I can't get the i2c to go below 100kHz. The minimum value of the frequency with which I see i2c-1 with i2cdetect - l is to specify in /boot/config.txt this value: In this case, @poleg-on we have the same output for i2cdetect -l
I also have an announcement to make! Compared to what I described in my first post on this forum, I've investigated further, and I've found a mistake I'm not proud of! In fact, we made a mistake on the serigraphy of the 4th layer of our version 1.0 PCB. I had inverted SDA and SCL on the silkscreen, so I made a mistake in making my cables to test the AstrAlim board with a BME on our i2c jack. I realised the error by analysing the signals on the i2c bus at different points using an oscilloscope.
In reality, apart from the error I made earlier, which may complicate the reading of the problem, it doesn't really change our misunderstanding of the difference in behaviour between the pi4 and the pi5. Should we add more pullups because the pi5 is more 'sensitive' than its predecessors? Why is it that when we connect a 1m cable with a BME280 and its 10k pullups, the INA219s are all discovered with their correct address? In our calculations for the pull-up resistors, we thought we'd paid attention to the recommendations! Have we missed something?
I soldered two 10k Ohms pullups under our board between SDA and 3.3v and SCL and 3.3v. It doesn't seem to be enough. In total, we have :
- pi5 internal pullups (unknown values)
- our hardware pullups: 10k (see https://oshwlab.com/pololamag/nafa-astralim)
- the two other 10k pullups that I soldered underneath.
The only way I've found at the moment to communicate properly with all the i2c devices connected is to connect a BME280 and its 10k additional pullups at the end of a cable about 1m long.
Here again is an illustration of the problem. In one case, the BME is unplugged, in the other the BME is plugged in and all the devices are clearly visible.
![Image]()
![Image]()
Code:
dtparam=i2c_baudrate=100000I also have an announcement to make! Compared to what I described in my first post on this forum, I've investigated further, and I've found a mistake I'm not proud of! In fact, we made a mistake on the serigraphy of the 4th layer of our version 1.0 PCB. I had inverted SDA and SCL on the silkscreen, so I made a mistake in making my cables to test the AstrAlim board with a BME on our i2c jack. I realised the error by analysing the signals on the i2c bus at different points using an oscilloscope.
In reality, apart from the error I made earlier, which may complicate the reading of the problem, it doesn't really change our misunderstanding of the difference in behaviour between the pi4 and the pi5. Should we add more pullups because the pi5 is more 'sensitive' than its predecessors? Why is it that when we connect a 1m cable with a BME280 and its 10k pullups, the INA219s are all discovered with their correct address? In our calculations for the pull-up resistors, we thought we'd paid attention to the recommendations! Have we missed something?
I soldered two 10k Ohms pullups under our board between SDA and 3.3v and SCL and 3.3v. It doesn't seem to be enough. In total, we have :
- pi5 internal pullups (unknown values)
- our hardware pullups: 10k (see https://oshwlab.com/pololamag/nafa-astralim)
- the two other 10k pullups that I soldered underneath.
The only way I've found at the moment to communicate properly with all the i2c devices connected is to connect a BME280 and its 10k additional pullups at the end of a cable about 1m long.
Here again is an illustration of the problem. In one case, the BME is unplugged, in the other the BME is plugged in and all the devices are clearly visible.
Statistics: Posted by pololamag — Fri Jan 26, 2024 9:08 pm