TCA6408ARSVR not responsing on I2C bus...

Hi Soren,

There are no known issues with using the device at 1.8v.

My first thoughts would possibly be incorrect pullup resistors.  What value are you using?

From there it is hard to diagnose what the problem could be without scope shots or other information.

Do you have a schematic?

Hi David,

Thanks for your quick reply. It's an OMAP design, and all in all I have 13 pcs I2C devices all running 1V8 on 3 I2C busses just using OMAP internal pull ups. Pulls were my initial idea as well, so I have just tried adding 1k0 on both SDA and SCL for the bus with the two failing TCA6408A chips connected to it, but unfortunately result is still the same:failing (and other chips on the bus still work :-)).

I'm however glad that you can confirm that there in general shouldn't be any problems running on 1V8.

For the given package datasheet doesn't show any GND pad in the center of the chip (compared to the other QFN package). Can you as well confirm that there actually is nothing there? Reason for asking is that I have free CU connected to GND in a pad under the chip. It doesn't contain any solderpaste and as said is just  a free CU area connected to GND. Please confirm if that could be a potential problem?.If there are no conductive elements on the bottom of the chip itself I guess not, as all stuff seem to measure correct around the chip => It for sure doesn't short any "official signals" to GND.

Last but not least, Please confirm that datasheet RSV package pinput is double checked and correct.

The chips I have does state the following on top:
ZVU
0HA

Can this be use for any further tracking? ZVU seem to be according to datasheet, but I don't know what 0HA mean? Can you help on that?

 Best regards and thanks again
  Søren

PS: Have tried two different boards as well, and both with same failing result => Right now I'm at 4 failing chips, which lead me to think of some kind of systematic wrong doing. Are chips i.e. in anyway particular sensitive to soldering heat?

Hi David,

I have today been able to get my hands onto a scope and schedule some time for doing measurement on this problem, and as expected all look 100% correct. I2C signals are 100% fine, and there are no glitches or other problem on neither any of the supplies (VCCI, VCCP, GND), nor the RESET or ADDR pins. All in all I'm therefore now back at square one not having any idea how to take it from here, except my feeling 100% point to a chip issue somehow...

Chips are bought from Digikey on July 2012, and have been stocked at the EMS company from then until like 1 week back when boards were produced.

I hope you will be able to look into above questions, as I'm running out of ideas. Next will be to get from a new batch of chips and as well try with the compatible NXP PCA6408A as I as said can't find anything what so ever in the system which seems wrong, except that TCA6408A chips just never ACK an addressing...

Best regards and thanks in advance
  Søren 

Hi Soren,

With having an exposed copper pad under a package like this, you run the risk of shorting a pin to the pad.

Can you confirm no pins are being shorted to GND from the pad?

Regards,

David

Hi David,

Thanks for your fast feedback. Hereby please find my response:

1) Are you sure there is an exposed pad under the RSV package type as well? Reason for asking is that it isn't shown in datasheet, where exposed center pad only is shown for the RGT package type. Please see below figure cut from datasheet:

2) I (un)fortunately was wrong in my 2nd post stating that we were missing coverlayer under chip. The layouter actually managed to get it in there. Please see below, where red is copper, yellow is assembly drawing, and purple is opening in coverlayer.All seem 100% correct to me, but hopefully I have missed something?

3) I can confirm that following signals measure 100% correct and aren't shorted to anything:

• I2C_SDA (1V8 in idle and less than 200mV when low)
• I2C_SCL (1V8 in idle and less than 200mV when low),
• RESET (1V8)
• ADDR (0V or 1V8 depending on chip)
• VCCP and VCCI (both 1V8)
• GND (0V and confirmed connected to system GND)

4) I haven't taken any special care on measuring any of the 8 GPIO pins, nor the interrupt pin, as I don't expect them to "dangerous" for a simple I2C addressing? I can however confirm that all of these are expected to be in "stable and not stressed condition", meaning either 0V, 1V8 or floating

As said I have tried two boards and 4 chips in total, and all behave exactly the same, so it seem to be some kind of systematic error, which I will most likely consider to either chip and/or assembly processing related? I hoping you can help me forward from here, as I'm really running out of ideas on this (which was expected to be a no-brainer based on my previous good experience with the TCA6408A, although until now always in RGT package and running at 3V3).

Best regards and thanks
  Søren

Hi David,

A small update on this one from my side. I tried mounting the PCA6408A from NXP and all worked 100% as expected. I have now changed 4 boards (8 expanders in total) and all work, whereas none of the original TCA6408A ACK'ed on I2C addressing. Hoping you can share any info you might have been able to find on what might have caused this, as it to me now for sure seems like either a chip or an assembly/process issue?

Next step I will try is running the non-functional TCA6408A expanders at 3V3, although it might be a little in the future as I'm right now running out of time for debugging this (previously considered to be 100% trivial and non problematic) component? :-(:-)

Best regards and thanks again
  Søren

Hi,

I have exactly the same issue as Søren mentions, but with the TCA6408ARGTR. I have used this part in previous designs without any issues at all. On a new board I have an identical design (same setup of the TCA circuit, but with a new batch of IC's). Everything measures ok - GND, VCCP, VCCI, I2C lines, ADDR, and RESET but for some reason I get no Acknowledge from the TCA when addressing it. I have other I2C devices on the same bus, and see no issues with these.

Can someone from TI, comment if there has been any issues with a particualt batch of IC's? (They have been bought from Digikey in January 2013.) Any special cares/concerns with handling this part during assembly?

Thanks and best regards,

Thomas