• State TI Thinks Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 2 answers
  • Subscribers 31 subscribers
  • Views 1495 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

TMP117: TMP117/ TMP117EVM connection timeouts at 60 or 90° process temperatures

Chester Medina
Prodigy 70 points
Part Number: TMP117
Other Parts Discussed in Thread: , LMT70

Greetings Everyone,

We're evaluating the TMP117's  accuracy in our thermal baths using the TMP117EVM. We're encountering a device disconnection/ connection time out from the TMP117 itself towards the EVM and or the GUI software. Out test samples have your TMP117 (both WSON and BGA packages) mounted on a flexible film which is then glued inside a stainless steel probe, welded shut. The tip of this said probes were submerged in thermal baths (distilled water).

At 23°C, the device functions as intended.

At 60°C, we can get intermittent disconnections, but we cannot reproduce the problem reliably.

At 90°C, the TMP117 disconnects from the EVM/ GUI software almost instantly (within 4 seconds) across all 10 sensors we have built. We can delay the disconnection by increasing the conversion averaging and conversion cycle, but the most we can get is 16 seconds before it inevitably disconnects as well.

Worst of all, even at its one-shot (shutdown) mode, the sensor would still disconnect. I'm assuming that the "Hardware not Connected" message shown on the lower left corner of the GUI whenever the problem occurs indicates a disconnection of the TMP sensor, not the EVM.

We're seriously hoping that this is a software issue and not the TMP117 itself. We're using the GUI software version 1.2.1 for the TMP117EVM. We did not see any mention of thermal shutdowns on the datasheet (just SHE effects). If this is indeed a thermal shutdown at 60/ 90°C then this sensor would be of no use to us.

Any help is appreciated!

/Ches

  • 0
    TI__Mastermind 44085 points

    Hi Chester,

    The GUI software will disconnect if there is a NACK from the TMP117 device on the I2C bus. It will also disconnect due to issues with the MCU or USB interface; the reported message does not distinguish between these issues. 

    I suggest using an oscilloscope to analyze the I2C bus at the time the failure happens. That will get us closer to root cause. We don't expect the TMP117 itself to have issues communicating at higher temperature; the device is specified to operate up to 150C. The USB dongle itself is not designed to operate beyond 85C. 

    thanks,

    Ren

  • 0 in reply to Ren Schackmann
    Prodigy 70 points

    Thanks for the quick reply Ren.

    I'll hook up a scope on the SDA line once the same thermal baths I used originally is free for me to use. I'll get back to you on Friday.

    /Ches

  • 0 in reply to Ren Schackmann
    Prodigy 70 points

    Hello again Ren,

    Here's what I've found so far. So just to be clear, at room temperature (both ambient air and thermal bath) the sensor works great and would run forever without any hiccups.

    Below is the SCL and SDA lines at channels 1 & 2 of our scope respectively at room temperature.

    When the connection "times out" we get this message at the lower corner of the GUI and both the SDA and SCL lines were flat at 3V.

    Occasionally we'll observe identical "noise" on both the SDA and SCL lines popping up every now and then, shown below:

    My setup is shown below... (the EVM to the right is for your other product, LMT70 and is not powered up during this test).

    Any ideas on what could be causing this problem? It bugs me that the sensor works completely fine at room temperature.

    /Ches

  • 0 in reply to Chester Medina
    TI__Mastermind 44085 points

    Hi Chester,

    The first oscilloscope picture shows normal communication; thanks for retrieving this. There is almost a charging effect as the ground level decreases throughout the transaction. It suggests there is some sort of reactance to the controller's ground path relative to the oscilloscope. The second oscilloscope picture shows a whole lot of nothing! It's interesting that there is some noisy activity for about the same duration as the valid transaction. Keep in mind I2C idles high, and relies on pull-up resistors to create logic high. The I2C devices are only capable of sinking current to ground to create logic low. It would seem there is a short that is keeping the controller from pulling down the line and creating the lows of the transaction. Is it possible there is some mechanical stress that is shorting your SCL/SDA lines to supply? I would also expect leakages to increase with temperature, but not enough so to win against the ~3mA driver. 

    Thanks,

    Ren

  • 0 in reply to Ren Schackmann
    Prodigy 70 points

    Hi Ren,

    It is interesting as you've said that the I2C "noise" had the same duration/ period as normal communication. We are also leaning into mechanical stress... below is our layout for the BGA package of the TMP117, track widths are 0.2mm. I have given it plenty of slack to compensate for any expansion/ contraction at higher/ colder temps.Just want to post it here and see if you see any red flags.

    It is also important to note that we don't have any underfill or glob topping material on these samples we've tested...

    Thanks

    /Ches

  • 0 in reply to Chester Medina
    TI__Mastermind 44085 points

    Hi Chester,

    The layout looks pretty clean to me; thanks for sharing. Having your SCL/SDA routed next to one another will cause the edges of each to appear on the other as a glitch. If possible, place a ground or supply trace between them. At any rate, this is not the cause of your issues, because it would take an extreme case of coupling to cause communication failure. The transaction would still be visible, but it would be interpreted incorrectly.

    Underfill or 'glob top' may exert stress on the temperature sensor causing a slight shift in temperature error. The errors we've observed due to package stress have been less than 0.1C, and could be calibrated out afterwards.

    Thanks,

    Ren

  • 0 in reply to Ren Schackmann
    Prodigy 70 points

    Hello Ren,

    We would like to see if we can make the prototypes work before we do another round of film prototypes. Is the accuracy measurements given in the TMP117's datasheet measured with glob topping/ underfill material? We're also planning on firing up our oil baths at 125 and 150°C process temps to see if we can exacerbate the connection issue.

    We're also going to replace our eval board as a sanity check.

    Thanks,

    /Ches

  • 0 in reply to Chester Medina
    TI__Mastermind 44085 points

    Hi Chester,

    Our specification can't cover stresses outside our control.

    Let me know of further developments with your connectivity issue.

    thanks,

    Ren