CDCUN1208LP: I2C communication

Gavin Macpherson

Part Number: CDCUN1208LP

Hi,

I have been trying to communicate with the CDCUN1208LP clock buffer via a NI-8451 USB interface and labview with an I2C mux level shifting from 3.3V to 1.8V, which is the supply level of the clock buffer. My issue is that when I have 2 CDCUN1208LP connected on the same I2C bus (with the ERC pin pulled high and low respectively to make sure they have separate addresses) the communication fails. The error is "The I2C master lost arbitration and failed to seize the bus during transmission of an address+direction byte".

When I disconnect one of the clock buffers the issue goes away. I suspect that one of the buffers (probably the one with the ERC pin pulled high) think its on the same address (0101000). Is there a certain value of pull up I should use on ERC?

Also I have 47k pull ups on SCL and SDA but only one for the bus, should there be one per device on the bus?

The I2C mux I'm using may complicate matters but I'm able to communicate with other devices using it so I don't think that is an issue.

Any thoughts/help would be appreciated.

Thanks,

Gavin

over 9 years ago

0 Tim Chao Li over 9 years ago

TI__Expert 4415 points

Hi Gavin,

I have been able to replicate a similar issue on an EVM and am looking into the cause. I don't see anything wrong with your setup- there should only be one pair of pullups necessary on the bus.

To clarify, if you disconnect one buffer (so that only one is left connected), are you able to communicate with it regardless of the setting of the ERC pin? The issue I am looking into is that when ERC is pulled high, even if there is only 1 buffer on the bus, it appears to not behave well and locks the bus up by holding SDA low. ERC pulled low seems to work fine.

Regards,
-Tim

0 Gavin Macpherson over 9 years ago in reply to Tim Chao Li

Prodigy 100 points

Hi Tim,

Yes, after further investigation I can only communicate to clock buffers that have the ERC pin pulled low (0101000). Is there any workaround to this? We would rather not have to respin our board.

Thanks,
Gavin

0 Tim Chao Li over 9 years ago in reply to Gavin Macpherson

TI__Expert 4415 points

Hi Gavin,

Thanks for confirming. I am looking into the issue and will keep you updated.

Regards,
-Tim

0 Gavin Macpherson over 9 years ago in reply to Tim Chao Li

Prodigy 100 points

Hi Tim,

Any progress? If there isn't a work around is there a replacement that is pin compatible?

Thanks,
Gavin

0 Tim Chao Li over 9 years ago in reply to Gavin Macpherson

TI__Expert 4415 points

Hi Gavin,

Apologies for the delay, we are still working on finding the cause of the issue and I will keep you updated. Unfortunately we don't have a pin compatible substitute with the same functionality.

Regards,
-Tim

0 Gavin Macpherson over 9 years ago in reply to Tim Chao Li

Prodigy 100 points

Hey Tim,

Its been 3 weeks, I'm running out of time on my project. Is there anything I can do or are they just not able to communicate when there are 2 on a bus?

Thanks,

Gavin

0 Tim Chao Li over 9 years ago in reply to Gavin Macpherson

TI__Expert 4415 points

Hi Gavin,

Unfortunately we verified that there is a logic issue with this feature for which there is currently no easy workaround, so slave address 0x29 is not supported.

Regards,
-Tim

Clock & timing

Clock & timing forum

CDCUN1208LP: I2C communication