What to look for when specifying an IPM driver
Besides obvious factors such as price and package, there are two functional value propositions to take into account: switching speed and noise immunity. In terms of switching speed, propagation delay is the key—but not the only—parameter. Switching speed requirements of the IPM driver are governed by the switching speed required for the IPM module in question, or the pulse-width modulation (PWM) dynamic range. For example, with a 20kHz switching period (typical for IGBT-based IPMs) and a design requiring modulation of the switching pulse from 10 per cent to 99 per cent, the minimum pulse width required will be approximately 5 µs. Correspondingly, if the minimum pulse width required is 1 per cent, a minimum pulse of 0.5 µs would be needed. The pertinent calculation is:
t = (1/Switching frequency) × Minimum pulse width required
A good rule of thumb is that the maximum propagation delay should be less than the minimum pulse width required.
The next switching issue of concern is dead time. Most switching applications arrange their output switches in a ‘push-pull’ configuration. In this type of switch arrangement, it is extremely important to assure that the low-side switch is turned off before the high-side switch is turned on, and vice versa. If both switches are turned on at the same time, a short is created across the high-voltage power supply. At best, this results in tremendous stress on the output power components, if confined to a very short time. At worst, the result is smoke and fire.
In order to minimise this momentary high-current flow, commonly known as ‘shoot-through,’ the turning on of one switch is delayed from the turn-off of the complementary switch; the delay between these two switching events is known as dead time.
When both the switches are off, nothing happens. As a result, increasing dead time decreases the efficiency of motor drives and solar inverters. Therefore it is desirable to have as much dead time as needed to prevent shoot-through, but no more.
The optimal case for dead time would be a value of zero, where one switch closes instantaneously and the other switches on instantaneously. This type of behaviour is possible to achieve with simulations, but not with real components.
In terms of the IPM driver, ignoring any delays in the switches themselves, a zero delay would be required when the high-to-low propagation delay (tphl) is exactly equal to the low-to-high propagation delay (tplh). Again, this is not a realistic condition as there is some variance between tphl and tplh.