Improvements in MOSFET parameters like gatedrain charge, Qgd, will not fully translate into improvements in DCDC converter performance as long as we continue to use inadequate packages and PCB layout techniquesSome modern packages like BGA and DirectFet have addressed this point and possess very little source inductances We will demonstrate that the results obtained in this paper are supported by simulation and lab test results.
Detailed Mathematical ApproachFigure 1b depicts a typical DPAK MOSFET package used in the power conversion field. This class of packages which also includes D2PAK, IPAK and TO220, has been in use for a long time and has been the work horse of the industry. Modern electronics and in particular personal computers (PCs), notebook computers and the myriad of handheld electronic devices have always required a continuous increase in power, features and functionality from the DCDC converters used while allowing for an ever diminishing PCB real estate. This has pushed the switching frequencies to the levels they are at today  i.e. from 200 KHz to 2 MHz  to facilitate the physical shrinkage of the PCB space taken by them. As we will demonstrate below, this increase in switching frequency can cause problems that lead to excessive power loss, lower efficiency and higher temperature operation leading to lower reliability.
1. MOSFET Turn OffFigure 1a depicts the control MOSFET in a synchronous buck converter equivalent circuit as it turns off. In order to write the equations that govern the drain current turn off we will use the representations shown in Figure 2 to help explain what happens inside the MOSFET. We have represented the gate threshold voltage as a simple battery of value Vgth typically between 0.8 and 3 V.


Figure 1a. Simplified Circuit Schematic depicting
the MOSFET t turn off, tf.

Figure 1b. Power package showing the source lead and bonding wires that for the source inductance, Ls.





Figure 2. GateSource Loop Parameters.


The relationship between the transconductance, gm, the gatesource voltage, Vgs, and gate threshold voltage, Vgth, may be written as where a is a basic MOSFET parameter which may be easily derived from Figures 3 and 4 below.


Figure 3. Gatesource transfer characteristics.
By taking the slope at different gatesource voltages,
we get the graph in Figure 4.

Figure 4. Forward transconductance as a function of gatesource voltage. The slope of this line is the parameter a.

With the aid of the circuit in Figure 1, one may write the equations:

(1) 

(2) 

(3) 

(4) 

(5) 

(6) 

(7) 

(8) 

(9) 
where Idt is the drain current as a function of time t, tf is the drain current fall time as a function of a drain current IL and the circuit parameters, and Vst is the MOSFET source voltage as a function of time t.
Using the above equations in Idt and Vst we can now explore the different influences of the MOSFET and circuit parameter on the fall time tf.


Figure 5. The voltage across the source inductor
as a function of time t.

Figure 6. The drain (or inductor) current as a function of time t.



Figure 7. Drain current and gatesource voltage
for initial current = 7 A

Figure 8. Drain current and gatesource voltage
for initial current = 40 A

The above two plots show very clearly the very strong dependency of the fall time tf on the initial current. A larger initial current results in a larger fall time tf. This observation is confirmed by the lab test results in Figures 9 and 10 below.



Figure 9. Captured scope image showing the control and synchronous rectifier currents, gate voltage, source voltage and gatesource voltage for low initial current ≈ 7 A in a synchronous buck converter

Figure 10. Captured scope image showing the control and synchronous rectifier currents, gate voltage, source voltage and gatesource voltage for high initial current ≈ 40 A in a synchronous buck converter

Figures 9 and 10 depict the results of actual lab tests done on a synchronous buck converter where currents for both the high side and low side MOSFETs are shown together with the gatesource voltage of the high side MOSFET and the switching node. Figure 11 depicts the simulation results using a very powerful device simulation software for two source inductance, Ls, values of 1nH and 10nH which clearly agree with both the mathematical formulae derived here and the lab test results.

Figure 11. This graph depicts simulation results using device simulation software.
This also agrees with the lab test and mathematical solution.
Figures 12 through 14 show the fall time, tf, as a function of different parameters using equation 10. There is a very strong nonlinear dependency of tf on the initial current, IL, the source inductance, Ls, and gate threshold voltage Vgth. The value of tf is independent of the converter's switching frequency making it a more substantial percentage of the duty cycle as the switching frequency goes higher resulting in higher power dissipation and lower efficiency.



Figure 12. The current fall time, tf, as a function of initial current and gate threshold voltage. Notice the clear nonlinear dependency.

Figure 13. The current fall time, tf as a function of the source inductance and initial current. Again, clear nonlinear dependency.


Figure 14. The current fall time, tf, as a function of gate threshold voltage and source inductance.
Similar plots and/or calculations may help the designer in the choice of a MOSFET's silicon and package

For completeness, it is worthwhile noting that the following approximations have been done to be able to obtain an exact symbolic solution that can help us understand the dependencies of the current fall time, tf, on the different parameters where the following parameters have been ignored: 1. The effect of the equivalent source resistance, ESR, of the inductor Ls since it is in the range of one or two milliohms 2. The effect of gatesource capacitance, Cgs, and the gatedrain capacitance, Cds 3. The effect of the driver output impedance 4. The effect of the intrinsic gate resistance 5. The effect of the gate trace inductance
2. MOSFET Turn On


Figure 15. Simplified schematics used for the current turn on time calculation.

Figure 16. Lab test graph showing MOSFET turn on followed by inductor current ramp.

For the derivation of a closed form equation for the drain current turn on time we will use the simplified schematic in Figure 15. In this analysis we started with the full gate voltage applied to the drain assuming the rise time ≈ 0. This assumption is a very reasonable one as can be seen in Figure 16. This also allows the derivation of a closed form formulae for all currents and voltages. Figure 16 depicts captured scope images showing the gate drive, drain current and drainsource voltages.

(10) 

(11) 

(12) 


Figure 16. The drain current as a function of time t.

Figure 17. Current as a function of the source inductance Ls and time t.



Figure 18. The drain current as a function of the gate threshold voltage and time

Figure 19. The drain current as a function of the parameter and time

Derivation of on time, ton
The on time, ton, can now be calculated by substituting IL in the drain current equation. This value represents the inductor or drain current in the synchronous buck converter at the time of turn on. When the equation was solved, we got two solutions ton1 and ton2. Only ton2 is the equation that represents the right solution. Figure 20 and equation 15 below were used to verify this conclusion.
Figure 20. The drain current as a function of time .

(15) 


Figure 21. On time, ton2, as a function of inductor current IL and source inductance Ls.

Figure 22. On time as a function of the source inductance Ls and gate threshold voltage Vgth.

Conclusion1. The parasitic source inductance controls the drain current rise and fall times. We have shown that a smaller inductance is better. In modern MOSFETs there is a minimum inductance dictated by the device package e.g. ≈ 1 nH for SO8 and more for DPAK and D2PAK. Though modern packages offer superior performance, improved PCB layout techniques are a must. 2. The nonlinear dependency of IL on gatesource voltage  gate threshold voltage plays a major role in switching times. Lower gm around the gate threshold will result in faster fall times.3. Gate threshold voltage plays a role in the rise and fall times. Higher Vgth will result in faster fall time but will result in slower rise time.4. The use of a bipolar gate drive voltage that swings both positive to turn on and negative for turn off will result in faster tf.
5. Given all of the above, improvements in gatedrain charge, Qgd, alone will not result in any significant improvements in switching times if the source inductance is large enough to adversely affect the rise and fall times. In order to take full advantage of the improvements in Qgd we must have:  PCB layout with minimum trace inductance  MOSFET package with minimum source inductance 6. The rise time can be made faster with the proper gate drive voltage. Larger gate drive voltage will result in faster on times but will increase the gate drive losses where fs is the switching frequency and Vgd is the gate drive voltage.7. The above conclusions should place designers on the right track to design DCDC converters that can perform at the highest efficiency possible for a given specification.
References
[1] Evaluating MOSFET Susceptibility for CrossConduction, Alan Elbanhawy, PCIM Magazine, Europe[2] Shoot through analysis with parasitic inductance, Alan Elbanhawy, PCIM Europe Conference 2004[3] Effect of Gate ESR on MOSFET Switching Losses, Alan Elbanhawy, PCIM Europe Conference 2005 [4] A. Elbanhawy, "Effect of Parasitic Inductance on switching performance" in Proc. PCIM Europe 2003, pp.251255 [5] A. Elbanhawy, "Effect of Parasitic inductance on switching performance of Synchronous Buck Converter" in Proc. Intel Technology Symposium 2003 [6] A. Elbanhawy, ?Mathematical Treatment for HS MOSFET Turn Off" in Proc. PEDS 2003[7] A. Elbanhawy, "A quantum Leap in Semiconductor packaging" in Proc. PCIM China, pp. 6064
[8] A. Elbanhawy, ?The Road to 200 Amps at one Volt VRM? in Proc. PCIM Europe 2004, pp. 5458[9] A. Elbanhawy and W. Newberry, ?Packaging Parasitic Resistance Frequency Effects? Power Electronics Conference USA 2004 PCIM San Francisco?[10] A. Elbanhawy and J. Ejury, ?Investigations of the Influence of PCB Layout Parasitic Inductances In DC/DCConverters on the Efficiency? in Proc. PCIM Europe 2004, pp. 3136[11] A. Elbanhawy ?Are traditional packages suitable for the new generation of DCDC converter? in Proc. IPEMC China 2004[12] Effect of gate ESR on MOSFET Switching Losses, Alan Elbanhawy, PCIM Europe Conference 2005[13] Effect of Gate ESR on Localized Shoot Through on the MOSFET Die, Alan Elbanhawy, IPEC 2005 Conference, Niigata Japan
Legal Notice: The copyright for this application is owned by the author(s). Neither Maplesoft nor the author are responsible for any errors contained within and are not liable for any damages resulting from the use of this material. This application is intended for noncommercial, nonprofit use only. Contact the author for permission if you wish to use this application in forprofit activities.