MOSFET AND IGBT fail in pwm speed control for 260VDC 2200W universal motor

strantor

Joined Oct 3, 2010
5,638
EDIT: just saw that strantor was talking about IGBT snubber capacitor modules directly mounted on the switch, these are snubbers , not just capacitors, from what I understand.
They're purely capacitors, but their Extremely high DV/DT & ripple current rating, and extremely low ESR & lead inductance make them perfect for gobbling up spikes (so I'm told, we will see).
 

praondevou

Joined Jul 9, 2011
2,942
They're purely capacitors, but their Extremely high DV/DT & ripple current rating, and extremely low ESR & lead inductance make them perfect for gobbling up spikes (so I'm told, we will see).
Ok, I see. I wonder how much this does increase power dissipation of the IGBT, since when turning it on the cap will be discharged through the IGBT.

Surely they are to be used for these big IGBT modules as shown in the pictures. For TO-247 IGBTs and smaller I've never seen or used such "big" values directly from drain to source of a MOSFET and IGBT.

For this "small" 2kW circuit I'd prefer RCD clamps on the MOSFET, they just cut off everything above the specified limit. :)
 

strantor

Joined Oct 3, 2010
5,638
Ok, I see. I wonder how much this does increase power dissipation of the IGBT, since when turning it on the cap will be discharged through the IGBT.
I would think that power dissipation in the IGBT would be no higher than using a "perfect" power supply with no inductance. I.E. If you ran it in a simulation, you wouldn't see a difference in power dissipation with or without the caps. In real life though, the current is a little slower, so a real life comparison would probably show an increase in power dissipation after adding them. In my case, I want to deliver as much power as possible, as efficiently as possible, so I'd rather store those spikes in a capacitor and reuse them later, rather than burn them off.
Surely they are to be used for these big IGBT modules as shown in the pictures. For TO-247 IGBTs and smaller I've never seen or used such "big" values directly from drain to source of a MOSFET and IGBT.
Yes they are. I did not have a better example handy. These wouldn't be the solution for OP, but something along those lines. Low ESR, high ripple current. That was just to get the juices flowing.
For this "small" 2kW circuit I'd prefer RCD clamps on the MOSFET, they just cut off everything above the specified limit. :)
That's probably a good idea for this application. My "experience" is only with about 10X what he's dealing with and it's only on paper. Your more well rounded actual experience is going to be more valuable from here on out.
another note, are we sure what kind of KW we're looking at? in post #38-#41 It seemed that there was a fatal flaw in the way power was being calculated, but we never got back to that. a big angle grinder like that draws a lot of Amps, especially with a load on it.
 

praondevou

Joined Jul 9, 2011
2,942
In my case, I want to deliver as much power as possible, as efficiently as possible, so I'd rather store those spikes in a capacitor and reuse them later, rather than burn them off.
I wonder what topology u are using and how u want to be able to use the energy stored in the cap which is attached to the drain / source of 1 MOSFET?

You open the transistor, cap gets charged, you close the transistor, cap gets discharged, through the very same transistor.

Unless you are talking about an regenerative snubber I don't see how it's being reused with 1 MOSFET/IGBT in the topology the OP used.:confused:

It's different in a fullbridge where a capacitor like yours is mounted on the bridge bus voltage. Every charge you give them will be reused when turning the bridge back on.

As for the OP's motor resistance/power I think it's exactly what you said. If there is nothing else to limit the current, it will be initially the 260V divided by winding resistance = a lot.
 

strantor

Joined Oct 3, 2010
5,638
I wonder what topology u are using and how u want to be able to use the energy stored in the cap which is attached to the drain / source of 1 MOSFET?

You open the transistor, cap gets charged, you close the transistor, cap gets discharged, through the very same transistor.

Unless you are talking about an regenerative snubber I don't see how it's being reused with 1 MOSFET/IGBT in the topology the OP used.:confused:

It's different in a fullbridge where a capacitor like yours is mounted on the bridge bus voltage. Every charge you give them will be reused when turning the bridge back on.

As for the OP's motor resistance/power I think it's exactly what you said. If there is nothing else to limit the current, it will be initially the 260V divided by winding resistance = a lot.
I'm using high side switching for current sensing/control reasons, but even when I was planning low side switching I planned to have the capacitors both across the motor and the mosfet. the capacitor bank is one of the areas I'm still focusing my studies on; I may be way off the mark on this.
 

strantor

Joined Oct 3, 2010
5,638
even when I was planning low side switching I planned to have the capacitors both across the motor and the mosfet.
...let me expound on that:
reason being, the the lead inductance from the batteries and the discharge rate of the batteries themselves may prevent the ideal amount current from being delivered during a pulse (especially at the beginning of the pulse). The capacitors (across the motor) would deliver the "on demand" current and absorb the inductive spike from the leads and the motor when the switch is opened
 

praondevou

Joined Jul 9, 2011
2,942
...let me expound on that:
reason being, the the lead inductance from the batteries and the discharge rate of the batteries themselves may prevent the ideal amount current from being delivered during a pulse
I really think you mean a capacitor connected like in the attachment

But maybe we leave this for your own thread :)
 

Attachments

Thread Starter

kandilx

Joined Oct 21, 2011
69
now i lost :S , so i will make wires shorter (the colored ones) and i will search for the appropriate "snubber" , and considering the resistance in series how then i can select the appropriate resistance and power value ; i guess i will increase the resistance to get the correct inst. amp, how about the power is it i^2*R or what?
 

Thread Starter

kandilx

Joined Oct 21, 2011
69
one more thing i have just remembered in my early trials i was searching for a zener 300v i didn't found but instead someone told me to use varisitor instead i did try it but it did bruned after short period and before the mosfet fail it was ~275Vac and 350DC i think
then when i read the snubber artilce i remembered
http://en.wikipedia.org/wiki/Snubber
citation>>
"More-sophisticated solid-state snubbers In some DC circuits, a varistor or two inverse-series Zener diodes (collectively called a transorb) may be used instead of the simple diode. Because these devices dissipate significant power, the relay may drop-out faster than it would with a simple rectifier diode. An advantage to using a transorb over just one diode is that it will protect against over voltage with both polarities if, connected to ground, forcing the voltage to stay between the confines of the breakdown voltages of the Zener diodes. A Zener diode connected to ground will protect against positive transients to the value of the Zener breakdown, and will protect against negative transients greater than a normal forward diode drop.
In AC circuits a rectifier diode snubber cannot be used; if a simple RC snubber is not adequate a more complex bidirectional snubber design must be used."<< end of citation
 

strantor

Joined Oct 3, 2010
5,638
one more thing i have just remembered in my early trials i was searching for a zener 300v i didn't found but instead someone told me to use varisitor instead i did try it but it did bruned after short period and before the mosfet fail it was ~275Vac and 350DC i think
You can put multiple zener diodes in series to achieve the zener voltage you need. I would consider a varistor a poor solution, if a solution at all. All it is, is a resistor, whose value changes as it heats. I don't see how that is supposed to act as a voltage clamp
then when i read the snubber artilce i remembered
http://en.wikipedia.org/wiki/Snubber
citation>>
"More-sophisticated solid-state snubbers In some DC circuits, a varistor or two inverse-series Zener diodes (collectively called a transorb) may be used instead of the simple diode. Because these devices dissipate significant power, the relay may drop-out faster than it would with a simple rectifier diode. An advantage to using a transorb over just one diode is that it will protect against over voltage with both polarities if, connected to ground, forcing the voltage to stay between the confines of the breakdown voltages of the Zener diodes. A Zener diode connected to ground will protect against positive transients to the value of the Zener breakdown, and will protect against negative transients greater than a normal forward diode drop.
In AC circuits a rectifier diode snubber cannot be used; if a simple RC snubber is not adequate a more complex bidirectional snubber design must be used."<< end of citation
I have seen transorbs in mosfet switching circuits and it probably wouldn't be a bad idea (wait for praondevou to chime in on this before you go out and buy anything). IIRC I have seen them on the gate as well, and remember that the gate is one of the most sensitive parts of the mosfet. an overvoltage spike on the gate will cause the mosfet to fail without any symptoms of heat.
 

praondevou

Joined Jul 9, 2011
2,942
one more thing i have just remembered in my early trials i was searching for a zener 300v i didn't found but instead someone told me to use varisitor instead i did try it but it did bruned after short period and before the mosfet fail it was ~275Vac and 350DC i think
then when i read the snubber artilce i remembered
There you have it. Your 350VDC varistor burnt.I know that the IRFP460 has a VDS max of 500V, but to burn your varistor you need more than just a few tiny transients, so it looks like the voltage got really much higher.
Btw, varistors or VDRs are similar to zeners and transorb diodes in their function. They can protect sensitive components by shunting the high voltage transients.

However, they are not supposed to work constantly, you put them in the circuit for an extrem condition. I would surely never put in a transorb if I knew that it's breakdown voltage will be constantly exceeded.

For your application I'd prefer the transorb as it usually can dissipate more peak power than a common zener. It doesn't necessarily need to be a bipolar one. Choose a voltage value lower than the max VDS of your MOSFET but not too low. For 500V VDS I guess 450V transorb voltage should be ok.
If excessive voltage spikes ARE the problem, then the transorb will heat up or finally burn, (right before the MOSFET does). If you don't have access to an oscilloscope the transorb will be a good (destructive) test to find out what your problem is.

I wrote earlier you don't need a zener at the gate, but of course strantor is right, if you choose to put a zener with a voltage higher than the gate pulse but lower than VGS max it'll help you to protect the gate.

Finally, you will only be able to develop and test an appropriate snubber if you know what voltage transients you have... so you still need an o-scope :)

Btw, the right way to improve your design is starting with less load and measure voltage transients and peak currents while slowly/gradiently increasing the load. With the right tools you will find out in a matter of minutes what your problem is... without destroying one component after another.;)
 

strantor

Joined Oct 3, 2010
5,638
Ah crap. When I said what I said about the varistor, I had varistor and thermister confused. so disregard what I said in the beginning of post #71.
 

Thread Starter

kandilx

Joined Oct 21, 2011
69
ok good news i did go to the university and i have test but only on the 12v motor using only one battery the square wave looks good and the charging time of the gate 1 uS and i did remove the 4 ohm resistance and connect directly it is the same time i wasn't able to measure the amp , is the time too much??!
 

Attachments

strantor

Joined Oct 3, 2010
5,638
ok good news i did go to the university and i have test but only on the 12v motor using only one battery the square wave looks good and the charging time of the gate 1 uS and i did remove the 4 ohm resistance and connect directly it is the same time i wasn't able to measure the amp , is the time too much??!
It looks good @ 285Hz. Do you plan to run @ 285Hz?
 

strantor

Joined Oct 3, 2010
5,638
Ok then, good deal. Did you happen to scope the output of the mosfet in comparison to the drive signal (I'm assuming your wave pictures are of the gate drive signal)? If your pictures are of the mosfet output then disregard. The reason I ask is that (*I'm not sure of this, but I think) I experienced in **this post that I had a good clean square wave for a gate drive, but mosfet was turning on really slowly. This probably won't be the case since your mosfet is relatively small.

*I was testing 2 different mosfets at the time. one was a small one like yours, and the other one was a huge mosfet module. I know I scoped the gate drive in comparison with the output smaller mosfet, but I don't really remember if I scoped the gate drive in comparison with the output of the large one. I want to think that the output of the mosfet will reflect the gate drive signal.

**please pay absolutely no attention to the circuit on that page, it is likely the poorest performing circuit ever.

All that being said I have a good feeling that you're working with a good circuit and should be ready to move on to the next step.

what's the next step? Have you given any thought to current limiting? I beleive we will see more blown mosfets in your future if you choose to disregard current limiting. Also (alternatively or in addition to current limiting) you may consider using more than one mosfet in parallel to share the load. mosfet paralleling is another learning curve if you want to parallel more than 2 or 3.
 

praondevou

Joined Jul 9, 2011
2,942
ok good news i did go to the university and i have test but only on the 12v motor using only one battery the square wave looks good and the charging time of the gate 1 uS and i did remove the 4 ohm resistance and connect directly it is the same time i wasn't able to measure the amp , is the time too much??!

Ok. If you post waveforms it would be a good thing to explain what they show. Like Strantor I also assume that it is the gate waveform. I repost one of your pictures. How much was the negative peak on gate turn-off? That's not a good thing to have on a gate driver output.

We presume that you might have a problem at exactly that moment I circled, so when gate turns off then the voltage from drain to source increases. But we are talking about a few tens to hundreds of nanoseconds, so you need to zoom into exactly this moment and measure the peak voltage from drain to source again. Just trigger the oscilloscope on the rising edge of the Vds voltage.
It is also a good idea to leave a gate resistor in there. Leave it at 10 Ohm, it won't hurt your MOSFET dissipation due to slower turn-on/off but it will help to protect your gate driver output. (I think I saw 2W for this resistor in one your schematics, this is not necessary)
 

Attachments

Thread Starter

kandilx

Joined Oct 21, 2011
69
Ok then, good deal. Did you happen to scope the output of the mosfet in comparison to the drive signal (I'm assuming your wave pictures are of the gate drive signal)? If your pictures are of the mosfet output then disregard. The reason I ask is that (*I'm not sure of this, but I think) I experienced in **this post that I had a good clean square wave for a gate drive, but mosfet was turning on really slowly. This probably won't be the case since your mosfet is relatively small.

*I was testing 2 different mosfets at the time. one was a small one like yours, and the other one was a huge mosfet module. I know I scoped the gate drive in comparison with the output smaller mosfet, but I don't really remember if I scoped the gate drive in comparison with the output of the large one. I want to think that the output of the mosfet will reflect the gate drive signal.

**please pay absolutely no attention to the circuit on that page, it is likely the poorest performing circuit ever.

All that being said I have a good feeling that you're working with a good circuit and should be ready to move on to the next step.

what's the next step? Have you given any thought to current limiting? I beleive we will see more blown mosfets in your future if you choose to disregard current limiting. Also (alternatively or in addition to current limiting) you may consider using more than one mosfet in parallel to share the load. mosfet paralleling is another learning curve if you want to parallel more than 2 or 3.
i didnt test the output signal of the mosfet :( so maybe today i will do it do you recommend any other tests or measurement b4 going to the uni?

considering current limiting i don't know the way to do it what is then the creiteria of the selection of the resistance or component will be used for current limiting ??

ok i think next time i will test the big motor i will be using 2 mosfet in parallel

last yesterday i did got to the electronics shop and bought a 370 v 40A varasitor and i ask for snubber but i did nt find the seller told me that i can just make one by a resistance and capacitor so my question is howa to select the values of the resistance and capacitor??
 
Top