Gate resistor value tradeoff
chris_inkubate, Mon Sept 15 2014, 06:03PMHi all,
I have been experimenting with gate driving a Littelfuse MG06400D-BN1MM half bridge brick using UD2.5 and don't fully understand the trade offs associated with the gate resistor value. This brick already has an integrated 2.5R gate resistor so I'm not sure if this is enough or if to add more. From tests adding an additional 4R7 caused the rise time of the gate waveform to be around 1us while testing at around 60KHz. Reducing this value to 2R near enough halved this to 500ns. I suspect that 1us is a bit slow but am also aware that dead time is created by this and is also desirable to avoid shoot through.
Can anyone clarify this for me please so I can determine and optimum value?
Many Thanks
IGBT:
Gate charge: 1.8uC
Integrated gate resistor: 2.5R
Re: Gate resistor value tradeoff
loneoceans, Mon Sept 15 2014, 06:13PM
I did a write-up on how to choose a good gate resistor value. Too small and you will be at risk of shoot-through. Too large and the gate doesn't turn on fast enough. The transistions can be seen by scoping the gate and collector and adjusting them for best performance. See here (somewhere in the middle with scope shots):
loneoceans, Mon Sept 15 2014, 06:13PM
I did a write-up on how to choose a good gate resistor value. Too small and you will be at risk of shoot-through. Too large and the gate doesn't turn on fast enough. The transistions can be seen by scoping the gate and collector and adjusting them for best performance. See here (somewhere in the middle with scope shots):
Re: Gate resistor value tradeoff
chris_inkubate, Mon Sept 15 2014, 07:18PM
Thanks for that link to your site - very useful information. Its made some sense to me so I can work towards perfecting the value when I also adjust my phase lead.
I know this may be a silly question but I assume from what I read that I scope the outputs of my full bridge where one connection goes to my primary and the other my tank cap? I've yet to test as my current scope probe isn't rated for this although having a USB scope and laptop should allow measurement without using two probes (I think!). I must admit its a scary prospect connecting my scope and laptop direct to my bridge!
Chris
chris_inkubate, Mon Sept 15 2014, 07:18PM
Thanks for that link to your site - very useful information. Its made some sense to me so I can work towards perfecting the value when I also adjust my phase lead.
I know this may be a silly question but I assume from what I read that I scope the outputs of my full bridge where one connection goes to my primary and the other my tank cap? I've yet to test as my current scope probe isn't rated for this although having a USB scope and laptop should allow measurement without using two probes (I think!). I must admit its a scary prospect connecting my scope and laptop direct to my bridge!
Chris
Re: Gate resistor value tradeoff
loneoceans, Mon Sept 15 2014, 09:15PM
You will have to be careful when scoping your bridge to avoid blowing things up. The best and safest way to do this is to isolate your entire bridge completely. I usually use a 60V isolated power supply for low power bridge testing, and then move on to an isolation transformer which then goes to my variac for higher power. Use one scope at the GATE of one IGBT and another at the Collector of the same igbt (for say the bottom IGBT so the ground goes to the emitter and the negative rail). This allows measurement of gate voltage and inverter output. For primary current, just measure the voltage across a burden resistor across a current transformer. You can make your own transformer just like making FB or OCD CT.
loneoceans, Mon Sept 15 2014, 09:15PM
You will have to be careful when scoping your bridge to avoid blowing things up. The best and safest way to do this is to isolate your entire bridge completely. I usually use a 60V isolated power supply for low power bridge testing, and then move on to an isolation transformer which then goes to my variac for higher power. Use one scope at the GATE of one IGBT and another at the Collector of the same igbt (for say the bottom IGBT so the ground goes to the emitter and the negative rail). This allows measurement of gate voltage and inverter output. For primary current, just measure the voltage across a burden resistor across a current transformer. You can make your own transformer just like making FB or OCD CT.
Re: Gate resistor value tradeoff
chris_inkubate, Mon Sept 15 2014, 10:10PM
Cheers for the info on testing. I'll add some waveform pix when I've done to show how I get on. I have a 48V SMPSU which I can use so I'll rig that up and check that its isolated first.
I made a little current transformer for testing that works well for current monitoring so I'll be able to finally measure and design this DRSSTC better then my previous attempt as I really don't want to blow a brick!
chris_inkubate, Mon Sept 15 2014, 10:10PM
Cheers for the info on testing. I'll add some waveform pix when I've done to show how I get on. I have a 48V SMPSU which I can use so I'll rig that up and check that its isolated first.
I made a little current transformer for testing that works well for current monitoring so I'll be able to finally measure and design this DRSSTC better then my previous attempt as I really don't want to blow a brick!
Re: Gate resistor value tradeoff
Dr. Dark Current, Mon Sept 15 2014, 10:33PM
Use the smallest resistor you can while the gate waveform still looks clean without too much ringing and overshoot. If the output resistance of your gate drivers is high enough, you can even get away with no resistor.
Dr. Dark Current, Mon Sept 15 2014, 10:33PM
Use the smallest resistor you can while the gate waveform still looks clean without too much ringing and overshoot. If the output resistance of your gate drivers is high enough, you can even get away with no resistor.
Re: Gate resistor value tradeoff
Steve Conner, Tue Sept 16 2014, 07:40AM
I believe too small a gate resistor can cause reliability issues due to excessive gate voltage overshoot and collector voltage spikes caused by turning the IGBT off faster than it was designed to go.
If you study the datasheet of an IGBT brick, you'll probably see a particular value of gate resistance that was used when measuring most of the parameters and charts. I take this as the recommended value.
To introduce deadtime, a larger gate resistor is commonly used along with a diode that shorts it out in the reverse direction, so the device turns off faster than it turns on. This is not ideal because the large gate resistor slows the turn-on rise time as well as the desired effect of lengthening the delay period. But it's simple and seems to work fine for most DRSSTCs.
Solid-state tesla coiling is all about connecting expensive things to dangerous things.
Steve Conner, Tue Sept 16 2014, 07:40AM
I believe too small a gate resistor can cause reliability issues due to excessive gate voltage overshoot and collector voltage spikes caused by turning the IGBT off faster than it was designed to go.
If you study the datasheet of an IGBT brick, you'll probably see a particular value of gate resistance that was used when measuring most of the parameters and charts. I take this as the recommended value.
To introduce deadtime, a larger gate resistor is commonly used along with a diode that shorts it out in the reverse direction, so the device turns off faster than it turns on. This is not ideal because the large gate resistor slows the turn-on rise time as well as the desired effect of lengthening the delay period. But it's simple and seems to work fine for most DRSSTCs.
Solid-state tesla coiling is all about connecting expensive things to dangerous things.
Re: Gate resistor value tradeoff
Dr. Dark Current, Tue Sept 16 2014, 03:48PM
Steve, I'm aware of that, even though the "too fast gate drive can damage a transistor" belief is somewhat unconfirmed.
With the gate charges of usual bricks and the way most people make GDTs, the "smallest resistor value while the gate waveform still looks clean without too much ringing and overshoot" will most often be larger than a value which could cause problems by too fast switching. I was just trying to be helpful without too much information
Dr. Dark Current, Tue Sept 16 2014, 03:48PM
Steve, I'm aware of that, even though the "too fast gate drive can damage a transistor" belief is somewhat unconfirmed.
With the gate charges of usual bricks and the way most people make GDTs, the "smallest resistor value while the gate waveform still looks clean without too much ringing and overshoot" will most often be larger than a value which could cause problems by too fast switching. I was just trying to be helpful without too much information
Re: Gate resistor value tradeoff
chris_inkubate, Tue Sept 16 2014, 05:32PM
Thanks for the additional information from you both and it is always interesting to know a little more stuff as these things sometimes make sense when a dodgy waveform appears on the scope!
The pointer to the test Rg was handy as the Td on/off etc were with a Rg o 3R which added to the internal Rgint of 2.5R gives 5.5R which is in the ball park of what many people seem to be using on IGBT gates. So I'll start with that and see what happens.
Chris
chris_inkubate, Tue Sept 16 2014, 05:32PM
Thanks for the additional information from you both and it is always interesting to know a little more stuff as these things sometimes make sense when a dodgy waveform appears on the scope!
The pointer to the test Rg was handy as the Td on/off etc were with a Rg o 3R which added to the internal Rgint of 2.5R gives 5.5R which is in the ball park of what many people seem to be using on IGBT gates. So I'll start with that and see what happens.
Chris
Re: Gate resistor value tradeoff
chris_inkubate, Sat Sept 20 2014, 09:17PM
I've setup my bridge with 2R gate resistors which added to the internal ones gives 4.5R. I put a resistive load across my bridge output and run it up on 20VDC just to check the basic output of the bridge with a scope. I notice there are some spikes on the rising edges from time to time - would this be a symptom of too small a gate resistance? I just want to get this as good as possible before putting the tank circuit in the mix!
Many Thanks
Bridge output (above)
Here is my gate waveform too (below)
chris_inkubate, Sat Sept 20 2014, 09:17PM
I've setup my bridge with 2R gate resistors which added to the internal ones gives 4.5R. I put a resistive load across my bridge output and run it up on 20VDC just to check the basic output of the bridge with a scope. I notice there are some spikes on the rising edges from time to time - would this be a symptom of too small a gate resistance? I just want to get this as good as possible before putting the tank circuit in the mix!
Many Thanks
Bridge output (above)
Here is my gate waveform too (below)
Re: Gate resistor value tradeoff
Steve Ward, Fri Oct 10 2014, 10:10PM
I didnt think turn-off too fast was ever a problem with IGBTs because of the tail currents being somewhat independent of the gate voltage. I use GDTs almost exclusively, and always use a diode to bypass the gate R for fast turn off.
Ive had good results using fairly large gate resistance (with reverse schottkey diode) on my IGBTs with the primary goal of adding dead-band to allow for zero volt switching (at turn on) of my tesla bridges. This requires a phase-lead type driver so that you can command the IGBT to turn off before the current zero crossings, this way the load current causes the companion IGBT VCE to transition to "0V" BEFORE the IGBT is commanded to be ON. I find this type of tuning gives the cleanest waveforms and minimal overshoot.
You can verify whether or not you have enough dead-time by checking an IGBT's Vce and Vgs simultaneously on an oscilloscope**. You want to see the Vce fall to 0 before Vgs hits the turn on voltage. You can observe the turn on voltage in the VGS waveform: it shows up as a little hesitation in voltage rise at 5-10V typically. If the IGBT turn on is too early, you will see large voltage transients (undershoot) in Vce (and Vgs for that matter) as the turn on is very rapid. If the phase lead is not great enough, then no amount of deadtime will work because this switching method relies on there being enough primary current to charge the IGBT junctions while in the deadtime portion of the switch cycle. I reduce Rgs to avoid excessive dead-time period, so this is more or less how to pick an Rg in my opinion.
For CM300s i found 4.7 ohms to work well with 24V drive through a GDT.
For FGH60N60SMD im using 15 ohms with 2:1 stepdown GDTS for 12V drive.
**Be careful about connecting your oscilloscope ground probe to your power electronics.
Steve Ward, Fri Oct 10 2014, 10:10PM
I believe too small a gate resistor can cause reliability issues due to excessive gate voltage overshoot and collector voltage spikes caused by turning the IGBT off faster than it was designed to go.
I didnt think turn-off too fast was ever a problem with IGBTs because of the tail currents being somewhat independent of the gate voltage. I use GDTs almost exclusively, and always use a diode to bypass the gate R for fast turn off.
Ive had good results using fairly large gate resistance (with reverse schottkey diode) on my IGBTs with the primary goal of adding dead-band to allow for zero volt switching (at turn on) of my tesla bridges. This requires a phase-lead type driver so that you can command the IGBT to turn off before the current zero crossings, this way the load current causes the companion IGBT VCE to transition to "0V" BEFORE the IGBT is commanded to be ON. I find this type of tuning gives the cleanest waveforms and minimal overshoot.
You can verify whether or not you have enough dead-time by checking an IGBT's Vce and Vgs simultaneously on an oscilloscope**. You want to see the Vce fall to 0 before Vgs hits the turn on voltage. You can observe the turn on voltage in the VGS waveform: it shows up as a little hesitation in voltage rise at 5-10V typically. If the IGBT turn on is too early, you will see large voltage transients (undershoot) in Vce (and Vgs for that matter) as the turn on is very rapid. If the phase lead is not great enough, then no amount of deadtime will work because this switching method relies on there being enough primary current to charge the IGBT junctions while in the deadtime portion of the switch cycle. I reduce Rgs to avoid excessive dead-time period, so this is more or less how to pick an Rg in my opinion.
For CM300s i found 4.7 ohms to work well with 24V drive through a GDT.
For FGH60N60SMD im using 15 ohms with 2:1 stepdown GDTS for 12V drive.
**Be careful about connecting your oscilloscope ground probe to your power electronics.
Print this page