Is a 100kHz I2C Signal "fast" enough to become a source of EMI problem?
Hans-Jürgen Hartmann
Support to implement Digital Twin in Electronics Design and Manufacturing
We talk a lot about "Signal Integrity" and you should do analysis of your boards because signals become faster, like Gigabit Signal in DDR4, PCIe and so on. Back in 1986, wiring up a microcontroller and it's EPROM and SRAM, no "power plane", the "Power Delivery Network" was just some wires. Hardware was working ... but I had to keep it away from antenna TV.
Signals on printed circuit boards are getting "faster" indeed. But "faster" does not only mean fast clock rates. “Fast” is predominantly determined by the edge rates of signals in V/ns. Think about a very "slow" signal like a 100kHz I2C or a 1Mbps SPI signal, "slowly" travelling from a Microcontroller towards another chip or connector. The signaling rates are not fast at all in common sense, but often, such signals are driven by a piece of silicon (the Microcontroller) that is designed for 100s of MHz signaling speeds. Often you see, that I/O pins driving an I2C or SPI signal with an edge rate of say 1ns rise/fall time or faster. So even if the intended signal is in range of 100kHz or 1MHz, significant energy is in the frequency ranges of 100s of MHz making the signal indeed “fast” enough to create trouble.
If you do not follow rules of “impedance planning”, “Termination”, “Topology planning”, well design of HF return paths, “Controlling Crosstalk” and so on, then your design will probably still work (like Hardware back in 1986), but not reliable enough and you will measure the bad performance in EMI chamber. I was motivated to compile a Webinar about Basics in Signal Integrity. I am sorry, so far only available in German language.
Get execited by the German language or send me a comment and I will compile it in English too.?WebinarVideo
Enjoy!
Hans
C.E.O. at Induction Heating Technologies
2 年Hello Hans I am glad to read your post and the 100khz subject attracted my attention. Our company was asked to do a project for a univercity for 50-100KHZ induction heater kit.. That neccessitated a low budget aproach. But the 100khz pcb layouts especially in the power stages, and 80's style layouts turned out not be an easy job especially at resonance. About 50-100khz Pwm signals generated from a 20 pins UCC2895 phaseshift IC are employed to drive a Mosfet fullbridge. The layout is 80's style , and the pwm signals are clean and work well. TCC4422 and TCC4421 are fed with the pwm signals to buffer them to drive the gates of the 4 mosfets. The PWM signals show clean with sharp rise times at the gates, but when driving the mosfets noise and instability is always noted which increases as resonance is approached. Any similar experiences or ideas to share with thanks to your post. N Jaber
PCB designer at Marquardt Schaltsysteme SCS SIBIU
5 年Great initiative on an interesting topic, can you please also create one in English?