

Simon @ SDR-Radio.com

First, 73's to Simon and a huge thanks for your contribution to the amateur radio world and the SDR world with your work...

I'll try a different approach and put an antenna on the unit to see if this corrects the problem and, assuming so, I'll report back here. That's good information to know.

Revisited: Today I spent some time "playing around" with an added antenna (instead of the terminator). I still have a significant peak (15 dB or so) centered on-frequency. After 4 or so hours of testing, I didn't get to see this effect go away.

Hello Everyone,
I have similar problems on the TX side, I'm monitoring the output on a spectrum analyzer, there is a +20 dB DC spike above the actual signal level (it moves with the spectrum when adjusting the tx gain). IQ correction is not improving the situation.
It clearly shows up only when using higher sample rates, interestingly, on an older notebook the spike appeared only above ~22 MHz sample rate, with a recent machine (8th gen. i7) I can go up to 40 MHz, but I really need the full 61.44 MHz.
I have tried adjusting tx buffer parameters as well, without any improvements, tried with gnuradio (to transmit wideband noise) and bladeRF-cli (playing back pre-recorded signals), tried a RAM buffered folder, etc, exactly the same issue.
Looking at the RX spectrum while recording with 61.44 MHz, the DC spike was not present.
I'd be really grateful for solutions (other than using an even faster/better PC, since I can't see which parameter exactly in the PC I'm using now is the bottleneck now - if it is at all)!

Thanks!
Geza