# How do you size a resistor when the current and voltage flowing through it are impulsive?

Given a 33 Ω resistor, I need to size the power it can handle. The resistor is in series between a pin where a SPWM voltage is applied and the gate of an IGBT.

In the following photo: blue is the voltage at its ends, red is the SPWM signal input to the gate controller IC IR2110 (before being adjusted to the required electrical amplitudes), green is the current flowing in the resistor

The switching frequency is 10 kHz, while the sinusoidal reference is 50 Hz. I read another answer present here on stackexchange, where a shunt resistor was sized based on the duty cycle of a simple PWM signal, however here the resistor is not current flowing in the dutycycle but only in the rising and falling edge of the LO_1 control signal. I was thinking that perhaps, a good approach might be to size the resistor in the worst case, i.e., at the peak of the sine wave where the SPWM signal is HIGH for almost the entire period. But again, the current and voltage are impulsive, for information, the reason two opposite pulses follow each other is to turn off the igbt, as the absence of voltage is not enough to turn it off.

I don't know if given the implusive nature, the classical model still applies whereby the resistor must be sized with a power of 4W, or in these cases you do it another way.

Below the screen of the design of a half bridge branch of an inverter and the 33 Ω resistor in question.

How can this resistor be sized? What would be the best resistor technology?

• Hold down Alt key and click on the resistor. Zoom out to several switch cycles. Hold down Ctrl and click on the trace name. What’s the average power? Commented Sep 18, 2023 at 21:19
• the average power is 11.63mW, so do I just need a standard resistor like the ones used for Arduino 1/4 W applications, or do I need a resistor of a specific technology? Commented Sep 18, 2023 at 21:25
• First time I hear the word impulsive in electronic context. Perhaps something for a stop-motion animation of components. Commented Sep 19, 2023 at 4:06

You need to ensure the average power dissipation is well within the capabilities of the resistor.

It would be best to use a resistor that has a pulse-power curve in the datasheet. Obviously you also want it to have low inductance. So probably a thick-film resistor would be best.

Here, for example, are Yageo's datasheet curves:

Plot the power vs. time in your simulation (and you can also find the average). For the former, alt-left-click on the resistor and power-vs-time will be plotted.

As you can see from the graph, 4W peak is not great for the smallest resistors. Also, note that Yageo has a disclaimer, so it would be prudent to allow some healthy safety margin.

• considering an average power dissipation of 11.63mW, is the thick-film resistor shown at the following link a good choice? it.rs-online.com/web/p/resistenze-smd/9056342 Commented Sep 19, 2023 at 14:39
• Looks quite conservative (good) to me. Commented Sep 19, 2023 at 14:48

In this example the load is the gate capacitance.

Energy dissipated by a resistor charging a cap is 0.5*cv^2.

So plug your gate drive supply voltage into V and the gate capacitance into C. Then multiply by your switching frequency and multiply by 2 again because you have to charge and discharge each switching period.

So freq*cv^2 is the power dissipation in watts. Pick a resistor rated for this.

Note there are some nuances to gate C (it’s nonlinear) that I’m not covering here. Consider diving deeper into these if your resistor rating is close.

• Note this applies to the total resistance, including the driver's internal resistance (or equivalent thereof, in case it has some constant-current range). This power is shared between the two components (driver and resistor). Commented Sep 18, 2023 at 22:09
• True and worth mentioning though driver impedance is typically low and varies pretty widely. IMO you want a resistor that can take the entire load (and in my experience the math doesn’t force you into a large part anyway). Commented Sep 18, 2023 at 23:37
• "Typically low" -- but check IR2110 in particular! Commented Sep 19, 2023 at 0:39