Solar-powered weather station

Thanks for the feedback Paul, I'll have to have a serious look at Grafana and see how I can incorporate it into the Club's future plans and projects.

This is an example of my node-RED powered Grafana dashboard, using 5 x 5 second feeds (running on a Pi).

You Tube degrades the video quality, but this is a static screenshot -

1 Like

OK Paul you have got me 'hooked' - that looks brilliant.
I really like the way you can zoom into a certain timeframe and expand the plot.

How long did it take you to create it?

Thanks for sharing your work - I really do appreciate it.

I'm not very familiar with databases, queries etc, so it was initially a steep learning curve, with lots of googling over many nights. Once I got my head around the concept, it all seemed to fall into place pretty quickly.

The most important lesson I learnt is to make sure you have your data correctly formatted in influxdb before even looking at Grafana. If your data is wrongly stored, it makes retrieving & charting much more difficult.

Yes, I believe that Grafana does the "right" thing and automatically downsamples the displayed points according to the chart size and resolution.

While Dave (the other dynamic one :slight_smile:) has done what he can to make the Dashboard chart a little smarter I think that he may have hit the limits on what it can do.

Of course, Grafana is a server in its own right and gets lots of its own development - I suspect that it gets a fair bit of money from enterprise stuff too. It has also been doing chart-based dashboards for a long time. So not surprising that it can outperform Dashboard.

With Grafana, you can also improve things further by specifying the minimum time interval allowed. For example, with my Grafana dashboard, I generally get sensor updates about every 50sec or so. So I set my minimum time interval to 1 minute for detailed charts.

But I only keep per minute data for a week. For anything further than that, I summarise the data to hourly and for those charts I set the minimum time interval to 60min



Another graphing possibility is InfluxDB's Chronograf. While not as advanced as Grafana it does integrate easily with the InfluxDB database and does not have a steep learning curve.

1 Like

I had problems between Grafana and Influxdb because of formatting issues whilst storing data to Influxdb. I was put on the right track by changing the way the data was stored and then Grafana 'just worked' and is very easy to add graphs, charts, dials, tables etc.
Here is a snapshot of mine which I was suprised to find out, stored the data so anyone can have a play.

Have a look

1 Like

Hi Martin @Cathprotech
Woo - your dashboard is impressive.
I must seriously consider introducing Influxdb and Grafana into my IoT classes.

Thanks for sharing your work.

Thanks David.

My biggest problem was writing the data to Influxdb in a format that could be read by Grafana.
Once that was done, honestly Grafana is so easy. What I have done is very simple yet the results are impressive.
Imagine what I could do if I actually knew what I was doing.
My Influxdb problem was solved by adding 'topics' to my data, it was quite simple but lots of asking questions to manage to get it right, once the format was correct, Grafana just reads the data and works fantastic.
Did you select the date ranges on my data and zoom in/out. That is just using a snapshot of published data!


@dynamicdave do your BME280's read an accurate temperature ? I have a couple, but they are all off by 5 degrees Celsius (it appears to be exact 5) I have no idea how this is possible.

Pardon me for butting in, but If you are talking "chinese" bme280s then I believe there are two (perhaps more) "types" around. One is a bmp280, but still labelled (in many cases) as a bme280. One includes humidity, and I think may be more accurate. Worth checking.

The ones I'm using are labelled BME/BMP280 - they sense temperature,humidity and pressure.
Temperature seems fairly accurate (within +- 0.5C) for what I need.
I've checked it against a digital thermometer over a 3hr period.
The sensor connects via I2C to a Wemos D1 Mini running ESP Easy and one thing I did do was to set the altitude in the 'device driver'.

BMP280 vs BME280 difference is the humidity sensor as far as I understand. I can read the temp/humidity/pressure and except for the temperature it seems to match the measurements of my other devices.

@dynamicdave does the altitude make a difference for temperature readings ? I thought that was for the pressure only, i could be wrong.

I think you're right about the altitude and the pressure reading.
I was just saying that was maybe a difference - I'm clutching at straws!!

Quite possibly, although I have read reports of various inaccuracies with them. Perhaps its their reliability that is the problem. It also doesn't help matters that some have been "mislabeled". Lucky they are cheap :wink:

I'm going to have a go at making one of these to shield the sensor from direct sunlight.
They use a DS18B20, but I'm sure I can adapt it to accommodate the BME/BMP.

PS: A set of metal-cased DS18B20s arrived this morning - you know what my next job will be?

The DS18.. is still one of my favs in terms of accuracy and precision.

I have a similar problem with the BME280 (The one with humidity). The temperature is about 3 + degrees higher than it should be, especially when temperatures are above 18 degrees C.
But this doesn't appear to be anything to do with being cheap Chinese apparently.

The BME280 from Bosch is prone to 'Self Heating', it is documented in the spec sheet, the BMP's are apparently not so it could be related to the humidity part of the sensor. (Well, it doesn't say so in the spec sheet for the BMP's anyway).

From the data sheet....

"This temperature value depends on the PCB temperature, sensor element self-heating and ambient temperature and is typically above ambient temperature. "

I find this very strange as it is supposed to be more accurate, but it is in the spec. The DS18B20's are more accurate for temperature but with larger steps (lower resolution) and are very cheap. In future, as I expand my monitor, I will use these for temperature rather than the BME.

What is unclear is whether it effects the humidity readings or not as humidity uses temperature in its calculations which is why most (if not all) the humidity sensors also do temperature.

I have several alternative sensors for humidity and out of 5 different type/models of humidity sensor, no two gave the same result for humidity yet all gave very similar temperature readings with the exception of the BME280.

I have mine powered on continuously, so I wonder if it was powered down between readings or placed in standby if then it would not suffer the self-heating and improve accuracy!

I also read in case they are used with an esp or something, keep them on distance from the board, apperently the wifi chip interferes as well.

The key purpose of the temperature sensor is for temperature compensation of the humidity and pressure sensors, so the important thing is that it measures the temperature of the device accurately. For the compensation it is not important if it does not measure ambient temperature accurately. So the temperature reading is probably accurate and to spec, it is just not measuring what you want to measure.