OPA549-HIREL: Query on Theta JC

Hey Mark,

After some further investigation it appears as though the HiRel part makes use of a different, lower thermal resistance process compared to the original die. Also, the qualification of the two devices occurred many years apart and significantly different methods were used. Therefore, it's a combination of the HiRel part being inherently better than the standard part, and the different testing methodologies used. Hope this answers your question!

Cheers,

Jon

Hi Mark,

1.) After discussion with a colleague, it seems as though the HiRel θJC value was likely derived from simulation/modelling of that cold plate test rather than empirically. This is the general procedure that is used to measure this parameter:

• An IC package normally containing a thermal test chip is mounted on a test PCB. This is normally a JEDEC-defined low-k 1s0p PCB which has low copper content to minimize heat loss though the PCB.
• The package is pressure fit to a Cu cold plate (a copper block with circulating constant temperature fluid) with the leads facing up and the case against the cold plate when the top of the case is to be measured. Otherwise, a Cu cold plate contact to the bottom of the package is provided through the PCB when the primary cooling path of the package is through a soldered plate into the PCB.
• Silicone thermal grease or other thermal interface material provides thermal coupling between the cold plate and package.
• Insulation is provided around the test coupon to minimize parasitic heat loss.
• Power is applied to the device.
• The junction temperature of the test chip is measured. 
• The temperature of the package surface in contact with the cold plate is measured by a thermocouple or other temperature sensor pressed against this surface.
• RθJC is calculated by dividing the measured temperature delta by the dissipated power.

This scenario would be simulated and θJC derived from the result. The reliability report mentions a qualification of the thermal resistance of the part, but only θJA is specifically called out, so I cannot confirm that the θJC element was empirically measured as part of this procedure. I will mention however that a simulated cold plate will of course be more "ideal" than a real-world water-cooled cold plate.

2.) From my knowledge of the parts, it would seem that while the T/S parts could very well have slightly better thermal conductivity than the 1.4 C/W specified in their datasheet, I would not expect to achieve the 0.1 C/W thermal conductivity associated with the HiRel part. The modified process is definitely the primary contributor to this improvement, with the better testing/simulation playing a secondary role, but without having detailed simulation results for the T/S parts it's difficult to compare apples-to-apples here. My perspective is that your typical θJC will be better than the 1.4 C/W value, but from a process corner perspective, I would still use this 1.4 C/W value when calculating potential heatsink sizing, max power dissipation, etc just to be safe. If thermal conductivity is a significant concern I would definitely advise the use of a heatsink with the non HiRel parts as described on page 11 of that datasheet. Unfortunately I don't have enough hands-on experience with this part to comment on the merits of the T package vs S package.

Hope this helps,

Jon