Technical Article

Avoidance of Reverse Recovery Ringing in Wide Band Gap Devices

May 07, 2016 by Neil Markham

This article discusses the benefits of nHPD2 packaging technology and the wider market need for operational high power wide band gap devices.

HITACHI highlights the necessity of High Power Density Dual nHPD2 packaging Attaining the highest levels of Silicon chip efficiency whilst paving the way to avoid reverse recovery oscillations commonly observed in WBG semiconductor modules.

2.46 million Google results, a US Presidential backing, and the European Union’s carbon emissions darling of the future. Savior of World? Who knows for sure, but Wide Band Gap (WBG) semiconductors continuously grab column inches in the specialist publications, the wider press, and the conference circuit. Investment is booming despite the wider macro-economy woes. A significant reduction in global energy consumption will occur, by WBG adoption in consumer switched-mode products for example, but the industrial Power Electronics sector has a demanding job ahead with one hand tied behind its back due to ringing behaviour in conventional packages.

The intrinsic behavior of the Wide Band Gap semiconductor with a unipolar structure offers significant energy efficiency improvements due to negligible recovery current. So what is the catch, the chip level reports look fine? The challenge is realizing laboratory level chip performance as a working solution at industrial power levels. This will typically require a power module which offers convenient mounting and isolation complimentary to various cooling solutions. Switching 1200A @1500V, for example, across a wide temperature range, whilst maintaining compliance with the Electromagnetic Compatibility (EMC) Directive 2014/30/EU, is one part of a multi-faceted challenge. Factor in WBG characteristics and managing high power within Directives becomes a significant challenge.


Recovery waveforms of conventional 1800A/3300V IGBT module with and without SiC SBD. Test conditions: 1800A, 1800V, 25°C, Lσ 90nH, Rg_on 4.79Ω
Figure 1: Recovery waveforms of conventional 1800A/3300V IGBT module with and without SiC SBD. Test conditions: 1800A, 1800V, 25°C, Lσ 90nH, Rg_on 4.79Ω


Suppressing recovery ringing

In Figure 1, the recovery waveforms of two conventional modules are presented. The right-hand side module adopts a Silicon Carbide (SiC) Schottky Barrier Diode (SBD) to form a “Hybrid SiC” module with Si IGBT. “Conventional module” refers to classic 190mm x 140mm packages, also known as IHM or HVIHM. Problematic oscillation is evident in the SiC hybrid device.

Whilst it is possible to mitigate oscillation without a significant impact to the losses using active gate control architecture, to monitor the diode current direction and to react in real-time to dynamically adjust the turn_on time, this can be a complex and expensive design process especially without costly investment. Controlling the oscillation by simple gate resistor tuning will significantly drive up losses and eradicate the WBG efficiency benefits. Other options also exist, but considering price-performance merits at a system level, it is worthwhile managing the system loop inductance as a viable alternative. With this in mind, the nHPD2 high power dual-module was developed.


LCR modelling

Special focus is given to IGBT turn-on and diode recovery where the ringing highlighted in Figure 1 was evident.


Equivalent circuit diagrams
Figure 2: Equivalent circuit diagrams


By considering the nHPD2 phase leg as a combination of LCR models similar to Figure 2., which differ according to the switching state and diode type (i.e. Si or SiC), a differential equation may be established:


Assuming a silicon IGBT, the turn_on is significantly faster than the diode recovery period thus can be ignored and Figure 2c) applied.

To be free from oscillation, the result of equation (3) must be zero. For the SiC SBD, acts much faster than and thus Figure 2d) can be considered.

It is now possible to simply identify the importance of the three key elements. Having identified that increasing will eliminate the beneficial WBG zero recovery characteristic, the IGBT’s is known to have a substantially higher capacitance than the diode due to its thinner charge carrier zone, then we can consider the beneficial influence of the system inductance and attribute a realistic package inductance to suppress the oscillation seen in Figure 1. Since it is a combination of the module package, DC link capacitor, and busbar, 25% of the total allowable stray inductance was defined as the target value for the module package.


Defining the allowable package inductance

Testing Hitachi’s first-generation* 3.3kV SiC MOS chip (rated 25A) and 3.3kV SBD (25A) in combination with different Rg_on and stray inductance values (package), a trade-off map was plotted to determine suitable WBG performance points. Refer to Figure 3.


Equivalent circuit Stray inductance versus Rg(on) turn_on waveforms
Figure 3: Equivalent circuit Stray inductance versus Rg(on) turn_on waveforms


Using the plotted data it was determined necessary to set the 3.3kV LV nHPD2 package inductance to 10nH, allowing for a 3.3kV system level of 40nH.


Optimizing package structure

A typical 3.3kV 1500A 190mm x 140mm IHM package the module stray inductance is about 6-7nH. With market acceptance for the next generation module footprint set at 100mm x 140mm, the 10nH target value does not disadvantage system output power requirements enjoyed today allowing package introduction using silicon technology whilst offering platform to mount SiC without the ill-effects of electromagnetic interference.

By way of electromagnetic simulation, the internal and external package design is investigated to acquire an optimum P-N terminal gap clearance. Whilst a wider gap offers simpler bus-bar connectivity, it adversely affects performance.

According to the existing IEC convention, strict rules apply to creepage and clearance. With consideration of functional isolation and a 4kV Voltage impulse (3.3kV IGBT), the minimum functional terminal-terminal distance shall be 7.5mm. A higher gap will linearly increase the module inductance and negate the WBG performance benefits. A higher gap will also increase design pressure on capacitor and busbar manufacturers to meet the system level goal, identified earlier as 40nH total stray inductance.



Using experimental tests, two type names were assessed using the same Hitachi advanced Trench gate HiGT structures, both adopting SiC SBD. Switching curves are compared in Figure 4. under three switching states.


Validation of oscillation suppression using low inductance package (nHPDD2)
Figure 4: Validation of oscillation suppression using low inductance package (nHPDD2)


Figure 4 confirms the genuine merit of the low inductance package solution for high power WBG applications. Turn_on and reverse recovery ringing is suppressed whilst enjoying WBG technology, as highlighted by the MBM450FS33F-C example. (Additional loss benefits exist but are not explored in this article. See published papers: PCIM2014 KAWASE APEC2016 SAITO).


LV nHPD2 (available) & HV (under development)
Figure 5: LV nHPD2 (available) & HV (under development)



By adopting nHPD2 packaging technology Hitachi has demonstrated its pioneering spirit to realize the wider market need for operational high power WBG devices. Benefiting from a 9nH module inductance (using the 3.3kV example), nHPD2 is able to realize potential reductions in turn_on losses up to 50% and almost 100% of the reverse recovery losses. For cost-sensitive applications requiring a silicon only solution, nHPD2 still delivers performance and lifetime advantages compared to conventional packages, in addition to several secondary cost benefits of adopting a modular design approach using a single standardized footprint. Referring to the opening title “Can changing your package really be that simple?”. Yes. The caveat? The best solution is a system solution. With the combined efforts of capacitor and bus-bar manufacturers, the system engineers, and the semiconductor industry, each striving to reach key stray inductance targets, yes it really can be that simple.


About the Author

Neil Markham works as the Product Marketing Manager of Power Devices Division at Hitachi Europe Lt., a European branch of the highly diversified Japanese company, Hitachi Ltd., that operates eleven business segments: Information & Telecommunication Systems, Social Infrastructure, High Functional Materials & Components, Financial Services, Power Systems, Electronic Systems & Equipment, Automotive Systems, Railway & Urban Systems, Digital Media & Consumer Products, Construction Machinery, and Other Components & Systems.


This article originally appeared in the Bodo’s Power Systems magazine.