• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 201 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.

TMP119: Query regarding mechanical mounting of TMP119 for good thermal path with sensed object

John Ash
Prodigy 40 points
Part Number: TMP119
Other Parts Discussed in Thread: TMP118, , , TMP117, ISOTMP35

Tool/software:

I am using a competeitor's device to measure the temperature of a (largely) metallic block. It's a WSON type pacakage and the board is mounted upside down so that when secured with a blob of thermal epoxy, the top of the IC is in contact with the block. We have had issues with this design (for different reasons other than the mounting method) which has led me to experiment with the TMP118/9EVM boards and they do not show the same issue which is great, but I am not sure if we can use the same technique for mounting as we currently do.

I have read SNOA986A and that is based upon the mounting of the PCB the right way up, so that the PCB layer is between the IC and the sensed object which seems as if you're building in errors from the start which need to be overcome.

I would like to mount it in the following way - is there a reason why this would not be considered appropriate?

We have been using this technique without issue to date with our existing IC.

It'd be great to discuss this with an FAE if that were possible?

Thanks

  • 0
    TI__Genius 9670 points

    Hello John,

    Thank you for posting on the e2e forum. I have connected you to our FAE manager through email to guide you to a FAE. In the meantime, I would like more clarity on what the block is. I'm assuming you are wanting to measure the temperature of the block. What material the block is made out of? What is the voltage range? What is the high voltage heat source?  Is there any pressure on the temperature sensor?

    Best Regards,

    Meredith McKean

  • 0 in reply to Meredith McKean
    Prodigy 40 points

    The block is 80% iron, coated in epoxy and teflon. We abrade the surface to aid adhesion provided by the thermal epoxy. Response time is not (that) critical; I have read the app notes on this issue and am comfortable that the sensor will be able to respond at least as fast as the block will. There are no voltages associated with the block. There is no pressure on the sensor either. 

    Since I posted my orginal query I have had a TMP119 (courtesy of the TMP119EVM) attached to a block as depicted, but via 10 deg C/W.m thermal paste (sufficient to gap fill and provide contact with the block) rather than epoxy and have had that running over the weekend. It performs as expected and is not susceptible to the issues we have seen with the device that we currently use.

    So my current thinking is that with no copper on the back of the board (top as depicted above) I can 'focus' the thermal path to the block. I have some concerns about the mechanical aspects of bonding via epoxy, I do have an option to use a thermal silicone instead which I believe would minimise the stress which could be associated with the use of epoxy as the bonding method and it's this which I'd appreciate some FAE input on.

  • 0 in reply to John Ash
    TI__Mastermind 43725 points

    Hi John,

    You've reached the temperature experts. Meredith and I work in the Temperature and Humidity Sensing product line within TI. We sit right next to the people that develop the temperature and humidity products. It's our job to support TMP119 customers to enable their success with these products. We don't support any other products in TI's portfolio. Why am I telling you this? For TI, the Field Applications Engineer (FAE) role is one that is co-located with the customer.  They have to support all the TI products that customer uses. They can't be subject-matter-experts on individual products, but sometimes they are. E2E is unique in that it connects you directly to the subject-matter-experts for a given TI product. Having said that, there are still limitations to our knowledge of temperature sensors. ;)

    I'm glad that you're happy with your TMP119 and TMP119EVM. (The EVM hardware, accompanying software and documentation is the responsibility of my team.) I'm also glad you read our app note SNOA986 and understood it. I agree that bonding the device to the block is a better thermal solution than using the PCB as an intermediary thermal conductor. Here are my concerns with this approach, and I apologize we don't currently have a published app note that addresses this topic:

    • Block/Sensor Assembly Process
      • If you are bolting the PCB to the block, or pressing down on it during epoxy cure, we don't currently have a clear specification from our IC packaging team on allowed clamping pressure. There are packages that are designed to be clamped to a heatsink, but our sensor packages are not. 
      • The TMP118 and TMP119 are DSBGA packages that could chip or crack. They are not encased in black plastic "mold compound." Most ICs are mounted on a leadframe, the pins are "bonded" using fine wires and then the whole thing is covered in black mold compound. The WSON that you used previously is created this way. The mold compound provides some protection to the silicon die, which the TMP118 and TMP119 do not have. The TMP117 is a member of this sensor family available in a WSON package. However, the lack of material on DSBGA packages does have advantages in size, thermal response and cost.
    • Block Voltage Potential
      • I realize you don't expect to have this issue in your setup, but I wanted to touch on it anyway. There are two risks here: one is voltage from the block arcing to a sensor pin, and the other is voltage from the block arcing through the mold compound to the silicon die.
      • For voltage arcing to a sensor pin, it should be fairly clear that industry standards regarding clearance/creepage need to be followed (specifically clearance.)
        • Even if the block is not designed to have electrical potential on it, it can still act as a conductor for ESD. It may need to be grounded to create a safe dissipation path, and further measures may be needed to protect the sensor from ESD strikes, such as standalone ESD devices. http://ti.com/esd
        • If you need to make an electrical connection for a thermal monitor (such as at a MOSFET tab,) our ISOTMP35 has a 3kV voltage isolation rating that allows it to be connected electrically.
      • Similar to clamping pressure, there isn't a clear specification for how much voltage isolation our packaging can offer the silicon die itself. Note that the DSBGA devices don't have mold compound at all, and may be more susceptible.
    • Temperature Accuracy in the face of Mechanical Stress/Strain
      • Like other precision analog IC designs, it is known that mechanical stress can alter the sensor's temperature output to some degree. The effect often goes unnoticed until a very high degree of accuracy is achieved, such as the 100mC accuracy achieved by these products. 
      • The stress can come in the form of tension/bending of the PCB, or it can come from epoxy surrounding the IC pressing on it. In the case of epoxy or pastes surrounding the IC, the stress may change over temperature due to the epoxy's CTE (coefficient of thermal expansion.) It's for this reason that epoxies and mold compounds used on ICs should have a low CTE. A thermal paste which does not cure may not present any stress or CTE concerns, but these products are only intended to fill small voids between heatsink surfaces. Such a paste may not be very useful in a sensor application as it cannot be used to secure the sensor.
      • TMP118 and TMP119 were specifically designed to minimize errors due to mechanical stresses. Refer to the Strain Tolerance sections in those datasheets. There you can see that even a device that was not designed to be strain tolerant is still only expected to produce less than 80mC error in the face of mechanical stress. It's for this reason that mechanical stress may not be a concern, depending on your error budget.

    thanks,

    ren

  • 0 in reply to Ren Schackmann
    Prodigy 40 points

    Ren,

    Thanks for your full and detailed reply - it's much appreciated! 

    Taking into account all you have pointed out, I am going to proceed with some trial builds using the TMP119 in the manner I have described (i.e. package down). I didn't make clesr in my orginal post, but there is no direct fixing between the PCB and the block, replying instead upon the thermal adhesive which surrounds the device to fix it to the block. 

    I will be trialling a thermal silcone adhesive in a bid to avoid the issues associated with epoxy and the stresses that may be induced through differing CTEs.

    We currently use a magnet on top of the PCB (opposite side to device) to keep it in place while the adhesive sets. I don't want to place too much pressure on the device, so I'm thinking about having 3-4 components (with a height slightly above the TMP119 but not electrically connected) around the TMP119 to provide mechanical 'stops' to prevent the device being pressed into the block too much by the magnet during the setting process. That would then allow the thermal silcone to set, providing a good bond from the device package to the block.

    Thanks again for your input, Temperature Experts!

  • 0 in reply to John Ash
    TI__Mastermind 43725 points

    Sounds great, John. Let us know how it goes.

    ren