Cumulus Support

Nearly two months after I ordered it, it arrived earlier this week and I put everything up yesterday. The installation went a lot quicker than I expected and was all done in less than an hour. The wind/temperature pod is around 30ft up in the air and the rain gauge is at about 8ft AGL which I appreciate is a little high but it's on a convenient post in the garden. I'm not doing this for absolute accuracy, I just want a decent approximation.

I set up CumulusMX on a Pi4 last night using the pre-built image and that was very straightforward as well.

I've not done any calibration - Should I?

The next thing I want to do is to try and embed a couple of the gauges into a Wordpress page but that's not something I've even looked at yet.

My Cumulus page (pretty much all defaults) is here.

Great!

If you have something accurate to use, I would probably calibrate - but most of us amateurs don't have!

Pi4 is way over-powered by the way - Pi Zero W is sufficient as there's no number-crunching going on!

ExperiMentor wrote: ↑Fri 13 Nov 2020 2:42 pm Pi4 is way over-powered by the way - Pi Zero W is sufficient as there's no number-crunching going on!

Oh that's good to know, thanks. I had assumed it would need a bit of power which is why I used my dev Pi4. I've just moved the SD card into a Pi Zero (not W) with an external ethernet dongle and you're right, it's working fine

Feek wrote: ↑Fri 13 Nov 2020 3:05 pm I've just moved the SD card into a Pi Zero

That's interesting! A couple of years back, I got pulled up over my 'how to install Cumulus on a Raspberry Pi' instructions for suggesting that one could build the SD card on a faster Pi then switch it to a Pi Zero. That used to work, then they'd made enough differences somewhere that once installed on one Pi version one could not switch the SD card to another. And now you're telling us it works again!

Mark Crossley (who knows more than the rest of us put together) says that a Pi Zero is noticeably slower when downloading accumulated data from a station if not been connected for a while, but otherwise copes fine. More than fine - it'll still be running at a low CPU%. My Pi Zero runs a variety of other "always on" tasks besides Cumulus - eg weather Tweet generator, uploading multiple webcams, running Pi Hole network ad blocker etc etc

ExperiMentor wrote: ↑Fri 13 Nov 2020 9:32 pmAnd now you're telling us it works again!

Yup, it certainly does and I can't say it's ever not worked for me - I've got a whole stack of Pis here doing various things (ham radio hotspots, Piholes, VPN, Unifi, DX Spider, ADS-B, internet watchdog, etc) and I've never had an issue swapping the cards around between them as long as the OS is supported. You can't build a Pi Zero with Stretch and expect it to work on a Pi4 but backwards compatibility has never been a problem.

Looking at htop on the Zero I'm using for Cumulus, it's sitting at around 10% CPU with the odd peak up to 18-20% but no higher. The web interface isn't quite so snappy as the Pi4 but that's understandable. I'm a big fan of the Zero and Zero W for single tasks and generally just add another one to the farm if I want to do something else rather than load multiple things up on the same Zero. Some things need a bit more grunt and so my ADS-B trackers are running on a 3B+ and a 4 and my Unifi controller is on a faster Pi but largely Zeros are very capable little boards.

Regarding downloading accumulated data, is that going to be relevant for me with a GW1003? I don't think that's a data logger is it, so if the Pi goes down, I'll just miss the real-time weather and it'll have a gap when it comes back, is that right?

I don't have a GW1003 or the receiver that you are accessing, but I think you are right that it has no logger (strange omission from a wonderful device).

Which reminds me of another topic for our off-topic chat. I've always advised people to set a cron job to do a daily reboot, to ensure thangs get cleared out and a restart if eg Cumulus has crashed. I've been ridiculed for that as 'not needed', but I've been rescued by that a few times. What's your take?

We're getting seriously off topic now but that's fine by me

I don't know Cumulus well enough yet to know how reliable it is and whether it needs a nightly reboot but I don't have a regular reboot set for any of mine. I would hope that Cumulus can keep running without the need for regular reboots.

I've lost track of how many I have here so I've just done a quick count. I have three Zeros, eight Zero Ws, three 2Bs, a 3A, three 3Bs (one or two of these are 3B+s) and a couple of 4Bs. That's not including the two or three spare Pi Zero/Zero Ws I've got in drawers, waiting for me to think of things to do with them.

Generally I'll go through them every so often and do OS updates and reboot them but it's not a priority really. A couple of them have an uptime of well over a year.

Many of them run a full dd backup of themselves once a week to my NAS and that's one thing that the Zeros are particularly slow at. My Pi4 tracker completes in about half an hour whereas the Zeros can take a couple of hours.

Feek wrote: ↑Fri 13 Nov 2020 11:33 pm I don't know Cumulus well enough yet to know how reliable it is and whether it needs a nightly reboot but I don't have a regular reboot set for any of mine. I would hope that Cumulus can keep running without the need for regular reboots.

Prior to doing regular updates of my prod instance of Cumulus for dev testing (!) I regularly ran my rPi and Cumulus for 6 months or more without rebooting, and then it was for update purposes not because anything had failed.

Definitely calibrate your station. You are at about 23 meters of elevation. That corresponds to about 2.8 hPa difference between your Absolute and Relative pressure.
Therefore to start off with I recommend that you set your Relative Offset to 2.8 hPa and you do this using the WS View mobile app.

If you look at your local METAR you are off by about that amount which I've mentioned.... 2.8 hPa.
https://en.allmetsat.com/metar-taf/unit ... ?icao=EGMC
https://aviationweather.gov/metar/data? ... e=&hours=0

How did I figure this out?
1. Got your coordinates from your website
2. Converted your DMS coordinates to Decimal degrees: https://www.directionsmag.com/site/latlong-converter/
3. Entered Decimal degrees coordinates into this site to get your elevation: https://www.freemaptools.com/elevation-finder.htm (shows 22 m)
4. Added a meter to the ground elevation to account for height of your barometric sensor (an estimate which you can correct but I'm going with 23 m)
5. Entered 23 m into this online barometric calculator - DO NOT CHANGE TEMPERATURE NOR STANDARD PRESSURE...LEAVE AT 15°C and 1013.25:
https://keisan.casio.com/exec/system/1224579725
6. Subtracted standard pressure 1013.25 from the calculator results using your elevation: 1013.25 - 1010.49 = 2.76 (which I round to 2.8 hPa)
7. Compared your current pressure to observed local METAR. Determined you are running a virgin system that is sending out Absolute pressure...a not calibrated system. You are off by the exact amount I expected.

That is how you calibrate. Now if you still need to fine tune your barometric pressure after entering in 2.8 hPa for the Relative Offset you would then make those corrections to the Absolute Offset. But it seems to me like you will not have to do this extra step.

galfert wrote: ↑Mon 16 Nov 2020 8:23 pmThat is how you calibrate. Now if you still need to fine tune your barometric pressure after entering in 2.8 hPa for the Relative Offset you would then make those corrections to the Absolute Offset. But it seems to me like you will not have to do this extra step.

That's great, thank you. I've just done exactly as you've suggested.

/edit - I've got a barometer app on my phone and it's currently reporting sea level pressure at 1016.94 and station pressure at 1013.46.

When I look at Cumulus, it's showing 1016.7, with the calibration, would I expect that to match what the barometer app is showing?

Within WS View it's showing Absolute Pressure at 1013.8 and Relative at 1016.6.

I have restarted the Cumulus service.

Feek wrote: ↑Mon 16 Nov 2020 9:30 pm

galfert wrote: ↑Mon 16 Nov 2020 8:23 pmThat is how you calibrate. Now if you still need to fine tune your barometric pressure after entering in 2.8 hPa for the Relative Offset you would then make those corrections to the Absolute Offset. But it seems to me like you will not have to do this extra step.

That's great, thank you. I've just done exactly as you've suggested.

/edit - I've got a barometer app on my phone and it's currently reporting sea level pressure at 1016.94 and station pressure at 1013.46.

When I look at Cumulus, it's showing 1016.7, with the calibration, would I expect that to match what the barometer app is showing?

I've not explored phones and their barometric accuracy. I'd say though that you are in the ball park. I wouldn't expect it to match perfectly. Barometric pressure is constantly changing and different barometric sensors take readings at different intervals. Also we don't know if with these multiple interval readings if you are seeing instant pressure or rather a running average of the past couple minutes.

Within WS View it's showing Absolute Pressure at 1013.8 and Relative at 1016.6.

Looks good. Here is the big take away. You are at 23 meters of elevation to your barometric sensor (where the WH32B is - indoor sensor). That elevation corresponds to 2.8 hPa of barometric pressure difference from sea level. This difference will never change. When you enter 2.8 hPa for Relative Offset into your station what you are in fact doing is telling the station that you are at 23 meters. Therefore the difference between Absolute and Relative will always be 2.8 hPa. That is why I said that if you still feel that you need to do some minor adjustments then that is done by adjusting the Absolute from its factory 0.0 hPa offset calibration.

Your phone thinks you are are at 28.92 meters of elevation.
1016.94 - 1013.46 = 3.48 hPa (this corresponds to 28.92 meters from the online calculations below)
https://keisan.casio.com/exec/system/1224585971
It doesn't seem as though the phone is that accurate. It is probably going off GPS elevation to then calculate what the sea level pressure is from the raw absolute sensor pressure. And we know GPS for elevation is not that accurate.

Yeah, GPS altitude is a bit hit and miss. It was even worse before the fiddle factor was switched off in the dim and distant past.

As I'm a radio ham, it's important to know my exact height ASL and I know from close examination of maps and contour lines, years ago that we're at almost exactly 22m ASL here. The sensor is just above floor level in the upstairs back bedroom which I've measured at 2.7m above the ground.

So that's 24.7m ASL. Slapping that in gives me a correction of 2.96 with gets rounded up to 3 as opposed to 2.8 so I'll go with that.

Sounds good. Looks like you got it.