Gate drive output is ringing..

[si2030]

Hi there,

I have a simple circuit using a TL494 on a breadboard which is outputting alternative square wave pulses. This is then fed into two gate drive ICs. The two ICs here are IDXN430's.

See attachment below for this IC's datasheet.

The TL494 is running on 15 volts... The gate drivers have 20 volts however I did try it on the same rail of 15 volts and got the same result. There is no ringing into these driver ICs. There is ringing at the start of each turn on and off for the square wave...

View attachment 62151

Wondering how to fix this..

Kind Regards

Simon

 

[ronsimpson]

Attached are some IC that do gate drive like you want. That way the transformer can go away.

Some of this type of parts have a circuit that will not let both MOSFETs to be on.

You are using LTspice so look up Linear's parts. I know I have used a MOSFET Gate driver with LTspice. I just can't remember what I used. I will look it up in the morning.

On the transformer; use a resistor across the gates to dampen the ring. (as I said) or use a RC.

 

[crutschow]

Where's the transformer?

 

[si2030]

I was using a toroidal gate transformer which I wound myself.. it seemed to work perfectly - just it was reflecting the ringing coming from these gate drives.



Incidentally might I also ask... with gate drivers... whats the significance of the source and sink peak current some say 2.3amps some are in the milli amps - 300mA some are much higher. I am using IRFP450 mosfets at the moment and at a later date IGBTs - irgp50b60pd1.

 

[crutschow]

.............................

Incidentally might I also ask... with gate drivers... whats the significance of the source and sink peak current some say 2.3amps some are in the milli amps - 300mA some are much higher. I am using IRFP450 mosfets at the moment and at a later date IGBTs - irgp50b60pd1.

MOSFETs have a large gate charge capacitance that must be rapidly charged and discharged to switch the device on and off. How much current you need depends upon the value of this charge and how fast you want the transistor to switch.

The data sheet shows that the IRFP450 has 150nC of Total Gate Charge that must be switched. Thus if you drove that gate with the driver capable of 2.3A peak current it would take 150nC / 2.3A = 65.2ns to switch the gate.

 

[si2030]

So if I understand this correctly you would only use the drivers with the larger source and sink currents where you have very fast switching requirements.. You mentioned 65ns... if my operational frequency is 45kHz then this represents ~1/170th of the switching cycle...

What is an acceptable rise time in relation to the frequency being used?

At what point would you consider using a driver with a bigger source sink current?

 

[crutschow]

So if I understand this correctly you would only use the drivers with the larger source and sink currents where you have very fast switching requirements.. You mentioned 65ns... if my operational frequency is 45kHz then this represents ~1/170th of the switching cycle...

What is an acceptable rise time in relation to the frequency being used?

At what point would you consider using a driver with a bigger source sink current?

It's generally a function of the allowable power dissipation in the transistor and desired efficiency of the circuit. A switch dissipates power during the switching period of roughly 1/2 times the ON current times the OFF voltage times the switching duration. Thus the faster the switching frequency, the more dissipation from switching. I would expect that keeping the rise and fall times below 1% of the switching period would generally be acceptable for most applications, but the faster the better.

 

[si2030]

Thanks for that.. I now have a better idea of the requirements etc for gates..

 

[creakndale]

First, is the ringing real? Measuring fast transition signals requires an exceptionally good scope probe ground lead. Use the shortest ground lead possible. Make sure the scope probe ground wire is making perfect contact with the clip (soldered to it) and the connection to the probe body is clean. You can jump through all sorts of hoops trying to reduce ringing only to eventually realize it's caused by the test equipment.

Try grounding the probe to a different ground point to see if it makes any difference.

They make electrical ground leads that connect right next to the probe tip so to minimize scope probe induced ringing, under and/or overshoot. Here's an example. To use something like this, your test point needs a ground pad right next to the signal you're measuring.
https://www.testpath.com/Items/Ground-Lead-117-256.htm

creakndale

 

[MrAl]

Hi Simon,


When we connect an inductor in series with a capacitor we create a potential oscillator. When we drive it with a pulse the inductor and capacitor begin to oscillate. If it is an ideal inductor and ideal capacitor with no losses that oscillation would continue indefinitely even if we remove the pulse source.

This oscillation contains sine and cosine components as well as an exponential part. The sine and cosine are the oscillations we see. If we include a series resistance that provides some damping so if we remove the pulse source eventually the oscillation stops. The resistance controls the exponentially decreasing part so how much it still oscillates (if at all) and how fast it stops oscillating depends on how much resistance we add in series. For a small resistance it takes a longer time to stop the oscillations, for a larger resistance we may not see any oscillation or it will stop sooner.

In the MOSFET drive circuit we have inductance in the two leads that go to the gate and source, and a MOSFET input capacitance. This inductance and capacitance are in series and so can oscillate just like an oscillator. There is also a series resistance such as the lead resistance, but it is often very very low so we might see oscillations at the gate.

To minimize this oscillation the leads between driver and gate and source have to be very short. Also, if we use a drive transformer that transformer had best be wound bifilar.

Those solutions alone dont always reduce the ringing. To aid in damping this oscillation one solution is to provide a series resistance. This resistance is placed in series with the gate. It is often a fairly low value on the order of 5 to 25 ohms. This reduces the ringing because now there is plenty of damping, but it also increases the rise and fall times to some degree so there's a slight disadvantage. This increase however shows up as an advantage on the output of the MOSFET, where the slower rise and fall times generate less overshoot on the output.
You can try low values and see how much it affects the ringing and output overshoot.

If the leads are longer than a couple inches because they have to be that way, then a twisted pair cable should be used, and the same length going cable to all the MOSFETs is a good idea.