Sebastian Krzyszkowiak @dos@librem.one

@martijnbraam Found a bug!

I have two part numbers for the same value & footprint:

C5, C2 1µF 1206 C1966364, C24497

The first one matches, the second one does not (it's 100nF) - but partlint doesn't report a problem with the second one until I remove the first one.

Nice tool and timing, you pushed it just in time for when I needed it 😁

@nokernel Yes, used it a lot! There were many more bridges at first 😂 Got rid of all the rest, but this one is highly resilient.

TBH, it's the first time I solder such tiny things, it's a wonder I got as far as I did already anyway :)

Joints seem to be better now, but I didn't mention one thing - no matter how much hot air I blow at it or what I do with soldering iron, I can't seem to be able to get rid of a short between D+/D- on a passthrough port 😭 It's not critical, the important parts are already operational, but it would be nice to get USB 2.0 passthrough working too! I'm assuming it's on the USB-C plug, but it all looks perfectly fine to my eye 😫 I should finally get some proper flux I guess... #soldering #diy

Took me 3 days of soldering, but it works!

I can connect it to a Librem 5 and it automatically does all the needed PD stuff to expose its UART over USB-C 😁 It's also capable of a lot of other PD magic as a USB PD analyzer / injector / filter.

However, while these solder joints made a good job lasting till I got the firmware PoC working, it doesn't seem like they're going to last much longer... More soldering ahead 😫

#debubo #kicad #stm32 #usbpd #librem5 #shotonlibrem5 #diy #electronics #oshw

@alaraajavamma @MissBehave @PINE64 You may have to load module-bluetooth-policy in PulseAudio to have it switch to HSP automatically - but you can also switch manually in the settings. Otherwise - yep, as long as you can connect to your Bluetooth device it should just work.

@alaraajavamma @MissBehave @PINE64 No, with normal cellular calls. In fact, there's no difference on Librem 5, as modem's audio is just another digital audio interface seen by ALSA.

@alaraajavamma @MissBehave @PINE64 BT call audio worked for ages on the Librem 5 (with SparkLAN card - Redpine does not support HSP, which is a bit of a hurdle 😜).

@ptrc I had to adjust open files ulimit on my system because I regularly ran out due to terminal tabs.

@melroy > modern.IE

Oh wow, what a blast from the past. It still operates???

@beacondb Awesome, thanks! Works fine now:)

BTW. I've just looked into related MM code and APIs closer, and turns out that QMI modems report neighboring cell information with signal strength (RSRP/RSRQ). However, only the serving cell has its cellId available, the rest have only TAC/LAC (always the same for all seen cells, at least in case of LTE) and PCI. Could this potentially be used to improve estimation accuracy? Right now the endpoint outright rejects cells without cellId set.

@beacondb Ah, Geoclue doesn't include that field, and ModemManager doesn't expose it in its location API (it does in cell info API though). It used to work with MLS this way and still works with Google. Could this be handled on beaconDB side?

Ultimately Geoclue could do a much better job here, but this makes the versions currently out there incompatible with beaconDB...

@beacondb I'm trying with the first LTE cell from my operator in the dump and get 404:

$ zcat MLS-full-cell-export-final.csv.gz | grep LTE,260,3, | head -n 1
LTE,260,3,11,1306213,402,21.0048378,52.2180052,0,1,1,1670110185,1670110185,
$ curl -X POST https://beacondb.net/v1/geolocate -H "content-type:application/json" -d '{"cellTowers":[{"radioType":"lte","mobileCountryCode":260,"mobileNetworkCode":3,"locationAreaCode":11,"cellId":1306213}]}'