Hi,
is the app compatible to win11?
I can't get a conection via USB. It opens the port .... checking for controller .... then disconection and the ESP32 freezes (no changes anymore on the oled).
Nearly the same behavior via accesspoint (.... checking controller ... disconnection, but no freeze.
Other issues:
Temp is always 20, ambient 50 and pressure 11000.00
Due to missing connection to the PC I can't sent you a debug-log.
Its only a wired setup (picture)
Input is from regulated professional power supply. R1, R2 and C6 added.
yellow = SCL
blue = SDA
orange +3,3V
red +5V
black = 0V
Other sensors left "commented" in the controllerconfig.h
What can I do?
Hi Ralf,
the values you see are the default values if no real values can be read from the sensor.
If you couln'd fix the problem meanwhile you should use the test-sketch that comes with the FW packet and observe the messages in the IDE monitor.
I would say there is a problem with the wrong I²C adress or a problem with the correct cabling.
I had to solder the adress pads on my Adafruit BME280 module.
The correct connectors are SCK = SCL and SDI = SDA
Regards
Hardy
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Merry X-mas, thanks for the tipps.
- I2C Scanner found 5 devices on 0x0 (???), 0x29 (TSL2591), 0x3C (Oled), 0x5A (MLX90614) and 0x76 (BME280).
- BME280 test: nothing on Serial Monitor
- MLX90614 test: frezze on "start: task timer"
- RTC on ESP32 test: not working, allways 01/01/1900, 00:00:00
- TSL2591 test: values changing, but not due to the light
- OLED display is working normally
The connection is so simple: I cant find a mistake:
- Pin D22 to SCL, yellow wire (+10k pullup to 3,3V)
- Pin D21 to SDA, blue wire (+10k pullup to 3,3V)
Removed all sensors except TSL2591, running the test. Again only bullshit.
Very frustrating. ..
trying to debug everything at once is not a recommended course of action.
Take a step back.
The I2c scanner has found some devices, so they must be connected correctly and they must be responding else i2c scanner would not know they are there.
try these retired tests
One sensor at a time - when finsihed disconnect it and put on the next sensor, one at a time
The I2C scanner found all (not some) connected devices, if that means that they must be connected correctly, than I am a step further.
But even only with connected light sensor and display, its not working. There but be a fundamental mistake. Did you have a look on the update screenshoot?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
One thing at a time. One device at a time.
use the tests that I attached (older-tests.zip)
I saw the photo you posted. Those resistors are pretty close to other pins, are you sure they are not touching a pin they should be?
I also see that you have the oled display working.
is the app compatible to win11?
yes
In past releases there used to be a mode called LOCALSERIAL. That is no longer supported.
You have to use TCPIP once the controller is running.
"Nearly the same behavior via accesspoint (.... checking controller ... disconnection, but no freeze."
The basic bare bones, is the controller itself.
Set the mode to ACCESSPOINT with everything else disabled (use //) except TSL2591 in controllerconfig.h
The reprogram the esp32
Upload the data skectch files (the esp32 will not behave unless this is done)
Wait 10s to allow the controller to startup, then
on the computer, connect to the wifi network mysqmplus
Once connected, open a web browser and enter the location
192.168.4.1/admin1
and you should see the first admin page.
Avoid opening multiple tabs, the controller only supports a number of connects at the same time.
Click on the next admin page button at the bottom of the page.
Wait for that to load, each page might take 3-4 seconds to load.
Once loaded, try the next page, and so on.
After viewing all the web pages, try
192.168.4.1
to view the home page. THERE IS NO NEED to click fresh or F5 on the home page, the values update themselves. You might not see any change because there are no sensors. But at least you should have a page.
have you installed all the correct required libraries
myAsyncTSL2591
mybme280e
myMLX90614AF
myOLED
If no, install them
Are you running Arduino 1.8.16 (NOT the web based version)
If no then update
Linux or Windows?
When connecting or removing a device, POWER must be off (12V and USB) - in other words the USB cable has to be disconnected first.
Some sensors require 3V3, others 5V and any 5V sensor will not work if powered from 3V3
So let us start with the TSL2591 sensor. It is a 3V3 sensor and if connected to 5V then you will likely damage the sensor
With power off, and cable disconnected, then connect the TSL2591 to the esp32.
Program the controller with the I2C scanner program and it should find the device and list its address as 0x29
If the I2c scanner does find it then good, but if not, you are going to have to figure out why. That involves checking your 4 wire connections again.
IF and ONLY IF the sensor is found,
Start arduino IDE
Load the TEST_TSL2591,ino
Open the serial port monitor
Set the serial monitor to 115200
program the controller with TEST_TSL2591
Watch what the serial monitor program displays
IGNORE - TSL2591 Sensor fail
TSL2591 ready: read lux
TSL2591 sensor ready, lux = some value
If you get this far, all well and good.
Get a small hand towel (dry) to block out most light to the sensor.
The reading should drop to a low value
Remove the hand towel
The reading should increase and stablize
Get a hand torch and about 1-2 feet way, turn the torch on,
shine the beam at the sensor surface and
the reading should further increase,
The program reads the sensor every 2s, so rapid movements of the torch etc will be missed becuase it is reading the sensor value - readinds are 2s apart
for 10s shine the torch at the sensor
The next 10s turn the torch off
For the next 10s cover the sensor with the hand towel
You should now have 3 different readings that are somewhat consistent
You can run these tests 2-3 times to verify the results (it does not matter if there is a small variation between results) - what you are expecting is
lowest value = sensor blocked from light by hand towel
mid value = sensor not blocked
highest value = torch on and beam shining on the surface of the sensor.
let me know how that goes.
I am not trying to teach you how to suck eggs, I am just concentrating on a strategy to verify the operation or failure of the sensor.
Regards
Robert
Last edit: brownrb 2021-12-26
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Maybe you should upload here your "controllerconfig.h". Remove personal info like passwords.
So we could check if you made the correct settings. Maybe you have overlooked something.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Robert,
Thanks for your detailed answer. You give me plenty of homework.
some details in advance: "I saw the photo you posted. Those resistors are pretty close to other pins, are you sure they are not touching a pin they should be?"
Yes, I am very sure
"I also see that you have the oled display working."
Yes, it's working on the same I2C as the (not working) sensors.
"So let us start with the TSL2591 sensor. It is a 3V3 sensor and if connected to 5V then you will likely damage the sensor"
If you check your circuit diagram, the TSL2591 is conected to J9 which is powered with 5 V, but this doesn't matter, because it have an own power regulator.
"Once connected, open a web browser and enter the location
192.168.4.1/admin1
and you should see the first admin page."
I see all the pages, but if I change the numbers e.g. coordinates, I can't set them.
"After viewing all the web pages, try
192.168.4.1
to view the home page"
I see values changing somewhat with the light, but after 3-5 minutes the ESP32 hangs it up and I loose conection.
*"have you installed all the correct required libraries
myAsyncTSL2591
mybme280e
myMLX90614AF
myOLED"
**I copied them into the library directory, while compilining I get no error message, but they don't appeare in the library manager. should they?
So because the light sensor somehow works, I reconected the BME280 and MLX90614
If not conected I get Hum 50, Amb. 20, Pressure 11000
Conected sensor I get Hum 0, Amb, nan, Pressure nan. It is recognized (see also I2C scanner) but I get no values. I changed the sensor (I mostly buy more than one), but the same.
mySQM+ Admin10
SENSOR STATUS
BME280 Found
GPS Not defined in firmware
MLX90614 Found
...
...
TSL2591 Lux Found
It's 2:30 AM local time, I have to go to bed.
Ralf
I would like to take this time to give a little more information that is not in the pdf or mentioned in the test programs.
The lux sensor has two registers that contain count values that represent the lux value.
To manipulate this the sensor uses two settings, gain and intergration time
The higher the gain setting (there are from memory 5 gain settings) the greater the ability to read low light values, the lower the gain then the brighter the light
The higher the integration time (how long we let the light fall on the sensor before it is read) is used for low light values, smaller integration times for bright days
Low Gain, Low Integration = very bright day
High Gain, High (long) integration times = very dark days/nights
The firmware uses an algorithm that senses the lux level and checks if any of these registers approx maximum value, and if so, adjusts the gain and intergration time accordingly to go to the next optimal gain/integration setting.
Conversely the firmware also looks to see if the register values are getting too low, meaning an increase in gain and/or integration time will be necessary.
For example, if the light was increaseing in intensity, then we reach a point where the registers overflow and we need to lower the gain and the integration time. Converse for light decreasing.
All of this takes time, adjust gain.time, check if registers are still full, adjust again etc
In addition to prevent sudden blips (a lightning strike - or light change that is transient in nature) the firmware also uses an averaging system to handle these outliers.
There is a lag between changing light levels and the sensor reporting appropriate lux levels (because of the adjustments to set an optimal gain/time setting and the averaging of lux values that is done).
In normal operation, the sun comes up over the horizon slowly, it just does not appear full in an instance, same with evening. These changes take minutes or hours to happen,
So, the lux plots using mysqmweb etc look very good because of what the firmware does. There are also some additional things the firmware does with the lux sensor that I won't go into right now.
The task timer controls the reading of lux values which about 10 sesonds apart.
When the firmware starts from a cold boot or reset, it is going to take some time before the sensor is fully configured for the current light. This can take up to 30s as the firmware checks the register values and adjusts the gain and integration settings to find an optimal setting. So after a boot, you see lux as 1000 (just a start value) then stay there for around 30+ seconds before it is changed to another value.
If the light source falling on the sensor changes - remember outside it does not change rapidly - by human hand by using a torch or cloth to alter the amount of light falling on the sensor, the lux setting will not change UNTIL the firmware has established the new register settings gain/time for the light. Again around 15 seconds required for this to happen
I point out again that during normal operation during the day/night we do not have these rapid sudden changes of light intensity. And when we plot the graph, it looks nice and smooth, but we are seeing hours of change. Imagine if one plotted a graph of instaneous samples over 20s - I leave that one for you to think about. (Attached is a sample graph of lux over 3 hours)
The problem with the test is this, you make a change, you think the next reading should be changed, it is not, you end up thinking it does not work, etc, etc
The controller remember is readjusting the gain/time settings to find a combination of gain/integration settings that gives a good range to measure the new light level, and that does take a heap amount of time. In daylight we might wait 12+ hours for the sun to go down, but we are trying to simulate that in an instant.
All of the test programs are under revision now, because firmware 125-xxx has a significant change in the way the controller operates. Which is why I posted the old tests.
In hindsight, the choice of using the tsl2591 for my example in previous posts was an error on my part. I apologize for that. A lot of what I just wrote I had forgotten, it all came back looking at the code today and running the test myself. So again, my bad.
I found it takes around 1m after boot to stabilize, when lux value of 1000 changes to a new value (the controller is busy adjusting gain/integration settings)
I found it takes 15-20s for a new value to appear after changing the light intensity (meaning if using a torch you have to keep it in the same place shining on the sensor for 30+ seconds), and when turning the torch off it takes 20-30s to stabilize again
thanks a lot for that explanation! I was also always a little bit surprised that alway the LUX value started with 1000 as well as it needs some time until a stable value is reached.
I placed my setup (just connencted via some cables) outside for a night and could see very tiny changes when sun raises or drops and it seems that every second of more or less light was notified. Also clouds with lower and higher density can be detected with this sensor. So from my point of view this sensor works better that expected.
Referring to Ralf´s problem I have some little ideas.
1. WHat I can see in his picture is that he uses a Light sensor from Adafruit as I do. This sensor has build in pull up resistors for the I²C bus. If you will have a look into the Adafruit PDF document of this device you will see that they do not use additional external resistors in their examples. I don't know what will happen if you spent too many of them in parallel? The other boards do have them not I think but maybe since at least it is now only 5kOhm this may have impact. So I would remove the external resitors just tocheck the bahaviour.
2. To be frank I have (only) 2 modules at the moment and both are from Adafruit and I use in addition external resistors additionally and it works. BUT I do not have any other module which maybe will be affected.
I am waiting for the UV sensor which is not from Adafruit and I will see if this has impact. In this case I would remove the external ones.
Interesting thing is, that every Adafruit module has these resistors but they do not tell us to limit the amount of modules on the bus. As can be seen in the Adafruit PDF there is also a FET transistor added to each I²C bus and I do not know the reason....? A HW expert can tell us.
3. The Adafruit PDF also includes how to test their modules with their standard libraries. SO I would use them in the Arduino IDE instead of the modified from RObert, just to validate the behaviour!
One additional question to Ralf. Did you install also the "other libraries"?
Also to Ralf, when I changed from the prevoius FW to the latest I removed the old Windows programm and installed the new one. Also I installed ALL libraries from the new package as well as the new sketch. I am using the Firfox browser and I'm doing all test in station mode using my WLan network.
Regards
Hardy
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Robert,
Thanks for your detailed answer. You give me plenty of homework.
some details in advance: "I saw the photo you posted. Those resistors are pretty close to other pins, are you sure they are not touching a pin they should be?"
Yes, I am very sure
"I also see that you have the oled display working."
Yes, it's working on the same I2C as the (not working) sensors.
"So let us start with the TSL2591 sensor. It is a 3V3 sensor and if connected to 5V then you will likely damage the sensor"
If you check your circuit diagram, the TSL2591 is conected to J9 which is powered with 5 V, but this doesn't matter, because it have an own power regulator.
"Once connected, open a web browser and enter the location
192.168.4.1/admin1
and you should see the first admin page."
I see all the pages, but if I change the numbers e.g. coordinates, I can't set them.
"After viewing all the web pages, try
192.168.4.1
to view the home page"
I see values changing somewhat with the light, but after 3-5 minutes the ESP32 hangs it up and I loose conection.
*"have you installed all the correct required libraries
myAsyncTSL2591
mybme280e
myMLX90614AF
myOLED"
**I copied them into the library directory, while compilining I get no error message, but they don't appeare in the library manager. should they?
So because the light sensor somehow works, I reconected the BME280 and MLX90614
If not conected I get Hum 50, Amb. 20, Pressure 11000
Conected sensor I get Hum 0, Amb, nan, Pressure nan. It is recognized (see also I2C scanner) but I get no values. I changed the sensor (I mostly buy more than one), but the same.
mySQM+ Admin10
SENSOR STATUS
BME280 Found
GPS Not defined in firmware
MLX90614 Found
...
...
TSL2591 Lux Found
It's 2:30 AM local time, I have to go to bed.
Ralf
Hi Hardy and Robert,
first of all many thanks for all the effort to help me!
Thanks for detailled explanation about working the light sensor...
Due to that, I think the light sensor is working:
it starts at 1000
after a while its now about 76.00000 (SQM 7.8....) indoor, no extra light
about 30000.00000 (SQM 1,7...) with torch
about 1.80000 (SQM 11,9...) covered by paper.
With the knowlege of the delay, the messurement is consitant.
So for me the light sensor is working, thats a big step foreward!
So I conected the IR and BME280, they are NOT working (recognized but no values). Maybe the reason is the parallel pull up resistor as Hardy mentioned. I will check this after lunch!
I am not very experienced with arduino, so maybe a silly question about "installing a library":
Up to now I "installed" all libraries I get a error (missing message) by checking the sketch.
- ArduinoJson and ESP32Time: I used the library manager
- For myOLED, myAsyncTSL2591, mybme280e and myMLX960614AF I just copied the files into the library directory until I got none error message, as a requirement to upload the sketch. But these libraries do not appeare in the library manager in the "installed library section". Is that normal?
I am surprised, that I can't find the onewire library neither in the directory nor in the library
manager (I was sure, it is already installed from other sketches, but due to the new Win11 PC it was not). There was also no need to install OneWire by checking the sketch...
(UPDATE: the installation of Onewire didn't change anything)
Ralf
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I did copy the zipped lib files (optionals and requiered) to the libraries folder and unzipped them. So each one gots a folder. Thats al I did. See screenshot.
To be frank, when compiling the sketch and you get no message about a missing library do did it in the right way (i think).
All libs with modified folder-names from Robert do not appear in my librray manager too.
Thats all for now. Good luck by experimenting with the resistors (have attached the PDF of the Adafruit unit I have seen on your screenshot. Look at the schematic where you will see the onboard resitors a dn the FET Transistors in the I²C bus were I do not know how they should work).
I do not know if any of the others are also from Adafruit. The design of the Stemma units are easy to identify.
So I conected the IR and BME280, they are NOT working (recognized but no values). Maybe the reason is the parallel pull up resistor as Hardy mentioned. I will check this after lunch!
unlikely. Should not make much difference, the more devices that u plug into the I2C bus the lower the total value of all pull-up resistors need to be. hence it is common practise to include resistors on each I2C sensor. without going into the maths, it is fine.
Note: There are things you need on the I2C bus when adding more sensors. It is possible for one sensor to be a little noisy and that causes issues with the signals SCL and SDA, so that is why the schematic has a number of capacitors across the data lines and VCC - c4, c5 and c6 should be fitted in place, else you may have issues once all the sensors are fitted,
But these libraries do not appeare in the library manager in the "installed library section". Is that normal?
yes, but you can also load those files using the add zip of arduino IDE, instead of copying the folders, not an issue though
I am surprised, that I can't find the onewire library neither in the directory nor in the library
manager
correct, that is a special part of arduino which is installed when running the Arduino IDE setup program, it is stored along with the IDE in C:\Program Files (x86)\Arduino i think from memory the libs folder
To test the BME280 use the attached - I tested this today and it is working fine for me, there is also a sample output text file to see what I got when I ran the program.
Please program the controller with this and only the bme280 sensor attached (remove all the other sensors)
the test files are ready to go, no changes necessary
Let me know what you get displayed in the serial monitor window
Hi Hardy and Robert,
first of all many thanks for all the effort to help me!
Thanks for detailled explanation about working the light sensor...
Due to that, I think the light sensor is working:
it starts at 1000
after a while its now about 76.00000 (SQM 7.8....) indoor, no extra light
about 30000.00000 (SQM 1,7...) with torch
about 1.80000 (SQM 11,9...) covered by paper.
With the knowlege of the delay, the messurement is consitant.
So for me the light sensor is working, thats a big step foreward!
So I conected the IR and BME280, they are NOT working (recognized but no values). Maybe the reason is the parallel pull up resistor as Hardy mentioned. I will check this after lunch!
I am not very experienced with arduino, so maybe a silly question about "installing a library":
Up to now I "installed" all libraries I get a error (missing message) by checking the sketch.
- ArduinoJson and ESP32Time: I used the library manager
- For myOLED, myAsyncTSL2591, mybme280e and myMLX960614AF I just copied the files into the library directory until I got none error message, as a requirement to upload the sketch. But these libraries do not appeare in the library manager in the "installed library section". Is that normal?
I am surprised, that I can't find the onewire library neither in the directory nor in the library
manager (I was sure, it is already installed from other sketches, but due to the new Win11 PC it was not). There was also no need to install OneWire by checking the sketch...
(UPDATE: the installation of Onewire didn't change anything)
Ralf
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
You are right, the pull up resistors dont make a difference.
Added all the captators, only BME280, running test_bme280, I get from serial monitor: see scrennshots. I used my 2nd BME280, the same. It's unlikely that both, new sensors are not working?
it gets even more mysterious:
- I2C scanner founds the BME280 at address 0x76, see screenshot
- loaded the test example which comes with the adafruit bme280 library, there I get Sensor ID is 0x0, could not find a valid sensor, check wiring, address, sensor ID...
- check wiring: there are only 2 wires SCL to D22 and SDA to D21, thats not very difficult....
- address is right 0x76
- what's about sensor ID???
I2C working for the light sensor and Oled, but not for the others sensors, how can this be?
(At the beginning I thought this is because of a wrong i2c address, so I changed the address (cut+soldier), but then I got the default values 50/20/11.000.)
Maybe the BME280 is in fact a BMP280? Google found some hints, that this can cause the indications. Ordered an other BME280....
Meanwhile I will now test the IR sensor
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Tested MLX alone:
- test from mySQM: stopps after "start task timer"
- test from adafruit: Its working, if I hold my hand over the sensor, object temp rises up to 31°C
- older test from mySQM is working as well!
The BME280 is recognized by I2C_scanner, the fact that is recognized means it is wired correctly.
Please do not try the Adafruit code, it is a waste of time and does not prove much at all. It just loads up all your libraries with stuff you do not need - I might write more about this later, but for now
First Delete the mybme280e folder that you copied into documents\arduino\libraries
then
Download the attached zip file checkbme280sensor
unzip the file
double click the ino file to load it into the Arduino IDE
Open the serial port monitor
program the esp32 with the code by clicking the ->
let it run for around 10s
Copy the text from the serial monitor window and post the results
I have includes the output that I got when I ran this a few minutes ago
regards
Robert
Good morni ng Robert,
only the BMP280 connected to the board,
deleted the mybme280e library,
programmed checkbme280sensor.ini onto the esp32,
It started first with "reading....", I made a hardware reset.
It founds a sensor, its not a BME280, because its a BMP280, as I told you (If you look exactly you can see that my sensor is more a rectangle than a square).
I think (hope) that's the reason why it doen't work. I awaite the real BME280 in some days.
Meanwhile I tested the MXL IR sensor, see post 18h ago: there ist only the older sensortest working.
regards
Ralf
Hi
the BMP280 is not compatible. It will not work. It would require a lot of code changes. makes no sense to do that. So you will need to wait till your bme280 sensor arrives.
As the ouput says, it is not a bme280 sensor, hence no values.
In my mind there was a possibility of either a library fail or sensor fail.
In the test program I send I added the sensor ID test in the code as well as placing the library files directly with the ino file. This would overcome any library install error, or identify a different sensor, so the library files worked, indicating no issue with the librry you installed, and identified that it was not a bme280.
Re mlx
Its past midnight in NZ right now. So I am winding down now.
I will put together a special mlx test for you to try (later my time). So in the meantime, have a rest?
Regards
Robert
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
is the app compatible to win11?
I can't get a conection via USB. It opens the port .... checking for controller .... then disconection and the ESP32 freezes (no changes anymore on the oled).
Nearly the same behavior via accesspoint (.... checking controller ... disconnection, but no freeze.
Other issues:
Temp is always 20, ambient 50 and pressure 11000.00
Due to missing connection to the PC I can't sent you a debug-log.
Its only a wired setup (picture)
Input is from regulated professional power supply. R1, R2 and C6 added.
yellow = SCL
blue = SDA
orange +3,3V
red +5V
black = 0V
Other sensors left "commented" in the controllerconfig.h
What can I do?
Thank you
Ralf
Hi Ralf,
the values you see are the default values if no real values can be read from the sensor.
If you couln'd fix the problem meanwhile you should use the test-sketch that comes with the FW packet and observe the messages in the IDE monitor.
I would say there is a problem with the wrong I²C adress or a problem with the correct cabling.
I had to solder the adress pads on my Adafruit BME280 module.
The correct connectors are SCK = SCL and SDI = SDA
Regards
Hardy
Bear in mind that LOCALSERIAL is now deprecated. latest versions of apps no longer support serial mode.
The serial port is now primarily used for debugging only. As of firmware 125-65 (in testing as I write this).
Hardy gave very good advice to use the test programs first, run each test program to verify that the parts and connections work.
That way, when you finally install the firmware there will be few suprises.
You cannot load the firmware first and then try to test 100 things at once because it did not work. Divide and conquer.
Yes, all the apps run under Win11. The versions currently available on SF were developed on a Win11 system and tested on Win11 before release.
regards
Robert
Merry X-mas, thanks for the tipps.
- I2C Scanner found 5 devices on 0x0 (???), 0x29 (TSL2591), 0x3C (Oled), 0x5A (MLX90614) and 0x76 (BME280).
- BME280 test: nothing on Serial Monitor
- MLX90614 test: frezze on "start: task timer"
- RTC on ESP32 test: not working, allways 01/01/1900, 00:00:00
- TSL2591 test: values changing, but not due to the light
- OLED display is working normally
The connection is so simple: I cant find a mistake:
- Pin D22 to SCL, yellow wire (+10k pullup to 3,3V)
- Pin D21 to SDA, blue wire (+10k pullup to 3,3V)
Removed all sensors except TSL2591, running the test. Again only bullshit.
Very frustrating. ..
Ralf
trying to debug everything at once is not a recommended course of action.
Take a step back.
The I2c scanner has found some devices, so they must be connected correctly and they must be responding else i2c scanner would not know they are there.
try these retired tests
One sensor at a time - when finsihed disconnect it and put on the next sensor, one at a time
The I2C scanner found all (not some) connected devices, if that means that they must be connected correctly, than I am a step further.
But even only with connected light sensor and display, its not working. There but be a fundamental mistake. Did you have a look on the update screenshoot?
One thing at a time. One device at a time.
use the tests that I attached (older-tests.zip)
I saw the photo you posted. Those resistors are pretty close to other pins, are you sure they are not touching a pin they should be?
I also see that you have the oled display working.
is the app compatible to win11?
yes
In past releases there used to be a mode called LOCALSERIAL. That is no longer supported.
You have to use TCPIP once the controller is running.
"Nearly the same behavior via accesspoint (.... checking controller ... disconnection, but no freeze."
The basic bare bones, is the controller itself.
Set the mode to ACCESSPOINT with everything else disabled (use //) except TSL2591 in controllerconfig.h
The reprogram the esp32
Upload the data skectch files (the esp32 will not behave unless this is done)
Wait 10s to allow the controller to startup, then
on the computer, connect to the wifi network mysqmplus
Once connected, open a web browser and enter the location
192.168.4.1/admin1
and you should see the first admin page.
Avoid opening multiple tabs, the controller only supports a number of connects at the same time.
Click on the next admin page button at the bottom of the page.
Wait for that to load, each page might take 3-4 seconds to load.
Once loaded, try the next page, and so on.
After viewing all the web pages, try
192.168.4.1
to view the home page. THERE IS NO NEED to click fresh or F5 on the home page, the values update themselves. You might not see any change because there are no sensors. But at least you should have a page.
have you installed all the correct required libraries
myAsyncTSL2591
mybme280e
myMLX90614AF
myOLED
If no, install them
Are you running Arduino 1.8.16 (NOT the web based version)
If no then update
Linux or Windows?
When connecting or removing a device, POWER must be off (12V and USB) - in other words the USB cable has to be disconnected first.
Some sensors require 3V3, others 5V and any 5V sensor will not work if powered from 3V3
So let us start with the TSL2591 sensor. It is a 3V3 sensor and if connected to 5V then you will likely damage the sensor
With power off, and cable disconnected, then connect the TSL2591 to the esp32.
Program the controller with the I2C scanner program and it should find the device and list its address as 0x29
If the I2c scanner does find it then good, but if not, you are going to have to figure out why. That involves checking your 4 wire connections again.
IF and ONLY IF the sensor is found,
Start arduino IDE
Load the TEST_TSL2591,ino
Open the serial port monitor
Set the serial monitor to 115200
program the controller with TEST_TSL2591
Watch what the serial monitor program displays
IGNORE - TSL2591 Sensor fail
TSL2591 ready: read lux
TSL2591 sensor ready, lux = some value
If you get this far, all well and good.
Get a small hand towel (dry) to block out most light to the sensor.
The reading should drop to a low value
Remove the hand towel
The reading should increase and stablize
Get a hand torch and about 1-2 feet way, turn the torch on,
shine the beam at the sensor surface and
the reading should further increase,
The program reads the sensor every 2s, so rapid movements of the torch etc will be missed becuase it is reading the sensor value - readinds are 2s apart
for 10s shine the torch at the sensor
The next 10s turn the torch off
For the next 10s cover the sensor with the hand towel
You should now have 3 different readings that are somewhat consistent
You can run these tests 2-3 times to verify the results (it does not matter if there is a small variation between results) - what you are expecting is
lowest value = sensor blocked from light by hand towel
mid value = sensor not blocked
highest value = torch on and beam shining on the surface of the sensor.
let me know how that goes.
I am not trying to teach you how to suck eggs, I am just concentrating on a strategy to verify the operation or failure of the sensor.
Regards
Robert
Last edit: brownrb 2021-12-26
Maybe you should upload here your "controllerconfig.h". Remove personal info like passwords.
So we could check if you made the correct settings. Maybe you have overlooked something.
Hi Robert,
Thanks for your detailed answer. You give me plenty of homework.
some details in advance:
"I saw the photo you posted. Those resistors are pretty close to other pins, are you sure they are not touching a pin they should be?"
Yes, I am very sure
"I also see that you have the oled display working."
Yes, it's working on the same I2C as the (not working) sensors.
"So let us start with the TSL2591 sensor. It is a 3V3 sensor and if connected to 5V then you will likely damage the sensor"
If you check your circuit diagram, the TSL2591 is conected to J9 which is powered with 5 V, but this doesn't matter, because it have an own power regulator.
"Once connected, open a web browser and enter the location
192.168.4.1/admin1
and you should see the first admin page."
I see all the pages, but if I change the numbers e.g. coordinates, I can't set them.
"After viewing all the web pages, try
192.168.4.1
to view the home page"
I see values changing somewhat with the light, but after 3-5 minutes the ESP32 hangs it up and I loose conection.
*"have you installed all the correct required libraries
myAsyncTSL2591
mybme280e
myMLX90614AF
myOLED"
**I copied them into the library directory, while compilining I get no error message, but they don't appeare in the library manager. should they?
So because the light sensor somehow works, I reconected the BME280 and MLX90614
If not conected I get Hum 50, Amb. 20, Pressure 11000
Conected sensor I get Hum 0, Amb, nan, Pressure nan. It is recognized (see also I2C scanner) but I get no values. I changed the sensor (I mostly buy more than one), but the same.
mySQM+ Admin10
SENSOR STATUS
BME280 Found
GPS Not defined in firmware
MLX90614 Found
...
...
TSL2591 Lux Found
It's 2:30 AM local time, I have to go to bed.
Ralf
What happened to just testing the TSL2591 ?
Running the test program for the TSL2591?
You can only enable TSL2591, not all the others. And don't connect them. I want to see the output of the test program,
Last edit: brownrb 2021-12-27
I would like to take this time to give a little more information that is not in the pdf or mentioned in the test programs.
The lux sensor has two registers that contain count values that represent the lux value.
To manipulate this the sensor uses two settings, gain and intergration time
The higher the gain setting (there are from memory 5 gain settings) the greater the ability to read low light values, the lower the gain then the brighter the light
The higher the integration time (how long we let the light fall on the sensor before it is read) is used for low light values, smaller integration times for bright days
Low Gain, Low Integration = very bright day
High Gain, High (long) integration times = very dark days/nights
The firmware uses an algorithm that senses the lux level and checks if any of these registers approx maximum value, and if so, adjusts the gain and intergration time accordingly to go to the next optimal gain/integration setting.
Conversely the firmware also looks to see if the register values are getting too low, meaning an increase in gain and/or integration time will be necessary.
For example, if the light was increaseing in intensity, then we reach a point where the registers overflow and we need to lower the gain and the integration time. Converse for light decreasing.
All of this takes time, adjust gain.time, check if registers are still full, adjust again etc
In addition to prevent sudden blips (a lightning strike - or light change that is transient in nature) the firmware also uses an averaging system to handle these outliers.
There is a lag between changing light levels and the sensor reporting appropriate lux levels (because of the adjustments to set an optimal gain/time setting and the averaging of lux values that is done).
In normal operation, the sun comes up over the horizon slowly, it just does not appear full in an instance, same with evening. These changes take minutes or hours to happen,
So, the lux plots using mysqmweb etc look very good because of what the firmware does. There are also some additional things the firmware does with the lux sensor that I won't go into right now.
The task timer controls the reading of lux values which about 10 sesonds apart.
When the firmware starts from a cold boot or reset, it is going to take some time before the sensor is fully configured for the current light. This can take up to 30s as the firmware checks the register values and adjusts the gain and integration settings to find an optimal setting. So after a boot, you see lux as 1000 (just a start value) then stay there for around 30+ seconds before it is changed to another value.
If the light source falling on the sensor changes - remember outside it does not change rapidly - by human hand by using a torch or cloth to alter the amount of light falling on the sensor, the lux setting will not change UNTIL the firmware has established the new register settings gain/time for the light. Again around 15 seconds required for this to happen
I point out again that during normal operation during the day/night we do not have these rapid sudden changes of light intensity. And when we plot the graph, it looks nice and smooth, but we are seeing hours of change. Imagine if one plotted a graph of instaneous samples over 20s - I leave that one for you to think about. (Attached is a sample graph of lux over 3 hours)
The problem with the test is this, you make a change, you think the next reading should be changed, it is not, you end up thinking it does not work, etc, etc
The controller remember is readjusting the gain/time settings to find a combination of gain/integration settings that gives a good range to measure the new light level, and that does take a heap amount of time. In daylight we might wait 12+ hours for the sun to go down, but we are trying to simulate that in an instant.
All of the test programs are under revision now, because firmware 125-xxx has a significant change in the way the controller operates. Which is why I posted the old tests.
In hindsight, the choice of using the tsl2591 for my example in previous posts was an error on my part. I apologize for that. A lot of what I just wrote I had forgotten, it all came back looking at the code today and running the test myself. So again, my bad.
I found it takes around 1m after boot to stabilize, when lux value of 1000 changes to a new value (the controller is busy adjusting gain/integration settings)
I found it takes 15-20s for a new value to appear after changing the light intensity (meaning if using a torch you have to keep it in the same place shining on the sensor for 30+ seconds), and when turning the torch off it takes 20-30s to stabilize again
Cheers
Robert
Last edit: brownrb 2021-12-27
Hi Robert and Ralf,
thanks a lot for that explanation! I was also always a little bit surprised that alway the LUX value started with 1000 as well as it needs some time until a stable value is reached.
I placed my setup (just connencted via some cables) outside for a night and could see very tiny changes when sun raises or drops and it seems that every second of more or less light was notified. Also clouds with lower and higher density can be detected with this sensor. So from my point of view this sensor works better that expected.
Referring to Ralf´s problem I have some little ideas.
1. WHat I can see in his picture is that he uses a Light sensor from Adafruit as I do. This sensor has build in pull up resistors for the I²C bus. If you will have a look into the Adafruit PDF document of this device you will see that they do not use additional external resistors in their examples. I don't know what will happen if you spent too many of them in parallel? The other boards do have them not I think but maybe since at least it is now only 5kOhm this may have impact. So I would remove the external resitors just tocheck the bahaviour.
2. To be frank I have (only) 2 modules at the moment and both are from Adafruit and I use in addition external resistors additionally and it works. BUT I do not have any other module which maybe will be affected.
I am waiting for the UV sensor which is not from Adafruit and I will see if this has impact. In this case I would remove the external ones.
Interesting thing is, that every Adafruit module has these resistors but they do not tell us to limit the amount of modules on the bus. As can be seen in the Adafruit PDF there is also a FET transistor added to each I²C bus and I do not know the reason....? A HW expert can tell us.
3. The Adafruit PDF also includes how to test their modules with their standard libraries. SO I would use them in the Arduino IDE instead of the modified from RObert, just to validate the behaviour!
One additional question to Ralf. Did you install also the "other libraries"?
Also to Ralf, when I changed from the prevoius FW to the latest I removed the old Windows programm and installed the new one. Also I installed ALL libraries from the new package as well as the new sketch. I am using the Firfox browser and I'm doing all test in station mode using my WLan network.
Regards
Hardy
Hi Robert,
Thanks for your detailed answer. You give me plenty of homework.
some details in advance:
"I saw the photo you posted. Those resistors are pretty close to other pins, are you sure they are not touching a pin they should be?"
Yes, I am very sure
"I also see that you have the oled display working."
Yes, it's working on the same I2C as the (not working) sensors.
"So let us start with the TSL2591 sensor. It is a 3V3 sensor and if connected to 5V then you will likely damage the sensor"
If you check your circuit diagram, the TSL2591 is conected to J9 which is powered with 5 V, but this doesn't matter, because it have an own power regulator.
"Once connected, open a web browser and enter the location
192.168.4.1/admin1
and you should see the first admin page."
I see all the pages, but if I change the numbers e.g. coordinates, I can't set them.
"After viewing all the web pages, try
192.168.4.1
to view the home page"
I see values changing somewhat with the light, but after 3-5 minutes the ESP32 hangs it up and I loose conection.
*"have you installed all the correct required libraries
myAsyncTSL2591
mybme280e
myMLX90614AF
myOLED"
**I copied them into the library directory, while compilining I get no error message, but they don't appeare in the library manager. should they?
So because the light sensor somehow works, I reconected the BME280 and MLX90614
If not conected I get Hum 50, Amb. 20, Pressure 11000
Conected sensor I get Hum 0, Amb, nan, Pressure nan. It is recognized (see also I2C scanner) but I get no values. I changed the sensor (I mostly buy more than one), but the same.
mySQM+ Admin10
SENSOR STATUS
BME280 Found
GPS Not defined in firmware
MLX90614 Found
...
...
TSL2591 Lux Found
It's 2:30 AM local time, I have to go to bed.
Ralf
Hi Hardy and Robert,
first of all many thanks for all the effort to help me!
Thanks for detailled explanation about working the light sensor...
Due to that, I think the light sensor is working:
it starts at 1000
after a while its now about 76.00000 (SQM 7.8....) indoor, no extra light
about 30000.00000 (SQM 1,7...) with torch
about 1.80000 (SQM 11,9...) covered by paper.
With the knowlege of the delay, the messurement is consitant.
So for me the light sensor is working, thats a big step foreward!
So I conected the IR and BME280, they are NOT working (recognized but no values). Maybe the reason is the parallel pull up resistor as Hardy mentioned. I will check this after lunch!
I am not very experienced with arduino, so maybe a silly question about "installing a library":
Up to now I "installed" all libraries I get a error (missing message) by checking the sketch.
- ArduinoJson and ESP32Time: I used the library manager
- For myOLED, myAsyncTSL2591, mybme280e and myMLX960614AF I just copied the files into the library directory until I got none error message, as a requirement to upload the sketch. But these libraries do not appeare in the library manager in the "installed library section". Is that normal?
I am surprised, that I can't find the onewire library neither in the directory nor in the library
manager (I was sure, it is already installed from other sketches, but due to the new Win11 PC it was not). There was also no need to install OneWire by checking the sketch...
(UPDATE: the installation of Onewire didn't change anything)
Ralf
I did copy the zipped lib files (optionals and requiered) to the libraries folder and unzipped them. So each one gots a folder. Thats al I did. See screenshot.
To be frank, when compiling the sketch and you get no message about a missing library do did it in the right way (i think).
All libs with modified folder-names from Robert do not appear in my librray manager too.
Thats all for now. Good luck by experimenting with the resistors (have attached the PDF of the Adafruit unit I have seen on your screenshot. Look at the schematic where you will see the onboard resitors a dn the FET Transistors in the I²C bus were I do not know how they should work).
I do not know if any of the others are also from Adafruit. The design of the Stemma units are easy to identify.
So I conected the IR and BME280, they are NOT working (recognized but no values). Maybe the reason is the parallel pull up resistor as Hardy mentioned. I will check this after lunch!
unlikely. Should not make much difference, the more devices that u plug into the I2C bus the lower the total value of all pull-up resistors need to be. hence it is common practise to include resistors on each I2C sensor. without going into the maths, it is fine.
Note: There are things you need on the I2C bus when adding more sensors. It is possible for one sensor to be a little noisy and that causes issues with the signals SCL and SDA, so that is why the schematic has a number of capacitors across the data lines and VCC - c4, c5 and c6 should be fitted in place, else you may have issues once all the sensors are fitted,
But these libraries do not appeare in the library manager in the "installed library section". Is that normal?
yes, but you can also load those files using the add zip of arduino IDE, instead of copying the folders, not an issue though
I am surprised, that I can't find the onewire library neither in the directory nor in the library
manager
correct, that is a special part of arduino which is installed when running the Arduino IDE setup program, it is stored along with the IDE in C:\Program Files (x86)\Arduino i think from memory the libs folder
To test the BME280 use the attached - I tested this today and it is working fine for me, there is also a sample output text file to see what I got when I ran the program.
Please program the controller with this and only the bme280 sensor attached (remove all the other sensors)
the test files are ready to go, no changes necessary
Let me know what you get displayed in the serial monitor window
Regards
Robert
Last edit: brownrb 2021-12-27
Hi Hardy and Robert,
first of all many thanks for all the effort to help me!
Thanks for detailled explanation about working the light sensor...
Due to that, I think the light sensor is working:
it starts at 1000
after a while its now about 76.00000 (SQM 7.8....) indoor, no extra light
about 30000.00000 (SQM 1,7...) with torch
about 1.80000 (SQM 11,9...) covered by paper.
With the knowlege of the delay, the messurement is consitant.
So for me the light sensor is working, thats a big step foreward!
So I conected the IR and BME280, they are NOT working (recognized but no values). Maybe the reason is the parallel pull up resistor as Hardy mentioned. I will check this after lunch!
I am not very experienced with arduino, so maybe a silly question about "installing a library":
Up to now I "installed" all libraries I get a error (missing message) by checking the sketch.
- ArduinoJson and ESP32Time: I used the library manager
- For myOLED, myAsyncTSL2591, mybme280e and myMLX960614AF I just copied the files into the library directory until I got none error message, as a requirement to upload the sketch. But these libraries do not appeare in the library manager in the "installed library section". Is that normal?
I am surprised, that I can't find the onewire library neither in the directory nor in the library
manager (I was sure, it is already installed from other sketches, but due to the new Win11 PC it was not). There was also no need to install OneWire by checking the sketch...
(UPDATE: the installation of Onewire didn't change anything)
Ralf
You are right, the pull up resistors dont make a difference.
Added all the captators, only BME280, running test_bme280, I get from serial monitor: see scrennshots. I used my 2nd BME280, the same. It's unlikely that both, new sensors are not working?
it gets even more mysterious:
- I2C scanner founds the BME280 at address 0x76, see screenshot
- loaded the test example which comes with the adafruit bme280 library, there I get Sensor ID is 0x0, could not find a valid sensor, check wiring, address, sensor ID...
- check wiring: there are only 2 wires SCL to D22 and SDA to D21, thats not very difficult....
- address is right 0x76
- what's about sensor ID???
I2C working for the light sensor and Oled, but not for the others sensors, how can this be?
(At the beginning I thought this is because of a wrong i2c address, so I changed the address (cut+soldier), but then I got the default values 50/20/11.000.)
Maybe the BME280 is in fact a BMP280? Google found some hints, that this can cause the indications. Ordered an other BME280....
Meanwhile I will now test the IR sensor
Tested MLX alone:
- test from mySQM: stopps after "start task timer"
- test from adafruit: Its working, if I hold my hand over the sensor, object temp rises up to 31°C
- older test from mySQM is working as well!
So let us take a step back
The BME280 is recognized by I2C_scanner, the fact that is recognized means it is wired correctly.
Please do not try the Adafruit code, it is a waste of time and does not prove much at all. It just loads up all your libraries with stuff you do not need - I might write more about this later, but for now
First
Delete the mybme280e folder that you copied into documents\arduino\libraries
then
Download the attached zip file checkbme280sensor
unzip the file
double click the ino file to load it into the Arduino IDE
Open the serial port monitor
program the esp32 with the code by clicking the ->
let it run for around 10s
Copy the text from the serial monitor window and post the results
I have includes the output that I got when I ran this a few minutes ago
regards
Robert
Last edit: brownrb 2021-12-28
Good morni ng Robert,
only the BMP280 connected to the board,
deleted the mybme280e library,
programmed checkbme280sensor.ini onto the esp32,
It started first with "reading....", I made a hardware reset.
It founds a sensor, its not a BME280, because its a BMP280, as I told you (If you look exactly you can see that my sensor is more a rectangle than a square).
I think (hope) that's the reason why it doen't work. I awaite the real BME280 in some days.
Meanwhile I tested the MXL IR sensor, see post 18h ago: there ist only the older sensortest working.
regards
Ralf
Hi
the BMP280 is not compatible. It will not work. It would require a lot of code changes. makes no sense to do that. So you will need to wait till your bme280 sensor arrives.
As the ouput says, it is not a bme280 sensor, hence no values.
In my mind there was a possibility of either a library fail or sensor fail.
In the test program I send I added the sensor ID test in the code as well as placing the library files directly with the ino file. This would overcome any library install error, or identify a different sensor, so the library files worked, indicating no issue with the librry you installed, and identified that it was not a bme280.
Re mlx
Its past midnight in NZ right now. So I am winding down now.
I will put together a special mlx test for you to try (later my time). So in the meantime, have a rest?
Regards
Robert
test for mlx90614
Thanks, Robert.
This sketch is working. What tells this you?
Regards
Ralf