My project team lead asked me to build a standby SensorGnome station, so we can immediately swap out remote units in the field that need maintenance for some reason. I sourced a new Raspberry Pi 3B+ from my local supplier, where they’d recently come back into stock, and set to work constructing it.
However, when I did an initial boot test, the RPi hung at the rainbow splash screen. I tested all the components (GPS HAT, external pushbutton and cabling, etc) with one of my original build SG stations, and they all worked; and when I built a standard RPi Debian boot image on an SD card, it booted correctly with the new RPi.
After a bit more research, I tried patching the start_ and fixup_ files from the new boot image back onto an SD card with the SensorGnome image. I managed to get the operating system to boot and command line login to come up, but I couldn’t persuade the SensorGnome software to start.
Is this a known issue with newer-build RPis, and is there a workaround for it? (I sourced a second-hand RPi from a friend who had it lying around spare and used that to build the spare station, and I’m stockpiling several more for possible future use, so it’s not anything urgent; but it would be useful to know if there’s a solution.)
Having an easily swappable SG on hand is a great idea when checking sites…
What process are you using to install the SG software? You should just be able to flash the SG image and go from there, without having to install RPi OS separately. There are instructions and links here: Initial software installation | Sensorgnome V2 User Guide
This is the “v1” software installation, unzipped from the SGPI-2018-10-12_LIWIXI.zip installation kit, not the new v2 software.
I should clarify that the spare secondhand RPi 3B+ I got hold of was an older build, not the new build, so it still works with the SGPI-2018-10-12_LIWIXI.zip install. It was only the unit I bought brand new that had a problem.
FYI, electronics distributors like Digikey often have the old models in stock, if you want to go down that route… (Not sure why you wouldn’t use the V2 software, but obviously I’m biased )
Commonality with our existing SG station array, and power consumption, mostly. All but one of our existing stations are remote installations powered by solar arrays, and the SGv2 stations use about twice as much power as the original SGs. Retrofitting them all with additional solar panels and extra batteries to support SGv2s is going to be a major project for us, and we’re only just at the point of installing and evaluating our first SGv2 station.
So at least for the time being, we need to cover various contingencies to support the project, which includes spares/ replacements etc for the existing units :).
Just to clarify, the V2 software actually has greater hardware compatibility than V1 (including the very low power Zero 2W). So you should be able to get around any power concerns. And the 3B+ you have should be able to run V2. There’s no active development on V1 anymore and though in many places it works just fine and there may be no reason to change, if you want to get all your SGs on the same platform, that might be a good enough reason to upgrade. (If you have BeagleBones in your existing array that won’t apply, though the RPi boards have been pretty affordable of late.)
Hmm, where do you get that from? I don’t know what your config is, but for most stations more than half of the power consumption is in the FCDs or RTL-SDRs, which don’t change between V1 and V2. The newer rPis may consume a bit more for the same workload, but I’m not sure that’s really the case. Do you have any power consumption comparison data?
It’s back-of-an-envelope until I can lay hands on a USB-C power meter to get the exact numbers, but my basic RPi 3B+ versions report 5V at about 0.45A and run quite happily on a standard cigarette lighter power converter with 1.0A current protection. Whereas the SGv2 built on an RPi 4B wouldn’t run at all with the same power converter, because the current protection kicked in; I went to another power converter rated for USB-3.0 QuickCharge at max 2.4A, and it barely managed to power the SGv2 (I kept getting undervolt warnings on the RPi console). Eventually I bought a single-board variable DC-DC power converter rated to 20W, and set it to 5.2V to accommodate a bit of voltage sag, and that seems to have done the trick.
tl;dr if one runs at under 0.50 A and the other trips the 1.0 A current protection, I figure “about twice as much power” roughly covered it.
The only differences between the original SG and the SGv2 were the Raspbery Pi model, and replacing the Adafruit GPS HAT with the Waveshare SIM7600G-H 4G HAT. (I couldn’t persuade the GPS on the SixFab HAT to work). They’re both using a single FunCube+ SDR.
I hadn’t actually tried running the v2 software on the RPi 3B+ yet, since I thought I’d read somewhere in the documentation that only RPi 4s were supported. I’m happy to be corrected, though! I might put one of those together on my new-build 3B+ and see what happens.
Thank you both for the followup - I didn’t mean to cause any consternation.
You are comparing whether the rPi boots using a 1A current protection. That’s a peak power measurement (depending on how fast the protection triggers). I would indeed expect an rPi4 to use more peak power than an rPi3, especially at boot time. It boots faster thus does more work per unit time which typically requires more power. However, that doesn’t mean that the average power consumption afterwards is higher.
If I understand correctly, you went from an Adafruit GPS HAT on the rPi3, which uses very little power (~10-20mA peak) to a SIM7600 cellular module on the rPi4, which itself uses over 1A peak, and uses significant power on average (I haven’t measured but would expect somewhere in the 100mA-500mA range). That has nothing to do with rpi3 vs rpi4…
You’re also using a single FCD, which itself also uses significant power. IIRC it’s 100-200mA average. This may be representative of your stations but many/most use 3x FCDs which makes the FCDs a significant portion of total avg power consumption “diluting” the difference between rpi3 and rpi4.
I look forward to your measurements with a power meter!
)