In my application I am seeing -0.7V undershoot for about 20ns on the SDA and SCL lines. This undershoot is present in standard mode (100kHz clock speed) on every clock cycle. The absolute minimum for the MP3320B is specified to be -0.3V. The I2C SDA and SCL each have a 2k pull-up resistor, which should limit the SDA/SCL sink current of the MP3320B to ~2mA.
In our application currently it would be difficult to change clock speed and drive strength to reduce the edge rate. Any capacitance on the line above ~150pF also begins to violate the NXP I2C rise/fall time specification. We would like to confirm whether the -0.7V undershoot, 2mA sink current can be tolerated by the chip. Please let me know if there is any more detail I can provide.
Is your scope probe compensated? Are you taking your ground from right by the 3320?
I just double checked my probe compensation for both probes I used to verify the measurement and the compensation looks adequate for both probes. I used a differential probe (RP7150) and a regular probe (RP3500A) with spring clip to probe across pins 6 (SCL) and pin 8/9 (ground).
In our application the I2C host and MP3320B are not on the same PCB. The I2C signals go over a 1m cable…in case that adds some context.
Hello,
I would not recommend allowing the voltage to drop below the -0.3V specified by the datasheet even though it is just for a short timespan. It may reduce the reliability of the part. I suggest using a small schottky diode to try and clamp the negative voltage. If you are worried about capacitance, there are I2C range extenders that add a push-pull to the lines and can increase signal integrity.
best,
-Kerr
Unfortunately we are not currently able to implement the range extender, diodes, or series/source termination resistors. What we would like to understand a bit more though is the input clamp current capabilities of the I2C SDA and SCL pins. Do you have an estimate of the allowable current?
Just following up on my previous question^
I would be surprised if they answer. How many you going to buy? What is their incentive for accepting the risk of saying sure no problem, you can exceed the datasheet? An IC is a big collection of transistors built on the same substrate. The transistors are isolated from each other through a mechanism called junction isolation. As soon as the substrate diode is forced to conduct that isolation is lost. Now you have electrons floating around in the substrate that could show up at a logic gate and flip it, or in a reference and alter its voltage. It is not a good thing.
Me personally as a rando on the internet I would say no problem, what an excellent and careful engineer you are to find this, most people don’t look and hence don’t bother to ask the question.
Have you seen a malfunction, how about in the hot where the Vf of the substrate is lower.
So absent observed problems, ignore it, or adopt their suggestions and protect it.
If I remember right, you can control the output port drive strength in the STM32. Try to have the software initialize the ports at low strength. If you can add components, add a small capacitor to the line and adjust the pull-up accordingly.