TMP116: TMP116

Part Number: TMP116

Hi,

I've been trying to figure out the problem for couples days now and finally found the issue.

Long story short: I have developed a driver for the TMP116 that worked fine with the TMP116evm. We broke the eval board to only keep the tmp116 and connected it to our custom board and everything works fine. However, with the TMP116 on our custom board we are unable to get the MSB of registers 0x00 and 0x01, as stated in the question. The only difference between both is the AD0 pin. The TMP116evm had AD0 pin to V+ and the TMP116 on our custom board had AD0 to GND. When we changed the AD0 pin to GND on the TMP116evm we encountered the same problem. We can read/write every register except for the MSB of reg 0x00 and 0x01.

Thing is, now we're stuck with PCBs where the AD0 is connected to GND. Are you aware of a solution other then reworking the PCBs?

We look forward to your reply.

Alex

Hello Alex,

There is a similar thread that has been posted at the same time as yours by another user. I believe both of them to be the same issue. Can you please confirm?

e2e.ti.com/.../750650

Hello Julien,

This is something that is not expected. Let me check the same on my EVM and take up the debug, before I can confirm the issue exists and possible WA's if any.

Yes, looks like the same issue.

Alex

We broke away the sensor from the EVM and installed it on our PCB. The resistor on the breakoff is set to R15 which corresponds to ADD0 being grounded and this doesn't work (we also removed the TMP116 from our PCB). When we set the resistor to R13 which corresponds to ADD0 being to V+ it works.

Hello Julien,

I tested my TMP116EVM with the GUI Composer Application. I confirm that on my board the resistor R15 is installed (ADD0=GND) and I can read both the MSB and LSB of the temperature configuration register for a room temp of approx 26 degC (which corresponds to 0xCXX).

Can you please send the I2C bus snapshot for the I2C access for the temperature register?

Hello Alex

OK, then I will join the two threads.

Hello Julien, Alex,

Any update on the thread? Also when you changed the connection for the ADD0 pin did you change the slave address?

Hello Amit,

I'll send you a snapshot of the I2C acces for the temp register tomorrow AM. And yes, we changed the slave address.

Regards,

Alex

Hello Alex

Also since you changed the slave address to 0x48 for the I2C access, could you also provide the following

1. Schematic showing the TMP116 connection to the MCU pins and circuit for the TMP116
2. Also please provide a register dump for TMP116 in ADD0 connected to V+ and to GND.

Hello Amit,

1. Here is the schematic of the TMP116 connections:

2. With regards to the register dump for the TMP116 with ADD0 to  V+ everything is as expected after a reset. Here is the register dump for the ADD0 to GND :

00h ---- 0003

01h ---- 0020

02h ---- 6000

03h ---- 8000

04h ---- 0000

05h ---- 41CB

06h ---- 00FF

07h ---- 0000

08h ---- 0870

0Fh ---- x116

Regards,

Alex

Hello Alex,

Except for the EEPROM registers which seem to have a non-zero value, I agree 0x00 and 0x01 register address do not seem correct. Were you able to get the I2C Bus waveform as well for 0x00 and 0x01?

Hello Amit,

Here are the snapshot for the I2c Bus waveform for 0x00.

The red arrow indicate where the previous snap ends.

Let me know if you need anything else.

Regards,

Alex

Hello Alex,

Thanks for the waveform. The format seems to be correct except on the last SCL where the MCU is supposed to NACK, but instead it shows that the data is being ACK'ed. Not sure how it could affect the transactions; but worth a shot to correct the behavior.

Also did you try to set one of the bits in the register 0x01 and read back the same to see if the bits are getting set in the upper byte.

Hi Amit,

Yes I did. The bits in the MSB were still 0 but I could set bits in the LSB and the read value was correct.

Regards,

Alex

Hello Alex,

Attached is a capture from the TMP116 EVM with the address pin connected to GND that I captured a few moments back.

Can you please check if the NACK issue on the read transaction is there on the ADD0=V+ connection as well?

Hi Amit,

Here is the snap of the end of the transaction with ADD0=V+.

And just to be clear, both configuration (all the snapshots that I've sent) were taken from the breakout of the TMP116evm connected to our PCB.

Regards,

Alex

And by the way, there is an error in the evm datasheet.

V+ isn't connected to R13 but to R14.

Regards,

Alex

Hello Alex,

As you can see for some reason the MCU does not place the NACK correctly for the last byte when ADD0=GND while it does for ADD0=V+. If you connect the breakout to the EVM does the breakout board work as expected?

As for the EVM table, let me double check and confirm.

Hello Alex,

Is there any update on the experiment we suggested earlier of connecting the EVM to the breakout board with ADD0 connected to GND?

Hi Amit,

We decided to go for the hardware fix (cut the AD0 trace to ground and short AD0 to VDD -> pin 4 & 5). I don't think I will push the debugging further since I already lost a lot of time on this.

Also, I'm still not sure if the bug comes from the TWI driver or TI's chip, but we never had a problem with the TWI driver and we worked with more then 10 other peripherals.

If you want to recreate the environment we're using nRF52832 from Nordic, our application is based on SDK15.0.0 and we're using nrfx TWI driver module from the SDK.

Thank you for your help with this issue!

Regards,

Alex

Hello Alex

We did test multiple TMP116 EVM and breakout boards with another MCU from TI (MSP432E4) and we did not see the behavior at all either. It is indeed intriguing why the TMP116 and the nRF52832 would not work with ADD0=GND. Hopefully if you do get the time please do look into checking it against the EVM controller.

Hello Alex.

Can you please say what is device supply voltage, I2C bus voltage, I2C bus pull-up resistors value, I2C bus clock frequency, what is bus length?

Hi Mihail,

TMP116 supply voltage is 3V and nRF52832 supply voltage is 1.8V.

I2C bus voltage is also 1.8V and is pulled-up by 5k resistors.

I2C bus clock is at 100K and the bus length is 152.4mm.

Alex

Just comments. The official input voltage range is 0.7*V+....6V. To prevent device self-heating it is recommended to keep Vih above 0.95*V+. If you still want to use 1.8V I2C bus, you can try to lower pull up resistors to 1 KOhm.

I recommend you look into SNOA986 on product website