Particle Powered Air Quality Sensor Logging to Google Docs

[jaredwolff]

@fotofieber I'm using I2C over here with great success. Let me know how it goes!

[FotoFieber]

@jaredwolff
Seems like I only get an IAQ value for the BME680 and no (e)CO2. Is this correct?

[jaredwolff]

@fotofieber correct! No C02. There's also temp, humidity and pressure. Here's a little code to get up and running with their C++ library.

[FotoFieber]

@jaredwolff
Managed to compile Bosch arduino demo code on ESP32 on platformio after 3 hours. :sunglasses:

But then....

BSEC library version 1.4.7.3
Timestamp [ms], raw temperature [°C], pressure [hPa], raw relative humidity [%], gas [Ohm], IAQ, IAQ accuracy, temperature [°C], relative humidity [%], Static IAQ, CO2 equivalent, breath VOC equivalent
BSEC error code : -2

Maybe I give up here, as I really don't like to use binary code in my sensors....

[jaredwolff]

@fotofieber integrating their static library is a PITA.

What processor are you using exactly? That wold dictate which version of the BSEC you'd integrate.

For Arduino, the library should handle the import of the .a file. If you're in any other environment you'll have to add a few lines:

LIB_DEPS += $(USER_LIB_DIR)/libalgobsec.a
LDFLAGS += -Wl,--whole-archive $(USER_LIB_DIR)/libalgobsec.a -Wl,--no-whole-archive

I had to add these to the makefile in .particle/toolchains/deviceOS/1.2.0-beta.1/firmware-1.2.0-beta.1/modules/xenon/user-part for the Particle stuff to work.

[FotoFieber]

@jaredwolff
ESP32 is the platform I use.

Linking is no problem (after trial and error of two hours :) ).

I seem to have made a call to the lib:

BSEC library version 1.4.7.3

Now adding DEBUG-Messages to the BSEC arduino library.

[jaredwolff]

@fotofieber said in Particle Powered Air Quality Sensor Logging to Google Docs:

Linking is no problem (after trial and error of two hours ).

Yea it took me a bit to figure that out too. :grimacing:

Have you fired up the standard I2C library to see if you can talk to the BME680? Or maybe throw a logic analyzer on it to maker sure it's communicating? I'd imaging the compiler would barf if you used the wrong static library. Bosch does have an ESP32 specific one.

Unfortunately, I don't have any way of testing that here. Keep me posted, I'm curious as to why it's not working.

[FotoFieber]

@jaredwolff
The data is read correctly:

temp 1883
hum 54739
press 94317
gas resistance 118267
call bsec_do_steps(..
return of bsec_do_steps -2

And -2 means:

 BSEC_E_DOSTEPS_VALUELIMITS = -2,                /*!< Value of input (physical) sensor signal passed to bsec_do_steps() is not in the valid range */

I wish I had the source code of the lib....

[jaredwolff]

@fotofieber the library does have an error I found earlier. Divide _data.temperature by 100.0f and _data.humidity by 1000.0f. This is in BSEC.cpp

See below:

/**
 * @brief Read data from the BME680 and process it
 */
bool Bsec::readProcessData(int64_t currTimeNs, bsec_bme_settings_t bme680Settings)
{
	bme680Status = bme680_get_sensor_data(&_data, &_bme680);
	if (bme680Status != BME680_OK) {
		return false;
	}

	bsec_input_t inputs[BSEC_MAX_PHYSICAL_SENSOR]; // Temp, Pres, Hum & Gas
	uint8_t nInputs = 0, nOutputs = 0;

	if (_data.status & BME680_NEW_DATA_MSK) {
		if (bme680Settings.process_data & BSEC_PROCESS_TEMPERATURE) {
			inputs[nInputs].sensor_id = BSEC_INPUT_TEMPERATURE;
			inputs[nInputs].signal = _data.temperature/100.0f; // Need to divide by 100 for fp
			inputs[nInputs].time_stamp = currTimeNs;
			nInputs++;
			/* Temperature offset from the real temperature due to external heat sources */
			inputs[nInputs].sensor_id = BSEC_INPUT_HEATSOURCE;
			inputs[nInputs].signal = _tempOffset;
			inputs[nInputs].time_stamp = currTimeNs;
			nInputs++;
		}
		if (bme680Settings.process_data & BSEC_PROCESS_HUMIDITY) {
			inputs[nInputs].sensor_id = BSEC_INPUT_HUMIDITY;
			inputs[nInputs].signal = _data.humidity/1000.0f; // Need to divide by 1000 for fp;
			inputs[nInputs].time_stamp = currTimeNs;
			nInputs++;
		}
		if (bme680Settings.process_data & BSEC_PROCESS_PRESSURE) {
			inputs[nInputs].sensor_id = BSEC_INPUT_PRESSURE;
			inputs[nInputs].signal = _data.pressure;
			inputs[nInputs].time_stamp = currTimeNs;
			nInputs++;
		}
		if (bme680Settings.process_data & BSEC_PROCESS_GAS) {
			inputs[nInputs].sensor_id = BSEC_INPUT_GASRESISTOR;
			inputs[nInputs].signal = _data.gas_resistance;
			inputs[nInputs].time_stamp = currTimeNs;
			nInputs++;
		}
	}

[FotoFieber]

@jaredwolff said in Particle Powered Air Quality Sensor Logging to Google Docs:

...
inputs[nInputs].signal = _data.humidity/1000.0f; // Need to divide by 1000 for fp;
...

WTF, they have a known bug in the library and we all have to discover it by ourselves?....

How cool is that, you seem to have solved the problem I had.... :)

BSEC library version 1.4.7.3
Timestamp [ms], raw temperature [°C], pressure [hPa], raw relative humidity [%], gas [Ohm], IAQ, IAQ accuracy, temperature [°C], relative humidity [%], Static IAQ, CO2 equivalent, breath VOC equivalent
End setup()
279, 21.37, 94698.00, 57.66, 80982.00, 25.00, 0, 21.37, 57.66, 25.00, 500.00, 0.50
3278, 21.21, 94700.00, 57.42, 111997.00, 25.00, 0, 21.15, 58.10, 25.00, 500.00, 0.50
6278, 21.22, 94700.00, 57.25, 127972.00, 25.00, 0, 21.16, 57.80, 25.00, 500.00, 0.50
9278, 21.22, 94702.00, 57.10, 135341.00, 25.00, 0, 21.16, 57.58, 25.00, 500.00, 0.50

I hope I will have a graph in two days and I can compare the Bosch sensor to the NIDIR-Sensors...

[jaredwolff]

@fotofieber I'm not sure they know.. haha I should tell them.

Good luck and let me know how it goes!

[jaredwolff]

They know about it. They keep their library up to date on Github:

Here's the diff:

https://github.com/BoschSensortec/BSEC-Arduino-library/commit/e9a10fe23739bb95650d2ab89ed37764a4ba3caf

Here's the library link:

https://github.com/BoschSensortec/BSEC-Arduino-library

You probably downloaded it from the website just like I did.

I'll have to update my repo and point it to this as a submodule.

[FotoFieber]

@jaredwolff
I downloaded the binaries from the Bosch site and the source library from the github repo.

In the binaries archive is an arduino library with an example with the error, you found.... The github repo has no binaries and I couldn't get it up with the binary.

The binary distribution of essential code makes it more complicated than it could be...

The integration in platformio is not to complicated in my code. I only had to add (after investigating for some hours)

build_flags =   -Llib/BSEC/src/esp32/
                -lalgobsec

[jaredwolff]

@FotoFieber how goes the testing?

The BME680 and the SGP30 both correlate in their readings with temperature. This means when the temperature goes up and down, i've seen the TVOC reading go up and down as well. Hardly useful for a sensor that is supposed to tell you if your air is bad or not.

Take a look at the screenshot below:

You can see that both sensors react wildly to temperature. Whereas the CCS811 has a bump but not significantly. If I go by the scales in the data sheets for the BME680 and SGP30, the room that these sensors are in require immediate venting.

Even right now, the SGP30 sensor is half way to the danger zone. 😵

Danger zone!

While testing that, I also threw everything together into an enclosure. Even added GPS:

So it sits up in the kitchen and chirps at me when I burn my food. I switch on my air purifier and the particulate level goes down pretty quick!

[Nca78]

Thank you for the feedback @jaredwolff, that is very useful.
Do you confirm that you use the BSEC library for the BME680 and not direct values from the sensor ?

[jaredwolff]

@nca78 I am using the BSEC library. I’m using the compensated humidity and temperature so I’m sure if that. I may have to take a second look at the code for both sensors. I may have mucked something up!

[FotoFieber]

@jaredwolff
My experiments with VOC sensors were all not satisfactory.

The NDIR-sensors reported more or less the same as the netatmo devices and I will therefore stick with them. Apart from that, using the NDIR-sensors in arduino is much easier and you don't have to store baselines or link precompiled code.

[Nca78]

@fotofieber said in Particle Powered Air Quality Sensor Logging to Google Docs:

@jaredwolff
My experiments with VOC sensors were all not satisfactory.

The NDIR-sensors reported more or less the same as the netatmo devices and I will therefore stick with them. Apart from that, using the NDIR-sensors in arduino is much easier and you don't have to store baselines or link precompiled code.

Of course sensors measuring CO2 will give better results than sensors trying to guess the CO2 levels by measuring something else.
But the power consumption is way too high, my hope was to be able to have a battery (li-ion) powered sensor able to run the BME680 in low power settings for a few months and just use it as a warning on when you need to ventilate a room. Unfortunately it seems to not even be fit for this limited use case :(

[jaredwolff]

@nca78 I mean, you can run it on a battery. Just need a really big one! 😬

Because of the internal heater it does make it hard to integrate into a battery powered application.

I haven’t measured the low power mode on the BME. What were you getting @Nca78

[Nca78]

@jaredwolff said in Particle Powered Air Quality Sensor Logging to Google Docs:

I haven’t measured the low power mode on the BME. What were you getting @Nca78

With 5mn interval it's supposed to use only 90uA, it's good enough for a li-ion battery of a reasonable size.