LM73: Temperature freezes at -32°C due to improper sensor reset

Hi Stani,

1) Is it correct to say that  the issue of this fixed constant is only found on one LM73 part? Are the other LM73 units that show this behavior fixed at -32°C? What is the temperature measurement they show?

2) As discussed in the previous thread— this is a POR issue. The section you described is the mechanism used to avoid errors.

Jalen

Hi Jalen,

please see the customers comments below:

1) Is it correct to say that  the issue of this fixed constant is only found on one LM73 part? Are the other LM73 units that show this behavior fixed at -32°C? What is the temperature measurement they show?

No, there were already several occurrences captured with different units. And even on the affected units, the issue is sporadic. It seems that value -32°C was wrongly interpreted by our debug signals, it was rather some excessive value (>200).

The critical question for us is to understand where the problem could arise. Below there is an example of the supply voltage (blue - 3V3) of LM73 during the whole system startup:

We have removed the second drop (emphasizes in cyan) from our system behavior-> however the issue was popping up again recently. So I believe that the ramp up after the first drop is the root cause. Can you advise me in more details when the issue can happen? Datasheet is mentioning only slow ramp up time, but it is not clear to me whether it refers to the interval between 0 - 2.7V or x - 2.7V (where x is a value below the min operating voltage)?

2) As discussed in the previous thread— this is a POR issue. The section you described is the mechanism used to avoid errors.

I understand, so if it is impossible to include this mechanism, there is no other solution to avoid this issue, right?

Best regards,

Stani

Hi Jalen,

the customer has reached out with some additional questions:

-------------------------------------------------------------------

we are still struggling with the LM73 improper reset during our system startup. Unfortunately, any HW change is not possible at this moment and we are not even able to implement the soft reset accordingly with the datasheet as the LM73 is controlled by FPGA which design is frozen at this point. So, to summarize, we know that we are in a situation where the voltage supply of LM73 drops below the min operating voltage of 2.7V and it is reactivated afterwards. So essentially we are facing a POR issue. Is it possible to give feedback on the following questions?

1) Would it help to leave the voltage to drop below some critical value to guarantee that LM73 would not stay in an unknown mode (e.g. let the power supply to drop below 0.5V or less for 100ms?) and so LM73 would then start up normally if the linear ramp-up >0.7 V/msec will be guaranteed from that point? 

2) In the referenced discussion

it is mentioned that even faster ramp up time could be problematic (contrary with the datasheet). Can you confirm this?

3) Is the 0.7 V/msec ramp up time required for the whole range of the rising power supply voltage (0-2.7V) ? We have confirmed some short-time period of power supply ramping up with a slower ramp up -> see the figure below (blue curve is LM73, critical point is emphasized in yellow). NOTE: the figure is from OK start up, I do not have yet the oscilloscope measurement from the faulty unit

----------------------------------------------------------------------------------------------------------

Best regards,

Stani