Lightning detector - Kickoff

I recently purchased a MOD-1016 chip for lightning detection in order to improve my weather station. The chip is based on the AS3935 chip and ships as a complete I2C-ready breakout module from embedded adventures.

First steps

I skip the following parts because I consider them trivial:

  • Soldering
  • Wiring to a Arduino Nano

The wiring part is actually the most tricky part, I will provide the schematics once I have a running system. For now I focus on getting the system online. The wiring on the following picture is accurate:

Wiring of the MOD-1016 to the Arduino Nano

I put everything together in a nice box to protect the electronics from the environment. In the end it will end up outdoors in my garden.
The box is IP55 compliant, so when deployed for real I will put it in additional plastic bag to avoid any issues that comes from rain. For the first experiments IP55 is fine. And this is how it looks like

Wired box, open

A small reader program is in my meteo repository (in the Lightning directory) on GitHub, and I let it run for 1.5 days.
I had some problems with the serial port on high baud rates, so i configured it for 9600 baud. The serial connection over this period was fine, but it seems that the location has too much interference.
All I got was constant "DISTURBER DETECTED"

Right now all I get out is "DISTURBER DETECTED"

Looks like I need some fine-tuning. I disconnected the device and will run some tests with my laptop on the go.

For now I have a running serial connection, the chip delivers some output, so I'm expecting that with some fine-tuning I should get this thingy running soon.

Sapphire Nitro Radeon RX470

About a month and half ago I purchased a RX470 and I want to write my experience with the card so far. I am not happy.

Originally I wanted to have a nVidia GTX 970. But the RX470 had 8GB RAM instead of 4GB and, more interestingly, the double precision float operations should be significantly faster. I had terrible experiences with ATI Cards and Linux drivers in the past, but I thought, since they have made some progress, it should be also a problem of the past.

I was so wrong.

So, I picked up the card and build it into my computer. Display was ok, GRUB and the BIOS displays are ok but somehow my Ubuntu doesn't come up with it.


Found out, AMD/ATI are not supporting Kernel Version larger than 4.8 - The DKMS driver doesn't build, because of an API change for linux/fence.c.

So I had to install Ubuntu 16.04 with Kernel 4.8. I had to uninstall my production system. I was really pissed.

After some some tries I managed to install the driver and was quiet happy for about a month, when a planned kernel update killed again the functionality. Of course, I forgot.'
So, again my production system was down. I though, well some time has passed, they have updated the driver for sure. Again, wrong. Even worse, AMD suggest to work with Ubuntu 16.04.02 and, quote

the recommended best practice is to install from 16.04.2 media but disable installation of updates.

Are you kidding me?!???? You are honestly recommending not installing Updates!??!?

I'm now stuck with the system and trying since two hours to get it back running and it's just a nightmare.

So, my conclusion so far: The hardware is cool and has nice specifications but the support for Linux drivers, even the mainstream versions, is still unacceptable.
I really hope they will improve their drivers for Linux.

I'm going back to nVidia.

StarParty Pfalzen/Falzes

Yesterday I have been an invited speaker at the Star Party 2018 in Pfalzen/Falzes in South Tyrol Italy.

I use this post to thank my former teacher Christof Wiedemair for the invitation. For me it has been a confidence boost - the audience was engaged and interested, like this it's fun to give a talk!

Star Party of the Dolomites still goes on on the Weekend, so if anybody around is interested on an insight of amateur astronomy, have a look at this Facebook event page - (Warning: Goes to Facebook) or the Empiricamente Event page.

I am honestly impressed by the passion and the professionalism of those people. It's worth going there!

Thanks again Christof, I'm looking forward to more opportunities like this 🙂

The joy of multicore processors

I recently came back from an amazing trip to Iceland. Had a glorious time there - it was an outstanding adventure!

After this trip I had to convert 3200 photos from RAW to JPEG. Turns out, that usually it takes something like 15 seconds for my computer to process 1 file. 3200 * 15 is roughly a bit more than 13 hours.

Some minutes later and a python script I could distribute the work my 8 processor cores of a FX-8350. The script is ugly and very very dirty, that's why I don't want to publish it.
I've started the script, went outside for a longer time and when I returned the process was completed.

This is just one example of things, where a large number of CPU makes sense on a workstation computer. Even normal users benefit more and more from increased CPU cores more and more.

AMD made good work when they were increasing the number of CPU's. And nowadays I willingly have a bit lower single-thread performance but twice as much CPU cores. Makes totally sense for multimedia converting, large picture editing, compiling, ecc.