I made a battery management system using one attiny per cell, communicating via I2C (using capacitors as a cheap way to shift voltage levels between the cells.). Each attiny has an external 100kOhm resistor to keep its capacitor at the correct offset. For the total bus, this means the effective pull-up is all of them in parallel (with 16 cells that's 6.25kOhm). Despite its high number of devices and high pull up resistance, it works fine (already for ~half a year). However it "only" runs in standard 100kHz mode and has long pauses between communications to let the connection capacitors settle to the correct voltage again.
Once i discovered I2C extender chips i stopped worrying about all this bus capacitance stuff. Those things are amazing. Just add them to yoir circuit and you can communicate over long diatances with no problem.
You can still lower the frequency to for instance 10Hz and add a huge amount of capacitance, right? It might not be a display on the other end, but a temperature sensor read every 10 minutes. It sounds like a fun project to make the master measure the rise time and configure the speed automatically.
If you measured the inductance of the following: SCL to GND, SDA to GND, and SCL to SDA, using an LCR meter would that correspond well to your calculations? SCL to SDA would measure unwanted coupling between the two, but curious about your thoughts on the other two measurements
Practically speaking, I don't think you could measure signal to GND with an LCR (without a fixture anyway.) And if you tried to measure signal to signal, you'd just end up measuring the effective capacitance (since these aren't DC signals). Regardless, I would expect the inductive effects to cause crosstalk while the capacitive effects will contribute to the (slower) rise times. Another way I first thought of your question is these are not DC signals and we're talking about impedance. By measuring (or calculating) the capacitance, we also account for the inductance. Maybe that needs more thought.
kind of seems like devices (like the oled screens) should have enough resistance to take care of themselves and their own built-in capacitance and then the "system" pcb's should have space for more.
Mh, maybe someone can correct me: I would like to trigger taps on capacitive touch screen (similar to a touch screen on your phone). I took the tip of a stylus, and soldered a long cable to it. The length of the cable alone already triggers on some screens enough capacitance; registering constantly as a touch. So I assume, I need to put enough „power“ in to the cable from my side, so it matches the screen... Would pull up resistors, knowledge about impedance etc in this video help in solving this scenario? Just need some pointers into what direction I should search learn - it is a learning project. Thx
Super useful.
I've never thought about it in that much detail before, but actually really useful to consider what you're using
Those are some nice big resistors in the thumbnail haha
Nah, they are only 1 kiloohms. ;)
Great video! Have always wanted to learn about I2C! Thank you!
very informative, thanks
I made a battery management system using one attiny per cell, communicating via I2C (using capacitors as a cheap way to shift voltage levels between the cells.). Each attiny has an external 100kOhm resistor to keep its capacitor at the correct offset. For the total bus, this means the effective pull-up is all of them in parallel (with 16 cells that's 6.25kOhm). Despite its high number of devices and high pull up resistance, it works fine (already for ~half a year). However it "only" runs in standard 100kHz mode and has long pauses between communications to let the connection capacitors settle to the correct voltage again.
until know I just used 10k as pullups..for no specific reason other than I saw others use 10k. well I guess its time to go measure...
I hope that's a key takeaway: make a measurement to verify you are in spec. If so, move on to real problems!
Once i discovered I2C extender chips i stopped worrying about all this bus capacitance stuff. Those things are amazing. Just add them to yoir circuit and you can communicate over long diatances with no problem.
Do you have one you can recommend? (I have an idea for an experiment.)
You can still lower the frequency to for instance 10Hz and add a huge amount of capacitance, right? It might not be a display on the other end, but a temperature sensor read every 10 minutes.
It sounds like a fun project to make the master measure the rise time and configure the speed automatically.
Nice thumbnail and even nicer video and theme.
If you measured the inductance of the following: SCL to GND, SDA to GND, and SCL to SDA, using an LCR meter would that correspond well to your calculations? SCL to SDA would measure unwanted coupling between the two, but curious about your thoughts on the other two measurements
Practically speaking, I don't think you could measure signal to GND with an LCR (without a fixture anyway.) And if you tried to measure signal to signal, you'd just end up measuring the effective capacitance (since these aren't DC signals). Regardless, I would expect the inductive effects to cause crosstalk while the capacitive effects will contribute to the (slower) rise times.
Another way I first thought of your question is these are not DC signals and we're talking about impedance. By measuring (or calculating) the capacitance, we also account for the inductance. Maybe that needs more thought.
Great thumbnail!😅
kind of seems like devices (like the oled screens) should have enough resistance to take care of themselves and their own built-in capacitance and then the "system" pcb's should have space for more.
is there a app that does the equation and users can input # of devices (for ppl who arn't good at math)?
I provided the Google Sheet calculators I used. Follow the link in the description to episode blog post to get that.
Mh, maybe someone can correct me: I would like to trigger taps on capacitive touch screen (similar to a touch screen on your phone). I took the tip of a stylus, and soldered a long cable to it. The length of the cable alone already triggers on some screens enough capacitance; registering constantly as a touch. So I assume, I need to put enough „power“ in to the cable from my side, so it matches the screen... Would pull up resistors, knowledge about impedance etc in this video help in solving this scenario? Just need some pointers into what direction I should search learn - it is a learning project. Thx