Spray Cooling of IGBT Devices

[+] Author and Article Information
Robert G. Mertens

 GammaPC, 2813 Kilgore Street, Orlando, FL 32803-6436gammapc@bellsouth.net and bob.mertens@gammapc.com

Louis Chow

College of Engineering and Computer Science, University of Central Florida, Orlando, FL 32816lchow@mail.ucf.edu

Kalpathy B. Sundaram

College of Engineering and Computer Science, University of Central Florida, Orlando, FL 32816sundaram@mail.ucf.edu

R. Brian Cregger

College of Engineering and Computer Science, University of Central Florida, Orlando, FL 32816bcregger@bellsouth.net

Daniel P. Rini

 Rini Technologies, Inc., 582 South Econ Circle, Oviedo, FL 32765dan@rinitech.com

Louis Turek

 Rini Technologies, Inc., 582 South Econ Circle, Oviedo, FL 32765louis.turek@rinitech.com

Benjamin A. Saarloos

 Rini Technologies, Inc., 582 South Econ Circle, Oviedo, FL 32765ben.saarloos@rinitech.com

Turn off times for IGBTs are typically much longer than turn-on times, often twice or three times longer. Switching times can vary from 20nsto450ns depending on the device, gate voltage, collector-emitter voltage, collector current, and junction temperature. A typical switching time for an IGBT is 75ns. A typical switching time for an IGBT module is 200ns

IGBT off voltages can be in the hundreds and thousands of volts, whereas the on voltage is generally around 2.5V—easily ignored and set to zero in this part of the calculation.

A duty cycle of 0% would not have any switching losses, but the calculation here assumes that switching takes place. This shows the heat flux due to switching without losses from conduction. This particular IGBT device is designated “ultrafast” by the manufacturer, which considerably reduces switching losses. Other IGBTs are not as fast, and switching losses are higher.

This was not a measured parameter but taken from known statistics on distilled water measurements. Selectively distilled water is said to have an electrical resistance of about 1MΩcm(11), while the most ultrapure (selectively distilled and de-ionized) water ever tested had an electrical resistance of 18.2MΩcm(12). Only a small amount of dissolved contaminant (especially salts) can cause water to become electrically conductive.

The constant current source actually requires about 2.5ms to settle. After this, the temperature data are relatively flat, as observed in the test set, indicating that the internal device temperature of the DUT does not change significantly during the read cycle.

The “3Y” identifies which IGBT DUT was being tested—kept here for future reference to the corresponding data set.

In fact, many IGBTs are specified to operate at their maximum ratings at 25°C and get severely derated at values nearing 125150°C. In many of these experiments, the IGBTs were operated up to (and occasionally beyond) the 150°C mark, at well over their rated current.

This is a theorized cause of failure, although heat stress failures in IGBTs are primarily due to thermal gradients caused by an uneven distribution of current flow (which causes heat) in the junction.

J. Electron. Packag 129(3), 316-323 (May 18, 2007) (8 pages) doi:10.1115/1.2753937 History: Received May 30, 2006; Revised May 18, 2007

The popularity and increased usage of insulated gate bipolar transistors (IGBTs) in power control systems have made the problem of cooling them a subject of considerable interest in recent years. In this investigation, a heat flux of 825Wcm2 at the die was achieved when air-water spray cooling was used to cool IGBTs at high current levels. The junction temperature of the device was measured accurately through voltage-to-temperature characterization. Results from other cooling technologies and other spray cooling experiments were reviewed. A discussion of electrical power losses in IGBTs, due to switching and conduction, is included in this paper. Experiments were conducted on 19 IGBTs, using data collection and software control of the test set. Three types of cooling were explored in this investigation: single-phase convection with water, spray cooling with air-water and spray cooling with steam-water. The results of these experiments show clear advantages of air-water spray cooling IGBTs over other cooling technologies. The applications of spray cooling IGBTs are discussed in open (fixed) and closed (mobile) systems. Current and heat flux levels achieved during this investigation could not have been done using ordinary cooling methods. The techniques used in this investigation clearly demonstrate the superior cooling performance of air-water spray cooling over traditional cooling methods.

Copyright © 2007 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Figure 1

IGBT switching losses

Grahic Jump Location
Figure 2

Heat flux versus switching frequency for various values of D (duty cycle) for a typical IGBT

Grahic Jump Location
Figure 3

Diagram of the spray cooling apparatus

Grahic Jump Location
Figure 4

Total resistance in the test set due to IGBT and water from the spray cooling of an IGBT base plate

Grahic Jump Location
Figure 5

Typical IGBT V-T characterization plot

Grahic Jump Location
Figure 6

Test IGBT from Fairchild Semiconductor Corporation (top (left) and bottom views)

Grahic Jump Location
Figure 7

Air-water system

Grahic Jump Location
Figure 8

IGBT 3Y, HD test, comparison

Grahic Jump Location
Figure 9

HD test, air-water, heat flux versus temperature

Grahic Jump Location
Figure 10

Open (fixed) IGBT spray cooling system

Grahic Jump Location
Figure 11

Closed (mobile) IGBT spray cooling system




Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In