This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
We are using four TMP75AIDGKT temperature sensors on our current design which have their Alert signals (pin 3) connected to a single wire net. We seem to be experiencing an occasional erroneous ALERT signal for the very first temperature sample after power up.
I’ve measured this occurrence to approximately 1 in 10000 power ups on average whilst the board is at room temperature and no warmer than 35 Celsius. This can be as many as 30,000 cycles and as low as a few hundred power cycles in any given test run, it seems to be a very random event.
The TMP75 is starting with its default values and so it would require the measured temperature to exceed 80C for the Alert pin to become active. No I2C reads/writes or MCU activity to the device take place until after this Alert Event has occurred.
I’ve attached a plot of the captured Alert Signal (Blue and Red Traces) to this thread below.
Would it be possible for the ADC inside this device to pass on a bad reading on rare occasions (due to environmental noise) and cause the Alert to be active for the 1st sample after power up?
Thanks
Hi Amit,
Thank you for your reply.
1) The readings (MS Byte) I get back from the 4 sensors after I see a power up alert are: 0x1F, 0x1D, 0x1F and 0x7F (31C, 29C , 31C and 127C)
One of the sensors is getting a full scale reading after power up causing the Alert to be triggered and this is consistent. The Config Register should be in the default state but I haven't captured this yet.
2) The power supply ramp up time is 2ms from 0V to 3.3V and there is a 0.1uF decoupling capacitor on the supply pins of each device with 0.6mm pad to pad spacing.
I do see some supply noise when the rail reaches ~2.6V as another supply rail is turned on. The 3.3V drops by 130mV for 100us.
The below table shows the I2C reads from the temperature register. I am also changing the config to Interrupt Mode (TM=1) to clear the Alert pin signal so a software workaround is possible.
Thanks
Ian
Hi Amit,
Yes we're now issuing an I2C general call reset which clears the alert condition and then setting the Fault Queue to 6.
This has resolved our issue with the TMP75.
We're looking to use the TMP1075 for future designs
Thanks for your help
Ian