Measuring Source Pin Temperature
As part of the design check-out procedure for a Tiny Switch Power Supply Design there is a suggestion to measure the source pin temperature and ensure it is at or below 110C for lowest input voltage and max continuous load.
Are there any suggestions on how to best get this measurement? Is a glued thermal couple the best option or can IR camera get better results. If the IR camera is used should the pins be coated in material that helps with the camera temperature detection or will the camera typically work ok with the standard pin finish? Is there an opportunity for electrical noise if a glued thermal couple used that can skew the measurement?
We've mostly run this test with glued thermal couples. I'm seeing some behavior that I don't quite understand. At a 95V DC input to the circuit and a fixed load we see a temperature near 50C, then increasing the input voltage to 270VDC with the same fixed load we see the temperature rise to 60C, then increasing the input voltage further to 400VDC we see the temperature decrease to 52C. I'm surprised to see the higher temperature in the middle of voltage range. This example data is using a 275. Any thoughts?
I'm looking for ways to decrease this rise a bit. I've looked at using higher part number Tiny Switches (started with a 274) and have been testing 275 and 277. I'm actually seeing the highest temperatures with the 277 - which also doesn't make sense to me. We've also looked at the drain current and over a long period of time (100's of ms) the average drain current does indeed continue to go up as the voltage goes down - which further confuses me on the temperature readings.
I know I can increase the heat-sinking copper area, but I'd prefer not to use that path. At the moment the pattern of the data is more concerning than anything. How do I get higher temperatures at voltages in between the max and the min? And is my measurement technique ok or is there a best practice for this?
Comments
Does the 110C target value for good thermal performance also apply to the hottest part of the top of the chip on the ceramic surface area or is there a different target value if that measurement is used instead of the source pin?
It applies to the hottest spot
It applies to the hottest spot
Just to make sure I understand correctly. You're saying that 110C is the target temperature to stay at or below under worst case conditions and that temperature can be either the source pin or the hottest part of the package? The 110C target value is for anywhere on the IC, wherever that hottest spot occurs?
There is only one hotter spot in the chip. It is the power mosfet. If you measure its temperature with camera, you can sense the temperature on the package. But if you are attaching an thermocouple, it is better to attach it to the source pin
If you have both, thermal camera and thermocouples, then you can take simultaneous measurements and see if you are reading the same temperature. When using the camera, you can cover the metal surface with a black marker ink. But also you can look for the hottest spot in the IC surface. Usually the hottest spot is on the top of the chip on the ceramic surface area.
About thermocouples, sometimes they pick noise from the circuit and you get wrong readings. One way to check how much your reading is being affected is by leting the power supply to run with the thermocouple connected. Let the circuit to settle to some temperature. Then abruptly turn the input voltage OFF. If you see an abrupt change in your thermocouple reading, then it means that you are getting much noise from the circuit. If you see the abrupt change to be about 2 degrees, then it means the reading is acceptable. One way to filter the noise is to wound the thermocouple wire inside of a ferrite core. This way you have an inductor between the temperature meter and the end of the thermocouple. This inductance filters the noise
We usually solder the thermocouple to the metal tab of the IC. If you use glue, then make sure you use thermal epoxy.
About thermal cameras, we had found that low cost cameras are not accurate instruments. So if your thermal camera is a low cost product, then better to use thermocouple.
I think you need to figure out if your instruments are accurate and decide how you are going to measure your temperature
About the reading of your temperatures with different input voltages, It is not impossible that your temperature changes as you describe. This could be because the switching losses in the circuit. But I also it can be due to the noise in the circuit affecting your reading if you are using thermocouple.