Fall time differs from datasheet on i.MX8QM I2C

[YukioOyama]

Hi all,

See the post below. I connected the i.MX8QM to an SE051 and checked the waveforms of the interface.

https://community.nxp.com/t5/Other-NXP-Products/Falling-time-of-SCL-and-SDA-to-SE051-is-earlier-than...

The I2C host is i.MX8QM and the follower is SE051. The clock speed is 400 kHz , so they work in fast-mode.

The fall time of the i.MX8QM clock is about 1.5ns, which is much faster than the fall time IC11 in the data sheet below. This is the same time for the SDA output by i.MX8.

YukioOyama_0-1699510216979.png

I think this seems to be a simple open drain Low time. And for the Fast-Mode fall time IC11 in the datasheet, an additional slope control seems necessary.

Does the i.MX8QM have a slope control somewhere? And do I have to turn it on to use Fast-Mode?

This fall time is a violation of the UM10204 specification to which SE051 adheres. So I would like to resolve this. Without reducing the speed to standard mode, of course.

Best Regards,

Yukio Oyama

[jimmychan]

Do you see any problem of the I2C communication?

[YukioOyama]

Hello Jimmychan-san,

Thanks for your reply.

I am not going to say there is no problem in every condition, but right now there is no problem.

However, I want to pass the specs on the datasheet with waveforms, before any problems occur on the software.

Best Regards,

Yukio Oyama

[fas3]

1. i2c has no spec. It is user manual now. Of couse, some version says spec. But not now vesion. 

2. the falling time tf in the UM10204 is Start(S)/repeated Start(Sr). It just gives time for sampling the start condition and make sure it is a start condition. For others, it should be as fast as possible to fall to get ready for next bit. For others(not detecting start condition), no meaning. Because the data sampling is on the raising edge. And if you read the UM10204, the tf is only tagged on S of the wavefrom. 

3. make sure, you measure the S/Sr. 

4. a little difference between S and Sr. In the UM10204, no tf of Sr, it says/measures in different way, accually. 

[YukioOyama]

Hello Fas3-san,

I understood that the falling time applies only to the starting condition. However, the TF time did not change whether in the start condition or in the middle of the data.

I could adjust the TF by putting a series resistor like the one below from UM10204 in the slave device.

YukioOyama_0-1700524360512.png

However, I do not think this is the right solution to this problem. i.MX8QM should satisfy the fall time in FastMode. And the iMX8QM datasheet says so.

Best Regards,

Yukio Oyama

[fas3]

Tf is changed in the start condition or in the middle of the data.

Even in the UM10204 uses two different figures. 

Note, there is "extreme" case, i2c Clock stretching from slave side. Slave pulls the clock low. 

UM10204

2023-11-21_084357.png

2023-11-21_091933.png