I don´t know how much they improved, and I think emissivity is still a problem. We use the camera to find hot spots to which we then attach thermocouples. My point was to ilustrate that a IR reading is usually not the same as a contact reading.

The case might be partially transparent for IR, as you can clearly see the shape of the die, but another factor will most likely be the high emissivity of the black matte chip surface.

 

If the FET is a more conventional package perhaps you could measure the case with the thermal couple then look at the die thru the top of the package and see how close it comes to the data sheet. I have no personal experience with the IR measurement of the black package, but would love to know how close it comes.

 

89% is low for a modern SMPS buck with synchronous rectification! I would expect 93-96% at mid currents.

It's very possible my figure is off by a few percent. In truth it's a WAG based on input and output power and gross assumptions about relative losses in the converter, FET and inductor.

My suggestion is to question the datasheet inductor size suggestion and see if you can go up a bit in inductance, and then increase the timing cap to reduce operating frequency. Push the frequency as low as you can without the inductor saturating (you will easily see that on the 'scope).

I can certainly try a higher value inductor, but Fsw is fixed; there's no timing cap. I've been through the ringer over the last couple of months getting the switching node to run without ringing and overshoot. Is an inductor change likely to require snubber changes as well?

And changing from 1oz to 2oz copper will make a very significant difference in how cool the SMD switching devices run, especially if you have >= area to the recommended cooling pad area.

That's excellent news. I had expected little improvement. My board is 3.125in\(^{2}\) and the manufacturer's demo board is 5.45in\(^{2}\). Both are two-layer and 1-oz copper. I believe my board has more unbroken copper area, both top and bottom, for thermal relief of the converter, but that clearly doesn't overcome a 2.3in\(^{2}\) difference in board size. In identical configuration, the demo board runs ~15-20°C cooler than mine.

There's a more thermally efficient version of the converter (7.5°C/W θjc and 60°C/W θja in a DFN-14 package) that I plan to use for production, in addition to the change to 2-oz copper.

Thanks for your suggestions!

 

Re temperature measurements, I have one thermocouple and one IR gun. Neither are calibrated instruments. Thermal imaging cameras are way beyond my R&D budget!

 

In my shade tree mechanic world, I make my own standards. For instance, using a bench vise and exercise weights to calibrate a torque wrench. My favorite way to calibrate temperature is a hollow core ceramic resistor with the thermocouple inside it. That's how I test leakage currents at elevated temperatures. Put your chip in the "furnace" of your choice, wait until the temperature has been stable for a minute or two, and and shoot it with the IR gun. That's as close to calibration as I can get. It might be enough for you. Just don't tell your boss. He/she might laugh at you.

 

...
I can certainly try a higher value inductor, but Fsw is fixed; there's no timing cap. I've been through the ringer over the last couple of months getting the switching node to run without ringing and overshoot. Is an inductor change likely to require snubber changes as well?
...

It's unusual to have no control of freq. If there is not a specific timing cap then often the circuit will switch based on inductor charactersitics, so increasing the inductance may automatically lower the swithing freq.

Re the snubber, that's based on the inductor current at turnoff usually. So with different freq BUT the same duty cycle and the same average and peak inductor currents you can probably run the same snubber.

Of course with SMPS design it always comes down to the testing (as you know).

 

My favorite way to calibrate temperature is a hollow core ceramic resistor with the thermocouple inside it. That's how I test leakage currents at elevated temperatures. Put your chip in the "furnace" of your choice, wait until the temperature has been stable for a minute or two, and and shoot it with the IR gun. That's as close to calibration as I can get.

Great technique! Can you point me to an example of a resistor you might use? I presume it has to be fairly large. I'm picturing those massive 4Ω/500W wire-wound jobs the car audio guys use as dummy loads. Probably not what you have in mind.

It's unusual to have no control of freq. If there is not a specific timing cap then often the circuit will switch based on inductor charactersitics, so increasing the inductance may automatically lower the swithing freq.

All the datasheet says about freq is this:

Oscillation frequency.....600kHz
Short circuit osc freq.....190kHz​

and this:

The internal oscillator runs at a nominal frequency of 600kHz and can be synchronized by an external clock in the range between 300kHz and 1.5MHz from EN/SYNC pin.​

So, I suppose I could put a crystal or clock chip on the board...

Re the snubber, that's based on the inductor current at turnoff usually. So with different freq BUT the same duty cycle and the same average and peak inductor currents you can probably run the same snubber.

Nice. I don't relish going through that process again soon.

Of course with SMPS design it always comes down to the testing (as you know).

And sacrificing chickens under a full moon.

 

I use a 4 ohm, 35 watt resistor because it is more than 1 inch outside diameter, several inches long, and because I have no idea where it came from, so I probably found it in a junk box for free.