Power Supply Transient Response for IXDI614 Gate driver

Thread Starter

Joester0064

Joined Jun 14, 2023
26
Hi All,

This gate driver IXDI614 can source/sink 14A. Since the switching time is around 20ns the power supply for this must be up to the job. Say I use a simple 12V regulator rated at 3A for this. Is this enough? Will the regulator be able to source this much current to make the 14A pulse since it is not a constant current? It will be running around 100Hz with at max a 50% duty cycle sp maybe the 3A rating is ok. Not too sure I just want to be sure there is 14A available for each pulse. Maybe just a large cap on regulator output would do it? Any ideas are much appreciated.

Thanks!

Joe
 

LowQCab

Joined Nov 6, 2012
4,052
Look at the Schematic in the Spec-Sheet.
Note that there is a 100nF Ceramic-Capacitor, and a 10uF Electrolytic-Capacitor.
These Capacitors provide almost all of the Current for Switching-Transients.

The Electrolytic-Cap is relatively "slow" and has
higher "Series-Resistance", and higher "Series-Inductance", when compared to the Ceramic-Cap.
The Ceramic-Cap is very "fast", but is rather expensive compared to the Electrolytic-Cap.

The 2 Caps together eliminate the slow response of the Power-Supply, and the
unavoidable Inductance created between the Power-Supply-Output and the Switching-Transistors.
.
.
.
 

MisterBill2

Joined Jan 23, 2018
18,433
L-QC is correct, and, in addition, the actual response time of the voltage regulator enters into the concern. "Fast" is much slower than "Instant", and the capacitor values given are minimums, not the recommended for all applications, values.
 

LowQCab

Joined Nov 6, 2012
4,052
How You choose the Capacitors usually depends upon ~4 main things ..........
How fast do You need the Switching to be for the required efficiency,
How much Gate-Capacitance do You need to Switch,
How much Board -Space do You have available, and,
How much Money do You have to spend on your Project.

"Good enough", or "I have about 200 of these Capacitors", is usually the solution.
.
.
.
 

MisterBill2

Joined Jan 23, 2018
18,433
Linear regulators versus switching regulators. My comment assumes a " a simple 12V regulator rated at 3A ".A switch mode regulator adds more considerations, as has been noted.
 

LowQCab

Joined Nov 6, 2012
4,052
You need a Circuit-Simulator to estimate ........
First, the ideal Switching-Gate-Current that insures minimum "Ringing" of the Circuit, and,
Second, the calculated Gate-Resistor required to achieve that required stability,
and only then, can You calculate what the Switching-Current-Spike will be.

It will, hopefully, be far less than a peak of ~14-Amps.
.
.
.
 

Thread Starter

Joester0064

Joined Jun 14, 2023
26
why do you say hopefully it will be less than 14A, I thought high current spike, faster gate capacitance charge and faster switching? Also, how does the ringing actually affect the circuit other than just showing up on the scope?
 

crutschow

Joined Mar 14, 2008
34,392
If the IXDI614 load is the gate capacitance of a MOSFET, than 100µF in parallel with 100nf across the IXDI614 power pin to ground should be more than enough to supply the peak current required.
 

ronsimpson

Joined Oct 7, 2019
3,037
The average current in the Gate Driver is very low.
You should have a resistor from Driver to Gate. That will limit the current and stop/control ringing and oscillation.
Driving the Gate too hard will not reduce the losses. It will probably increase the losses in the big transistor.
 

LowQCab

Joined Nov 6, 2012
4,052
If You try to cram a Square-Wave, into a Capacitance, or into an Inductance,
You will always create some sort of Oscillation, or "Ringing".

This may, or may not, cause a problem, depending upon all of the surrounding Components.
.
.
.
 

ronsimpson

Joined Oct 7, 2019
3,037
Why is that?
There is normally a Gate Resistor on the MOSFET. It is there to (more or less) slow down the MOSFET.
In a non-perfect world, there are stray inductance in every lead of every part, and in every piece of PCB copper. There are also capacitors between parts and between every trace and the next trace or part. These Ls and Cs will ring, and maybe cause the MOSFET to oscillate.
Look at the current rise and fall time with a fast scope and fast current probe. I have seen the current go from 100A to 0A then ring back to 30A then 0A the 20A, 0A 10A 0A. This ring heats up the part. Even if the 30A at 800V is there for only 50nS, it is heat. During the ringing the diodes may also having a hard time.
I understand the logic, that the faster you switch the less time you spend in the linear area where power is lost. If you ever ring, then you may be in the linear region more times per cycle than you think.
Look at the data sheet. They will tell you what Gate Drive resistor they used. That is the smallest resistor you should use.
Fast switching MOSFETs and diodes cause RF interference. Another reason to slow down the MOSFET.
 

MisterBill2

Joined Jan 23, 2018
18,433
R.S. is correct about every part, including all the conductors, having an effect in fast switching regulators. In that aspect they are about as complex as RF power amplifiers.
THAT is why I buy the switcher supplies from manufacturers that I know and respect, because it is cheaper than de-bugging a swtcher. Also, it avoids production issues.
That may not apply for production runs of thousands, but it certainly applies for a build of a dozen machines.
 

Thread Starter

Joester0064

Joined Jun 14, 2023
26
Thanks all for such great insight.

Here is another related question:

I understand now that the capacitance on the supply rail for the totem pole side of the gate driver facilities the current pulses to the MOSFET gate.

What would be the lowest acceptable current rating for a voltage regulator that needs to supply 15V to the drive side of an IXDI614 switching a SiC MOSFET with a frequency of 100Hz and 10% duty cycle?

Thanks again folks!
 

MisterBill2

Joined Jan 23, 2018
18,433
Given that this is an important function, my suggestion is for twice the gate drive current requirement. hat would allow for the capacitor losing capacitance over time. And probably not add to the cost of having the supply at all.
 
Top