Aliases and Mirroring Issues

[bpadalino]

I've been able to reproduce the problem here and it concerns me very much as I don't remember this occurring before either.

I tried doing some bisections on the code but I couldn't come up with a good working set.

scancapecod - you said that before it didn't happen, but now it does. Is there any way possible you may know the versions of the working setup versus the non-working setup? Can you verify that the "old" setup did work as you expected?

Sorry for being quiet on this. I verified it last night, tried to investigate it today and didn't make very much progress so hearing any version numbers which may or may not work might be a good start here.

If I find a smoking gun, I'll let people know - but this is definitely a huge concern.

Brian

[scancapecod]

Brian,

That was actually Colin (CC90) that commented regarding it not happening. It's been happening to me since day 1 of owning the BladeRF (1/24/14 as I recall). I've used many of the Build-O-Matic firmware and FPGA files as well and I've noticed this issue in all of them. I have not tried any firmware earlier than 1.6.1 or any FPGA earlier than 0.0.3.

I wonder if you made this confirmation using SDR-Radio? I still need to find the time to set up a virtual host with Linux so I can try gqrx with it.

I will reiterate that reception is superb, weak signals very monitorable, etc.. It is definitely stronger signals that are contributing to the aliasing and mirroring in my case, anyway.

Thanks.

[bpadalino]

I also notice it only with strong signals, and the aliases/spurs showing up maybe 25 to 30dB below the main large signal. IQ compensation only removes the single conjugate image which shows up.

Colin - if you're still monitoring the thread - can you let us know what version you fell back to which doesn't have the issue?

Brian

[CC90]

Hi,

Sorry it's taken me a while to reply.

I had real problems with mirroring mid January after loading fpga v0.0.2 (see https://www.dropbox.com/s/5xnx81y3f52uy ... 143555.png). I'm afraid I can't recall exactly which firmware I was using at the time, but I think it was v1.6.0 (it is 1.6.0 now and I don't remember changing it). I rolled back to fpga v0.0.1 and things improved markedly. I should point out that I do see aliases of very strong signals, but not to the extent that Scott has in his screenshots.

I haven't updated the firmware or fpga since as I'm in the process of moving workshop. I'll try the latest builds as soon as I have my set-up usable again.

Colin

[scancapecod]

Thanks for responding Colin.

I'm also seeing strong signals showing up that aren't even in the current sample range of frequencies. I have a DMR signal for our local power utility company that is at 451.125. During sampling of the 470-475 MHz range last night I noticed what looked like a DMR signal at 471.125. Using the available DSD software I was able to recover the DMR audio and it was an image of the signal at 451.125, exactly 20 MHz below.

Hopefully this will be addressed as more information is provided. I'll continue to try to do so as much as I can.

[scancapecod]

Brian: any further thoughts regarding this? Would it help if I provided range in MHz between actual signals and images? Anything I can do to provide input that will help move us to a solution, please let me know and I'll try.

[freqmod]

I flashed the builds from hoopycat from 2014-03-20 04:57:22 for firmware and fpga40 (using linux because windows bluescreens when flashing firmware from the installer), Installed windows driver 1.2 from http://nuand.com/downloads/ (without selecting to flash firmware).

If i reconnected the bladerf and started sdr console v2 (2.2. build 1676) i got mirrors.

If i closed the sdr-console (or havent started it) then disconnected the bladerf, and ran bladerf cli. In bladerf cli i did help calibrate, and ran calibrate without arguments, and with the arguments listed in help. Then i closed the cli and opened sdr console. Now i have no mirrors issues. (until the next time i disconnect the bladerf, or maybe reboot).

[scancapecod]

Thanks for the info. I followed what you did above. I need more time to watch....I'm still seeing some mirroring but naturally the signals I've been watching to gauge this have decided to take the night off.

Actually one of them just lit up....unfortunately for me I'm still seeing aliases although they "may" be reduced a bit. I'll have to continue to watch. I'm looking at a 5 MHz span from 460-465 MHz.

When I ran calibrate I got this message:

bladeRF> calibrate
[WARNING] Never converged - DC_CLBR_DONE: 1 DC_LOCK: 0
[WARNING] Never converged - DC_CLBR_DONE: 1 DC_LOCK: 0
bladeRF>

But when I ran each calibrate parameter individually I received no errors, only a return to the bladeRF> prompt.

Just to be clear on what I did, the sequence I used was:

1. run the CLI.
2. Flash to the firmware you suggested.
3. Recycle power.
4. Run the CLI again.
5. Load the FPGA.
6. Run Calibrate
7. Exit the CLI.
8. Run SDR-Radio.

Thanks,

Scott

[bpadalino]

It seems like he also ran the calibrate with the arguments listed in the help (tuning, txlpf, rxlpf, rxvga2).

I have a new calibration routine for rxvga2 but it didn't seem like it was giving me a great difference.

I'll push it out this weekend to the master branch so we can all play.

Sorry for being so quiet on this. Lots of things happening simultaneously.

Brian

[scancapecod]

I ran just calibrate and got the warnings. When I ran each individual command they all passed without any issues and just returned me to the prompt.

I know you guys have a lot of stuff going on. It speaks volumes to the diversity of this device. And here I am, just using it as a receiver. I'll get brave eventually....must walk before running.

[freqmod]

bladeRF> version

bladeRF-cli version: 0.10.3-git
libbladeRF version: 0.12.1-git

Firmware version: 1.6.1-git-b876a96-buildomatic
FPGA version: 0.0.3


Commands used:

calibrate tuning
calibrate txlpf
calibrate rxlpf
calibrate rxvga2


ps i needed to disconnect the bladerf, run the calibrate, and start sdr-radio. Running calibrate in the cli after running sdr-radio without disconnecting did not have any effect. The FPGA and firmware was flashed into the chip using linux tools (it is likely possible using the most recent bladerf-cli on windows too).

It is also important that the gains and filters for the bladeRF, which is controlled by sdr-radio is set to sensible values (not too low, and not over driving the radio; that can cause aliasing and mirroring by itself) . Try to play around with them.

PS: It looks like the scales for the gain and filter controls are flipped or messed up in some other way. A gain of 0db give me a much stronger signal than a gain of 30db.

Screenshots of calibration :
Before:

After:


(btw it looks like the forum cuts some parts of the image dependent on your screen site. Try to open the images in a separate window or tab to see all parts of the images )

[scancapecod]

Thanks for the detailed information. I too wondered if the gain settings in SDR-Radio might be in need of some work. That said I have tried a lot of tweaking with them, and while I can minimize the alias and image issue, it winds up being at the expense of weak signals, which I enjoy chasing.

I did the calibration before running SDR-Radio, and then again after quitting SDR-Radio, disconnecting the USB connection and reconnecting. So I think I must be doing it correctly unless I'm missing something.

[freqmod]

After looking around some more i found mirrors of tv (DVB-T) signals after calibration, so it is not perferct even if it looks better at 430 mhz. I have unfortunalty not tested how the mirrors look like without calibration.

[scancapecod]

Well now I'm a little curious about the calibration command.

When I type "calibrate" by itself I get the two warning lines that I think I posted somewhere above.

When I type calibrate followed by one of the four individual commands shown when "help calibrate" is typed, I get returned to the command line.

Just for giggles tonight I typed something other than one of the commands, i.e. "calibrate this". I received no error message and was returned to the bladerf command line.

So....what is the proper method? Just "calibrate"? If I run SDR-Radio Console and type calibrate at the bladeRF command prompt I see an immediate increase in the DC spike at the center of the FFT. After about 30 seconds it calms down.