Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > Power/Alternative Energy
Power/Alternative Energy??

Impact of GaN technology on EMI

Posted: 24 Jul 2015 ?? ?Print Version ?Bookmark and Share

Keywords:gallium nitride? GaN? MOSFETs? buck converter? EMI?

While attending DesignCon several months ago, I had the chance to hear an interesting keynote presentation by Alex Lidow, CEO of Efficient Power Conversion, Inc., speaking on the upcoming development in gallium nitride (GaN) technology for high-power switching devices. I also had the fortune of meeting Steve Sandler, author of the book, Power IntegrityMeasuring, Optimising, and Troubleshooting Power Related Parameters in Electronic Systems, who was associated with measuring the picosecond edge speeds of these devices (see his article in the References section).

Because of the fast switching speeds and related higher efficiencies of these new power switches, we'll expect to see them primarily applied to switch-mode power supplies and RF power amplifiers. They may broadly replace existing MOSFETs and have lower "on" resistance, less parasitic capacitance, are smaller, and faster. I'm already noticing new products using these devices. Other applications include telecom DC-DC, wireless power, LiDAR, and class D audio. Obviously, any semiconductor device that switches in a few picoseconds is likely to generate large amounts of EMI. In order to evaluate these GaN devices, Sandler arranged for me to test some evaluation boards. The one I chose to characterise was a half-bridge 1MHz DC-to-DC buck converter from Efficient Power Conversion (EPC9101, figure 1). Refer to the References section for additional information on this demo board, plus several others.

Figure 1: The demo board used to characterise GaN EMI. The GaN device is circled and I'll be measuring the switched waveform on the left end of L1.

The demo board takes 8 to 19V and converts it to 1.2V at 20 amps (figure 2). I ran it at 10V with a 10-, 2-W load resistor.

Figure 2: The block diagram of the half-bridge DC-DC converter. The waveforms were measured at the left end of L1 to return.

I tried to capture the edge speed by using a 1.5GHz single-ended probe (R&S RT-ZS20, figure 3) and probing at the switched end of L1, but the equipment available had too limited a bandwidth to capture it faithfully. The best I was able to capture (figure 4) was a 1.5 ns rise time (which, from an EMI point of view, is pretty fast to begin with!). To accurately record the typical 300 to 500ps edge speed would require an oscilloscope of 30GHz bandwidth, or more. For more on measuring the switching speed, refer to the article by Sandler in the References at the end of this article.

Figure 3: Measuring the leading edge using a Rohde & Schwarz RTE1104 oscilloscope and RT-ZS20 1.5GHz single-ended probe.

Figure 4: The captured rise time showing ringing at 217MHz. The fastest edge speed indicated was 1.5 ns, but in reality, the measurement was bandwidth-limited.

EMI emissions
While I wasn't able to capture the actual rise time, I did evaluate the ringing at a frequency of 217MHz. As you'll see later, this resonance will cause a peaking in the broadband EMI once we start looking at the frequency domain. Both the signal pin and the ground return connection to the R&S RT-ZS20 probe were very short, so the ringing was not due to the probe, but to parasitic resonances in the circuit. Depending on the robustness of the circuit design, this level of ringing would be sufficient to cause a failure in radiated emissions.

Next, I measured the conducted EMI on the power input cable and through the load resistor to characterise the conducted EMI (figure 5).

Figure 5: Conducted high frequency currents were measured with a Fischer F-33-1 current probe.

Figure 6 shows that there were very high 1MHz harmonics throughout the 9kHz to 30MHz conducted emissions band. These were riding on top of approximately 9MHz-spaced harmonics, of which I've not yet identified the origin. These were especially high on the load resistor circuit. I suspect this amount of EMI would likely fail the conducted emission compliance testing without good quality line filters.

Figure 6: The high frequency currents in the power input cable (violet trace) and the 10? load resistor (aqua trace) as measured with a Fischer F-33-1 current probe. The yellow trace is the ambient noise level. The 1MHz switching spikes are prominent, riding on top of about 9MHz harmonics. From my experience, the level of the aqua trace is very concerning and would likely fail a conducted emissions test.

1???2?Next Page?Last Page

Article Comments - Impact of GaN technology on EMI
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top