Did this ever get sorted so the script works or was the problem a one off?
Give LimeSuiteCalibrate --freqStart=433.525e6 --freqStop=433.525e6 replacing the frequency with your own a try.
I just gave using the LimeSuiteGUI calibration a go as a single frequency calibration, seemed to work.
I was tuning ADBS and seeing ghosts of the signals on the other side of the DC spike using SDR-Console v3 so I stopped the SDR Console (not shut down, just stopped the radio connection), started up LimeSuiteGUI and set SXR to 1090, calculate then tune then off to the CLKGEN tab to calculate and tune the main oscillator.
I disconnected from the LimeSDR and the GUI then restarted the SDR-Console and now I get clean spectrum on either side (as I should with the filter on the port) , lower noise floor (not by much) and the ADBS signals aren’t ghosting.
I think that may be your immediate solution to the single frequency calibrate question.
I ran the LimeSuiteCalibrate program last night and today it shows the .db file is populated probably correctly but the SDR_Console program doesn’t seem to use that file so I was still uncalibrated.
Might be my issue though and placing the .db file in the wrong place on the Win machine.
have you tried above 2ghz?
if so, did the DC spike go down and noise floor drop some?
I hadn’t till you suggested.
Here is what I did and got:
Started the laptop, got SDR Console V3 running at 2304038400, I chose there because there were some spikes and they looked mirrored.
The centre spike was large enough to have the notch filter activated dragging it’s middle third down below the nf but leaving these two big humps on either side of the notch15dBm above the nf. Several bouncing smaller spikes on either side and the display 1 MHz wide.
Stopped the SDR Console, started up LimeSuiteGUI, connected to the LimeSDR and used the Chip -> GUI button to populate the interface with the SDR Console settings then enabled DC offset and calibrated the receiver.
I then pressed the GUI -> Chip button.
I then disconnected from the LimeSuiteGUI interface and hit the start button on the idle SDR Console interface and the results were pretty obvious.
The nf actually rose by 5dBm but the extraneous spike noise mostly vanished and the central spike is no longer large enough to be notching down and is now around 20dBm above the noise floor.
Not perfect but waaaay better.
There were no ghosting images of an actual signal in this test as it was the ADBS signal mirrored on the other side of the spike that made me head for calibration last night. Might be something else but it did fix the ghosts and I got better signal out of it.
Ok . i tried what u did with a SXR of 3415 … and i get an error of sorts …
First try i get a response from LimeSuiteGUI that im not connected, check by seeing if i can pull chip info … yep still connected, then try again and i get this error …
I decided to do a git pull / rebuild – my limesuite is a few months old now … not sure what else might be a good idea to update
Welp … after rebuild of latest LimeSuite … i no longer get that error …
The image rejection at 3.4ghz got better but only by 10db or so … still a 30db image.
Yeah, that spike is still pretty big after auto calibration.
Central spikes are just kinda one of those things in SDR I understand, the HPSDR derived gear doesn’t have that issue but it also uses a different method of DDC that is immune to it. Form what I understand from testing in HPSDR the values that converge to cause the spike in most DDC systems actually diverge in the HPSDR type gear.
Image suppression is the main goal here, seemed to work pretty good for the 1090 signals.
The center (DC) spike is from the DC blocking capacitor in front of the ADC … nothing can be done about it in hardware so its removed in software … but requires good I/Q calibration first.
Do you have anything that transmits signals above 2Ghz?
ever thought of getting a HB-100 and modify it to Tx FM?
Only the LimeSDR itself, I should pick up a signal generator that can cover the SHF stuff.
I also noticed somethimg realy worng going on with calibration cache. With removed cache gqrx spectrum looks like this(which is nice):
However when i restart gqrx without removeing cache(new cache was created during previous run) it looks like this:
Configuration was not changed. Temporary fix is to remove cache.
No to be picky …
But LNA gain is max, tia 0, pga maxed then next pic, LNA &TIA are 0 and only PGA is maxed
Hello, what is the up to date status of the I/Q calibration ?
I use the LimeSDr through the SoapySDR interface and I observe I/Q inbalance at low gains,
apparently not at high gains, could anybody confirm the effect ?
There is a notice about calibrating with LimeSuiteCalibrate but the python scrypt is no longer there
and there is another notice about autpomatic calibration.
Indeed I see messages about calibrating when starting the stream.
Thank you.
Same directions for limesdr-mini?