WEBINAR: Thermal Management Guidelines for GaN Transistors

With Dr. Roy Hou, Technical Marketing Director at GaN Systems

GaN Systems is back with another of their highly successful webinars, this time it’s all about thermal management.

In This Webinar You’ll Learn:

  • The performance impact of thermal management
  • What the most important measurements and calculations are
  • How to maximize performance with GaN Systems top-cooled and bottom-cooled transistors

Questions & Answers

We provide two IMS designs with different sizes. Please refer to slide #20. You can also find the IMS user manual >

The GaNPX package is quite thin. Hence, we can measure the temperature on the very top of the bottom-cool device (or the very bottom of the top-cool device) and this package temperature is a good approximation for device junction temperature. The difference is within 1 degree C between the measurement temperature and the junction temperature.

Yes. With high current capability and ease of parrallelling devices, our devices have been applied in systems greater than 25kW.

In general, IMS achieves better thermal performance than top-side cooled, primarily because a TIM is not used with the IMS. However, the trade-off really depends on your converter system structure and its layout. Some users prefer top-cooled device, as its heat transfer path and electrical current path are also not overlapped. Refer to the IMS user guide >

To achieve best accuracy, we highly recommend to do double pulse test and measure the actual Eon and Eoff. A proper de-skew is critical on switching loss measurement.

Slide #25 and 27 show the thermal resistance comparison among different thermal designs, including top-cooling with thermal pad and bottom-cooling with IMS.

Yes, Eon/Eoff are dependent on drain current. Eon/Eoff can be estimated by SPICE models. Refer to our app note GN008 for details >

Yes. Our customers have developed up to 25kW inverters and have shown 3x power density improvement and up to 10% system cost savings.

You can use a thermal camera to measure the device temperature.

The detailed 4 stage junction-to-case thermal model is available in our datasheets.

The heat sink is not connected with the switch device. A TIM is placed between the heat sink and switch for electrical isolation. For increased EMI performance, you can the floating heat sink to earth.

Do you have a detailed CAD model of the GaN that can be used for FEA thermal simulations?

Yes, the parasitic capacitance between the switching node and base plate is ~100pF,  depending on the polygon area.  The effect on switching performance is negligible.  The effect on EMC also depends on the grounding of the base plate, but generally speaking, the EMI is similar to PCB based design.

For turn-off loss Eoff, it shall be only the VI overlapping loss. The capacitive Eoss loss actually is part of the Eon loss. GaN obtains low turn-off loss and gate driver loss. As you go to higher switching frequency, the losses that require attention are mainly the Eoff loss and gate driver loss.

Magnetic flux cancellation design is applied between the IMS board and the bottom layer of the driver board.  The distance between the two boards determines the cancellation effect. The gate loop inductance is ~6 nH in this design, with a 4.5mm distance between the two boards. It is slightly higher than the single PCB design (~4nH), but is still good for ~MHz applications. For more information, refer to this link >

IMS material can be attached to the heatsink directly. Refer to this link for more information about IMS-based design >

On the IMS evaluation board, we have its heatsink floating. To achieve better EMI performance, you might want to connect the heatsink to earth.

This is an application suitable for GaN. However, we do not have a publicly available grid-tie inverter reference design. We have published a technical paper regarding the IMS-based T-type inverter. Please refer to the paper with the link >

This is because for top-cool device, the GaN layer is already on the very bottom (near the PCB). Therefore, the Kelvin source on the package is not necessary. For external layouts, you may still want to layout a pseudo-Kelvin source to separate the driving loop and power loop.

This depends on your switching frequency, load current, and thermal conditions, etc. Generally speaking, for GaN-based hard-switching half-bridge, you can expect it introduces power loss in the range between 0.2% to 0.7%

Yes. We now offer 2 levels of SPICE model: Level 1 and Level 3. Level 1 is a simplified version and Level 3 offers detailed thermal model and parasitic inductances. The models are available in the product section of our website

We measure the heatsink temperature and then we can calculate the junction-to-heatsink Rth.

It is the maximum temperature in the chip.

We use a thermocouple to measure the temperature near the device for the ambient temperature measurement

Higher power (higher current) devices for 48V applications is part of our roadmap. Look for more information on this product in the short term.

Eon and Eoff are both low. Eon increases with increased temperature. Eoff does not change with temperature.

There are designs with up to 8 devices in parallel without issue. Limitation is not of concern. Designers do need to use good layout technique. Refer to our parraleling appnote >

They are quite similar. It really depends on how you want to cool the device, your system structure, and converter circuit layout.

We use FloTHERM for all the simulations in this webinar

The capacitive Eoss loss is listed on the datasheet and it occurs in hard-switching on. If the application is ZVS soft-switching, this loss is not in the system.

We use FloTHERM for all of the simulations in this webinar

No, the drain current does not impact the top side temperature reading.

It depends on the thermal capacitance value. From junction to case, it can be several micro seconds. To heatsink or ambient, it depends on the specific PCB, heatsink design, etc.

It is possible for all SMD power switches.

50 PSI is ~344 kPa, 100 PSI is ~689 kPa

Although not available today, this is a feature in our roadmap

The capped vias are under the thermal pad of the device

从本研讨会下载演示文稿

    Speaker: Ruoyu (Roy) Hou

    Power Electronics Application Engineer, GaN Systems Inc.

    Dr. Ruoyu (Roy) Hou is a Power Electronics Application Engineer at GaN Systems Inc. He received his M.S. degree from the Illinois Institute of Technology, Chicago, IL, USA and his Ph.D. degree from the McMaster University, Hamilton, ON, Canada, both in electrical engineering.

    Formerly an electrical engineer with GE Transportation, Dr. Hou was a post-doctoral research fellow at McMaster Automotive Resource Centre (MARC), a Canada-based Excellence Research Center. His interests include power electronics, modeling and loss analysis of wide-bandgap (WBG) semiconductor devices, and GaN-based high-power converter and its magnetic design.

    Dr. Hou was a recipient of the ECCE Best Paper Award in 2016 and a co-recipient of the Chrysler Innovation Award for the Automotive Partnership Canada (APC) project in 2014.

    Moderator: Paul Wiener

    VP Strategic Marketing at GaN Systems

    Paul Wiener is GaN Systems’ Vice President of Strategic Marketing. Prior to joining GaN Systems, Paul led the power magnetics business unit at Eaton. Paul brings more than 25 years’ experience in operations, sales and marketing, and business development. His experience includes vice president of sales at Fultec Semiconductor Inc. and several management roles at Genoa, BroadLogic, and Raychem.