Gate drivers ICS Dying( UCC37322 and UCC37321 )

[prabhatkumar]

So is that making my waveform slightly unequal on each side ??
http://electronsbefree.blogspot.com/p/electronics.html
Here it's written why we need the resistors

[davekni]

The 1N5819 diodes are a good idea.  Most gate drivers aren't used with inductive components on their outputs.  All the rest is likely compensating for poor circuit board layout and/or poor component performance (high ESR capacitors in particular).  That exact fix happened to work for some particular layout.  The ground resistors in particular will cause problems in most uses, and may be the cause of your specific issue.

How are your 3732x chips actually wired?  The reference you pointed to shows driver chips with single ground pins.  Did you wire the 3732x grounds together, then through 1-ohm to ground, or is it only the output side with 1-ohm resistors and the input side directly grounded?

[prabhatkumar]

I am attaching a reference schematic from mads own website which I followed and made a pcb of it . And by the way I have used uf4007 instead of the in5819 as I heard they are better here .
The grounds are wired together and then through 1 ohm resistor they are grounded.
EDIT: I am attaching some waveforms of the gates of the two mosfets and the current through the GDT. I have also increased the resistance of the series resistor though the GDT for measuring the current through it and also restrict the current through to a certain value.
Now my problem is that as it can be seen in the wave forms that the voltage at gates of one mosfet (YELLOW COLOUR) rises to a higher value then it goes below it ( The difference is around 0.5 V). So is that much of variation allowed or something is wrong ??
EDIT 2: I have now tested it with a voltage of 5 volt which is separate unit ( Its a low power wall wart), from the bench power supply i am powering slayer exciter with. The input to the bridge is hence 5 volt with a series 22 ohm resistor to prevent any over current if it happens. Now the drivers DIDN'T BLOW. But then again gate waveforms didn't look very promising as one side is rising higher than the other . So I am not yet sure to increase the input voltage to the bridge. Also I notice ringing is dominating now .Please have look and tell what is wrong and how could it be made better.New wave forms after powering the bridge

[davekni]

How are you measuring GDT current?  You mentioned increasing a resistor in series with the GDT.  What value is that resistor?  Are you measuring current by using a differential scope probe on the resistor?  I see 2.27V peak, but don't know how to translate that to amps.

Looking back to the beginning of your post, I see that this is driving a pair of TO247 parts, not a larger IGBT brick.  The gate capacitance and therefore current may be low enough that the 1ohm ground resistors aren't causing issues.

The uf4007 diodes are plenty fast, but aren't schottky, so have higher forward voltage drop.  They aren't as useful for protecting the gate driver outputs as 1N5819 or similar schottky parts are.  However, they likely do contribute some to protection.

I wouldn't be concerned about the 0.5V gate-drive voltage asymmetry.  That could easily be due to the duty cycle not being exactly 50% and/or the non-linear gate capacitance of the TO247 IGBTs (or FETs - I didn't search back to see which you are using).

The ringing is presumably a result of the IGBT switching.  Hard to say if ringing will scale up enough at higher IGBT voltage/current to become a problem.  The key will be good low-inductance bypass and bulk capacitance across the IGBT half-bridge power rails.

[prabhatkumar]

Thanks for the reply davekni. I am using an oscilloscope whose ground terminal I insulated so that I I can take non ground reference measurements without a differential probe. I am using a 2.7 ohm resistor in series to the gdt. Also I have 5.6 ohm resistor on the gates of the mosfets and regarding the voltage asymmetry , it almost fixed itself I don't know why that happened at the first place but it's still there a little bit but then as you said I don't need to worried about this

[davekni]

The gate-drive signal may be low enough impedance to work with a floating scope.  In general, a scope's floating ground still has quite a bit of capacitance to the power line, primary in the power supply's EMI filtering, perhaps 10nF.  I'd recommend against connecting the "floating" ground to any signals with significant source impedance.  The scope ground impedance may be causing the asymmetry, or at least contributing to it.

2.27V on 2.7 ohms is under 1A peak, so under 1V peak across the 1-ohm ground resistors.  That's low enough to avoid unintended logic switching.  I still recommend against any ground resistor.  If the chip needs current limiting or spike filtering, the supply resistor should be sufficient.  Since the two chips switch opposite directions, there's a supply resistor in series with the gate drive signal for every transition.  Double the supply resistors to 2 ohms each if you want the same total resistance as 1 ohm ground and 1 ohm supply.

[prabhatkumar]

The gate-drive signal may be low enough impedance to work with a floating scope.  In general, a scope's floating ground still has quite a bit of capacitance to the power line, primary in the power supply's EMI filtering, perhaps 10nF.  I'd recommend against connecting the "floating" ground to any signals with significant source impedance.  The scope ground impedance may be causing the asymmetry, or at least contributing to it.

2.27V on 2.7 ohms is under 1A peak, so under 1V peak across the 1-ohm ground resistors.  That's low enough to avoid unintended logic switching.  I still recommend against any ground resistor.  If the chip needs current limiting or spike filtering, the supply resistor should be sufficient.  Since the two chips switch opposite directions, there's a supply resistor in series with the gate drive signal for every transition.  Double the supply resistors to 2 ohms each if you want the same total resistance as 1 ohm ground and 1 ohm supply.

Hey devkni thanks for the quick reply . But I am a newbie yet , so there a quite a lot of things I didn't understand in your statement. What do you by that the voltage is low to avoid any unwanted switching.
Second : and there isn't any supply resistor in series to chip . So it's it required ??
And what do you mean by 1 ohm ground and 1 ohm supply in the last line .
And what exactly happens with a high impedance load and a floating scope connected across it ?

[davekni]

From your post yesterday you mentioned that you were following this recommendation:
http://electronsbefree.blogspot.com/p/electronics.html
And from your next post yesterday:
"The grounds are wired together and then through 1 ohm resistor they are grounded."

Did you add only the ground resistors from the electronsbefree schematic?  I see no need for either ground or supply resistors.  However, if you are going to include resistors as per electronsbefree, I recommend adding only the supply resistors, not the ground resistors.  This is in response to your question "And what do you mean by 1 ohm ground and 1 ohm supply in the last line".  Those 1 ohm resistors are the ones in the electronsbefree schematic, which I thought you had copied.

The resistors in series with the driver chip grounds are a problem only if the voltage across them is high enough to make an incoming logic 1 level (+5V) look like a logic 0.  In other words, if the gate-drive chip ground spikes up to 3 to 4V, then a 5V logic input will be only 1 to 2V above it's 3 to 4V ground, so will register as a logic 0.  For 1-ohm ground resistors, it would take 3 to 4 amps to reach that voltage.  Your gate-drive currents appear to be just under 1A (2.27V / 2.7 ohms).  That will create spikes on the gate-drive grounds of just under 1V, not enough to cause trouble.  You could scope the ground pins (voltage across the 1-ohm ground resistors) to verify that voltage spikes are low enough.

Concerning your question: "And what exactly happens with a high impedance load and a floating scope connected across it?", the issue is current flowing through the scope probe "ground" lead.  Even with the scope "ground" isolated from the power line safety ground, the scope "ground" will have capacitance to the power line hot and neutral leads.  This capacitance will inject some line-frequency (50 or 60Hz) current into the circuit at the node connected to the scope "ground".  The capacitance will also load down higher-frequency signals, as current from the signal is fed through the scope ground, through the scope's capacitance, then through the hot and/or neutral power leads to ground at the line breaker panel.

How much problem the scope "ground" current is depends on circuit impedance and scope "ground" impedance.  Larger scopes that draw more line power are likely to have more capacitance from hot and neutral to "ground".  You could experiment by scoping a floating signal as you've been doing, but add a resistor between the scope "ground" and the signal, say 10k ohms.  Compare the traces with 10k in series with the "ground" and separately with the 10k in series with the probe tip instead.  If the two scope traces look roughly the same, then your scope "ground" load is less than I'm guessing, so less of a problem.

[prabhatkumar]

Thanks a lot davekni for the reply and explaining so well. By the way now I am experiencing another problem that the bridge doesn't supply any current even though it appears to be switching . I have started a new topic ( NO CURRENT THROUGHB HALF BRIDGE )about that so it would helpful if you could have aook in to that also . Thanks !