AM2632: SCL frequency difference for I2C0 but not I2C1/2

Mari Tsunoda

Expert 5726 points

Part Number: AM2632

Tool/software:

Hi Team,

Please check the related question for context.

Previously you answered:

The mismatch between programmed I2C SCL speed and measured SCL speed is due to the rise/fall time and latency of the synchronizer inside the module. While the I2C CLK is configured by software to be exactly 400KHz, the actual transfer rate is slower than the value calculated and implemented in software.

Please refer to section 13.1.2.4.2.1 I2C Clocking in the AM263x Technical Reference Manual for more information.

My customer only sees this difference for I2C0 and not I2C1 and I2C2. Is this correct?

Best regards,

Mari Tsunoda

over 1 year ago

0 Brennan Hartigan over 1 year ago

TI__Genius 11151 points

Mari,

I2C0 differs from the other AM263x I2C[1-3] peripherals in that Only the I2C0 instance is a true I2C Open Drain buffer. I2C[1-3] are implemented with the typical LVCMOS voltage buffer and should be properly configured to operate as an Input/Output Open Drain signal type.

I can check this on a scope tomorrow, if that is ok.

Regards,

Brennan

0 Mari Tsunoda over 1 year ago in reply to Brennan Hartigan

TI__Expert 5726 points

yHi Brennan,

I see.

Brennan Hartigan said:
I can check this on a scope tomorrow, if that is ok.

Yes please keep me updated.

Best regards,

Mari Tsunoda

0 Brennan Hartigan over 1 year ago in reply to Mari Tsunoda

TI__Genius 11151 points

Mari,

I measured 375KHz on I2C1 today. What is your customer seeing?

Regards,

Brennan

0 Mari Tsunoda over 1 year ago in reply to Brennan Hartigan

TI__Expert 5726 points

Hi Brennan,

Please check the images below:

I'm not sure if it's just the latency of the synchronizer module. They are also seeing that I2C0 module has a duty cycle of 60%. Is this due to the fact that it is open drain buffer?

Best regards,

Mari Tsunoda

0 Brennan Hartigan over 1 year ago in reply to Mari Tsunoda

TI__Genius 11151 points

Hi Mari,

Are they using a TI EVM (LaunchPad / controlCARD) or a custom PCB?

Regards,

Brennan

0 Mari Tsunoda over 1 year ago in reply to Brennan Hartigan

TI__Expert 5726 points

Hi Brennan,

This is for custom PCB.

Best regards,

Mari

0 Ralph Jacobi over 1 year ago in reply to Mari Tsunoda

TI__Guru*** 135355 points

Hello Mari,

Brennan is out this week so I might be lacking some background here but reviewing the thread - given this is on a custom PCB, what is on the I2C pins on their circuit? And definitely want to know value of the pull up resistors on the I2C pins, but are those the only external components on the lines?

Usually for an I2C bus the pull up affects the rise time, but not the fall time. The signature with the slow delay on the fall of the signal is unusual. I can test the I2C0 bus on Monday on one of our EVMs.

Best Regards,

Ralph Jacobi

0 Ralph Jacobi over 1 year ago in reply to Ralph Jacobi

TI__Guru*** 135355 points

Hi Mari,

So turns out none of our EVMs have I2C0 brought out as an I2C bus meaning they lack the pull ups needed. I'll have to see if I can get some installed but will depend how much time I have to go into lab to re-work one of them. Otherwise when Brennan is back he'd be able to do so.

Best Regards,

Ralph Jacobi

0 Brennan Hartigan over 1 year ago in reply to Ralph Jacobi

TI__Genius 11151 points

Mari,

I probed the I2C0 SCL line and measured 375kHz:

Because your customer is using a custom PCB, I cannot accurately replicate their setup. They must follow the recommended pull-up values given in the datasheet and hardware design guidelines. Layout/BOM differences in the customer's PCB may be the cause of what they are seeing.

Regards,

Brennan

Arm-based microcontrollers

Arm-based microcontrollers forum

AM2632: SCL frequency difference for I2C0 but not I2C1/2