We have an extremely weight-sensitive application for the TX2, and would like the option of integrating the thermal dissipative qualities provided by the TX2’s TTP (Thermal Transfer Plate) into the design of our enclosure, such that the TX2 module could be directly coupled to the (milled aluminum) enclosure.
The question of whether the TTP can be removed has been addressed in these forums before by another user of the TX1, with an Nvidia moderator concluding that it could not be done without voiding the warranty – which makes sense, given that the devices are extremely thermally-vulnerable. Typically, we would address this situation by designing our own board around the necessary Nvidia chipset, and developing a thermal dissipation solution from there – however, for those of us building devices of limited quantity, it’s nearly impossible to obtain the Nvidia chips that we would normally use to design our way out of the situation. So we find ourselves stuck between a rock and a hard place :)
Our engineers are capable of devising a solution that provides equivalent power dissipation, thermal resistance, and mechanical mating characteristics (downward force, spread, etc). Is there any way that Nvidia would work with us to “certify” our enclosure/heat-sinking implementation such that we could still retain a warranty for the module? If not, is there any way for us to procure a reference design and parts in small volumes, so that we could design our own solution?
I have a similar issue; I have an embedded application which will be conduction cooled very similarly to the description of the thermal plate in the TX2 thermal design guide. We however have strict component height limitation and the TX2 is too high. We would like to use it without the TTP. Our thermal plate will cool not only the TX2 components but other components on our board as well. I understand that you do not have a team to support this application and its risk. We will do the thermal analysis and mechanical design of our plate, and TIM selection. All we need are a few details shared with our engineers such as a mechanical model or drawing with the different component heights. And most importantly the power dissipation and temperature limits of the critical TX2 components. If you could not supply us with a tested TX2 assembly before the TIM and TPP is applied, understanding that your warranty would be voided, we could remove the plate and clean the TX2 assembly before assembly onto our board. Would Nvidia be willing to share this information (under NDA of course) so that we could move forward with our application?
Thank You
Bob W