LT1129 SMD Thermal resistance

[741]

I'm confused by the DS. Rjc is not given for the SMD parts and the existence of a large reverse side copper area, with thermal conduction only through the PCB substrate is assumed.

What I'd like to find first is just Rja for a front-side "SMOBC" copper-foil heatsink of some given area. The Rjc value would also be nice to know.

I came across a document (http://datasheets.chipdb.org/IBM/x86/486/40011.PDF) where equation (2) seems to allow Rja to be calculated.

The formula given is
   Rja = (Tj(max) - Ta)/P,
with Tj(max) = maximum permitted junction temperature, Ta = ambient temperature and P = application-specific power.

The equation has the right form, but I don't yet understand the reasoning. It seems to assume that for any applied power, the junction will always hit Tj(max)

[Kleinstein]

For the equation to work the power and T_junction have to be for the same case / conditions. So either both for the application case or both at the maximum and thus more a test case to determine Rja, not  mormal use (maybe using the maximum clock for a CPU).

[AnalogTodd]

For any surface mount package, heat dissipation occurs mostly through the board, with the device transferring heat down through the leadframe and into the copper on the board and then spreading it out. There are a huge number of variables that come into play to determine net thermal resistance from junction-to-ambient. Copper thickness and area that the device is mounted to are primary, but then you can add in vias (thermal vias make a huge difference), number of layers on the board (and thicknesses of each copper) and the thickness of the board material itself (most PCB material is a poor conductor of heat but when it is thin and there is a big area, net thermal resistance drops). Surface mount devices rarely have heat sinks available for them, which is why you see the tables using board area and copper area to show thermal resistance as opposed to the TO-220 which is meant to mount to a heat sink.

If you want some basic package thermal information, there is a page available on the website under 'Design Center' -> 'Package and Layout Resources' -> 'Thermal Resistance Table' or https://www.analog.com/media/en/package-pcb-resources/package/thermal-table.pdf.

With regards to the equation you showed, it only works if you know you are operating the device at Tj(max) and therefore know the Tj. I've used that equation to calculate the thermal resistance of a part on a board many times, the only sticky part is ensuring that you absolutely KNOW (within a couple of degrees) what the junction temperature is. The way we would do it is to take a part like the LT1129 and run it in a minimum power situation (minimum input voltage and no load, total power dissipation in the uW range). We would then place the board in an oven and slowly increase the ambient temperature until the thermal shutdown of the part was triggered (the part must have a thermal shutdown for this to work). It has to be done slowly as thermal time constants can be long (tens of seconds to minutes) for a board and package, and you have to have the board and part fully in ambient conditions (suspended from thin wires or similar so it doesn't have anything it can transfer heat into such as sitting on a solid surface). You then know a fixed temperature that Tj will be at and can now repeat the experiment except (same board and part as the thermal shutdown temperature can vary) with some significant power dissipation (1W would be typical) and slowly increase ambient until thermal shutdown is once again triggered and therefore calculate the theta-JA. Do be aware that you must also avoid airflow across the board for measurements as this changes the thermal resistance.