TS5A23157-Q1: Noise coupling with each channel

Roy Chou1

Genius 12595 points

Part Number: TS5A23157-Q1

Hi Team,

We occur the issue using TS5A23157-Q1. Below is our connection. We used TS5A23157-Q1 to switch the I2C signal. (400kHz)

Below is the config.

SW = 1

CH1: U801_SDA, CH2: U801_SCL, CH3: U802_SDA, CH4: U802_SCL

CH3: U802_SDA, CH4: U802_SCL

We find U802_SCLK has coupling noise from U801CLK or U801 SDA.

Remove chip and find the result is ok.

The occurring rate is 1/4. We did the A-B-A swap and find the issue is followed by IC.

Could you let me know how to fix it?

The issue will let our panel flicker. (The ch which has noise is LED driver I2C)

Regards,

Roy

over 2 years ago

0 Bryan Marshall over 2 years ago

TI__Genius 10974 points

Hi Roy,

Thanks for providing these details. So what you are observing is something we define as crosstalk for our multiplexers. Essentially, this is the amount of signal that is present on off channels when another channel is active. My colleague has a good explanation on how this occurs and how you can mitigate this in your system( a lot of the time this is caused by poor layout, high frequencies, or large loading):

https://e2e.ti.com/support/switches-multiplexers-group/switches-multiplexers/f/switches-multiplexers-forum/981172/ts5a23157-x-talk-between-no-and-nc/3624912#3624912

Let me know if you have any other questions or need additional clarification.

Thanks!

Bryan

0 Roy Chou1 over 2 years ago in reply to Bryan Marshall

TI__Genius 12595 points

Hi Bryan,

Thanks for information. It's clear information for X-talk, but could you let me know how to improve or fix our issue maybe by adding components?

For layout, I'm not sure the if this is the critical point because some of boards had issue and others didn't have.

Regards,

Roy

0 Roy Chou1 over 2 years ago in reply to Roy Chou1

TI__Genius 12595 points

And one more information. The issue didn't occur in every PCB.

0 Bryan Marshall over 2 years ago in reply to Roy Chou1

TI__Genius 10974 points

Hi Roy,

The fact that is does not occur on every PCB is quite interesting. Are there any setup differences between boards at all? Also, would it be possible to provide the loading on problematic signal? As noted in the linked post, the crosstalk parameter is characterized when the non-selected input is grounded through a 50 ohm resistor. So if you have a large output resistance on that channel, the crosstalk will increase as the loading is increased, so there may be a higher impact at lower frequencies.

Thanks!

Bryan

0 Roy Chou1 over 2 years ago in reply to Bryan Marshall

TI__Genius 12595 points

Hi Bryan,

The end of channel is another device I2C input. I think it is high impedance, but our signal is 400kHz. And we ensure that the issue is followed by chip instead of PCB.

Is it possible to send suspect chip to factory to test the related spec by using frequency analyzer? We would like to check difference between ok and non-ok part.

Regards,

Roy

0 Bryan Marshall over 2 years ago in reply to Roy Chou1

TI__Genius 10974 points

Hi Roy,

I would like to see all the connections the multiplexer is going to for further evaluation (devices connected to the mux, loading conditions, and supply caps). Can you provide me the full schematic for this application? In an ideal situation, there should be nothing wrong with how the connections to the mux are being made in the diagram you provided, but there is clearly something amiss and would need to see more of the circuit to make sense of what is happening.

Thanks!

Bryan

0 Roy Chou1 over 2 years ago in reply to Bryan Marshall

TI__Genius 12595 points

Hi Bryan,

Thanks for information. Let's discuss internally.

Roy

Switches & multiplexers

Switches & multiplexers forum

TS5A23157-Q1: Noise coupling with each channel