A fee, are you kidding? I see this as a community project. The more you centralize or throttle or keep proprietary any aspect of it, the less useful it becomes to the world.

Yes I know storage costs money. But some of us would be willing to host our own data. I’d like to use IPFS somehow, but not sure yet whether there is a database that is a good fit for this kind of time-series data on IPFS. Maybe will evaluate OrbitDB, but it’s written in JS so I’d have to run it on NodeJS… ick.

I have experience with InfluxDB. I have doubts about its ability to scale, based on what I’ve heard at work (we use it for another purpose: automated software benchmarkes and test results and such). But on my system at home ATM it’s only taking up 157MB of memory, and I’ve been recording the data from one uRadMonitor at a time more or less continuously since February 2017, and a few other data points from other devices here and there. We’ll see how much it increases later. (I was just thinking that I should use influx to log its own memory usage, so that I can graph it and look for a trend.) My crappy little script to insert data (which I run from a cron job) is at https://github.com/ec1oud/uradmonitor-influxdb-inserter But I suspect if you try to store all the data worldwide it will be much more demanding.

I used to use https://en.wikipedia.org/wiki/RRDtool (a round-robin database) for weather data years ago. It didn’t have performance problems (being designed a couple of decades ago for the computers we had back then), but you have to decide in advance how the data will be resampled or otherwise reduced as it ages. The idea is to have high-resolution data for the most recent time period and summarized data for longer periods. I’m not fond of the idea of losing any of this data too quickly; who knows how many years is really enough. Whereas with Influx you can just start storing data very easily, and change the “retention policy” later on when it gets too unwieldy.

Anyway if you can’t afford the storage, it would be nice if you can provide a worldwide aggregated stream of some sort so that the rest of us can experiment with alternate means of storage. Perhaps a multicast channel would work, sending out a big JSON packet once per minute with all the readings from all the devices? It would not require much bandwidth.

Downgrading my compiler to 6.2.0 didn’t fix it.

Will keep poking around.

I was missing -funsigned-char – the display is working better now. The web server still doesn’t seem to be working quite right though.

@radhoo I received the A2 yesterday. It works great, and I’m glad to see that the temperature/pressure sensor is already installed. Thanks!

But that reminds me to install the BME on the Kit1, get back to trying to update the firmware and start adding features to it. So as a first step I tried to reproduce a working build. I started with the latest code from github, forked it, and wrote a Makefile to build it on Linux. I got a firmware built, flashed it, and it kindof works but some characters on the display are garbled, while others aren’t. It also doesn’t appear to have a functioning web server – it gets connected to the network but doesn’t respond when I try to connect from my browser. Today I fixed all the compiler warnings; that didn’t help. My fork is here https://github.com/ec1oud/uradmonitor_kit1

Any ideas? Has anyone managed to rebuild the firmware and start hacking yet?

Fortunately I downloaded the original firmware from the chip, so I can restore it and have it working again. That shows that there is no problem with the hardware or the fuses being set in prog.sh – it’s just a bug in the open-source version of the firmware, or that it compiles differently with my version of avr-gcc than it does with Radhoo’s version, or something like that.

I also managed to bend the legs of the fuseholder such that it’s possible to solder both ends, like Robert described. They strike me as quite a flimsy holder for the tube; but on the other hand, they are also too close to the mounting holes. So since they are so flimsy, it wasn’t too hard to bend them over a bit crooked, to make it possible to bolt the board into a case I designed, which is here: http://www.thingiverse.com/thing:1644939 The case includes struts which, if you have bent over the fuseholders the same way I did, will help hold the tube in position, so that it doesn’t fall out.

I’d suggest using sturdier fuseholders, but also making the board just a little bigger, to get those mounting holes away from the tube. Also, it would be nice to have holes on all 4 corners, not just 3. Moving the ethernet module over might make it possible.

I will leave it plugged into the network at home most of the time I guess, but designed the case so that I can slip a Nokia BP-4L battery into the back, for checking other places around town, etc. 3.8V is too high, so I used a diode to drop it down around 3.2-3.3V. At least this way I didn’t need to bulk it up to make space for a pair of AA’s. But using this type of battery is fiddly. It’s nice that the battery can clamp onto the battery contacts in the phone, instead of needing spring-loaded contacts. But you need something resembling the contacts in the phone. I used diode legs as battery contacts. Maybe I should make a PCB which can hold the actual surface-mount Nokia battery contacts that can be found on ebay, and have the board double as the inside plane of the battery holder, and slot into the case somehow. It could have a battery charging controller too. For now I will need to use a separate charger.

Attached pics are simply the board during assembly. Thingiverse has more pics of the case design.