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

Thread Starter

kandilx

Joined Oct 21, 2011
69
Im unclear on this. Are you saying that you have six irfp460 mosfets in parallel or you have three irfp460s in parallel with three other types? If you meant the latter, I really don't recommend that, because the MOSFET with the lowest Rdson will take the bulk of the current. It is best to have all of the mosfets the same part number and from the same production batch (I.e. Buy them all at the same place at the same time) and match parameters of them. Most important parameter is turn on/ turn off time, followed by Rdson. If you have one MOSFET that turns on slightly before the others or turns off slightly after the others, then if brief periods of time, it will be carrying the full load current, which is not good. It is also important to note, that even with perfectly matched mosfets, once you solder them into the circuit, they may have drastically different turn on/ turn off times due to parasitic inductance in your pcb. This is where Good circuit layout comes into play. Make gate traces all as short as possible and all the same length. Also note that more mosfets is not always better. If you look at the numbers, you may get the idea that it would be best to just throw 20 mosfets into the circuit, but remember that the more mosfets you use, it gets exponentially harder to make them share the load and you have to start derating them by higher and higher factors. The manufacturers recommend that if you are going to parallel more than 4 discreet mosfets, that you start looking into MOSFET modules (expensive)
the 6 mosfet are the same type and batch
before i will test again i will try to make the gate wires all same length as possible as i can
well i get 6 mosfets because one can hold 20A continuous and the start amp is 100 A : 100/20 =5 mosfets the 6th is for safety
 

Thread Starter

kandilx

Joined Oct 21, 2011
69
First and most important objective is to make sure the peak VDS voltage doesn't exceed the maximum breakdown voltage of the MOSFET, 500V in case of the IRFP460.
The picture in my previous post I found on the internet and is actually not exactly correct. By cutting of the upper peak of the ringing voltage the ringing itself will decrease. But the more you cut the more you need to dissipate in the snubber resistor. I would go with the clamp circuit using it directly on the MOSFET and make sure my voltage stays under the 500V.
ok this one is clear


I don't understand this data.
You don't need so many MOSFETs if your continuous current is 10A max.
2 would be sufficient (with a heatsink of course)
i will explain first one >4 ohm gate resistance / 2W power of the resistance :stwitch on time 0uS for only one mosfet / off switching time 1uS for one mosfet , on switching time for 6 mosfets 1uS / off switching time for 6 mosfets 1uS

You mean "to the ground of the battery" or "to the source of the MOSFET" or to the "COM pin of the IR2110". It's not the same thing and it helps us to analyse your results. Start to think of a wire/trace as an electronic component, an inductor.
no it is directly to the ground of the battery , btw is there any other reason not to consider them as one terminal other than the wire inductance??!
1. I consider 6 MOSFETs being to many.
2. Did you use an individual gate resistor for each of them?
3. You can build a more powerful driver with a combination of 2 n-channel and 2 p-channel MOSFETs and a few resistors. But I do not recommend to start yet.
1-so what is the appropriate quantity do you think ?
2-no all connected on 1 gate resistance
3-i did not find a lot of variaty of p mosfet as i did find in n mosfets but i can search again for p mosfet
I do recommend to start with a lower frequency to turn your motor (not 15kHz). If after changing to 15kHz switching losses in the MOSFETs are too high you can still opt for a more powerful driver circuit.
For now we want to determine if voltage transients are the problem.
You still didn't post a picture of the current setup. I'd like to see distance of 2110 to MOSFET, gate wiring, power wiring, motor and freewheeling diode.
well i don't get what your are saying is 15kHz good or bad should work above or below this value knowing that i am testing at around 14400Hz

untill now i didn't prepare for the big motor test i can post the experiment for the small one including the freewheeling diode if you want
and for the the big one i will assemble the snubber circuit first
btw i did find this oscilloscope that uses the soundcad of the pc , i was thinking to use it for the small motor instead of going to the uni every day
http://xoscope.sourceforge.net/hardware/hardware.html
 

praondevou

Joined Jul 9, 2011
2,942
What's the freewheeling diode model number / manufacturer??

no it is directly to the ground of the battery , btw is there any other reason not to consider them as one terminal other than the wire inductance??!
For everything other than pure DC they are not one terminal. If you want to measure a signal then do it as near as possible to the affected device.

Example: If you wanted to know wether the negative spike on the gate driving output of the 2110 is exceeding the 2110's specs then you measure at the IC pins. If you want to see if the negative voltage is exceeding the max negative voltage rating of the MOSFET, then you measure at the MOSFET pins. Same for other measurements. So the reference for Vgs and Vds is the source pin of the MOSFET, not the battery because it's probably several tens of centimetres away.

1-so what is the appropriate quantity do you think ?
2-no all connected on 1 gate resistance
3-i did not find a lot of variaty of p mosfet as i did find in n mosfets but i can search again for p mosfet

Looking at the datasheet again I actually think that 220mOhm is quite high, this + switching losses will dissipate a good amount of power. However, I think 3 MOSFETs for a start should be ok. Usually I use 1 gate resistor for each MOSFET.

well i don't get what your are saying is 15kHz good or bad should work above or below this value knowing that i am testing at around 14400Hz
You probably chose 14.4kHz because the motor won't make so much noise, right?
For a start I would use a lower frequency because switching losses will decrease accordingly. The motor will emit more audible noise, but that shouldn't be a problem for testing purposes.

untill now i didn't prepare for the big motor test i can post the experiment for the small one including the freewheeling diode if you want
and for the the big one i will assemble the snubber circuit first
btw i did find this oscilloscope that uses the soundcad of the pc , i was thinking to use it for the small motor instead of going to the uni every day
http://xoscope.sourceforge.net/hardware/hardware.html
The oscope is AC-coupled and since it uses the soundcard I imagine that it's frequency response is not suitable for your needs. You are measuring frequencies much higher than 14.4kHz.
 

praondevou

Joined Jul 9, 2011
2,942
i was searching for snubber design and i did found this page
http://www.maxim-ic.com/app-notes/index.mvp/id/848

what i dont understand is the difference between these two images


what is the difference between rate of rise and voltage clamp snubbers?
1st image: RCD clamp:

When voltage peaks start to appear on the MOSFET, the diode conducts and starts charging Cclamp. When the voltage on the anode is lower than on the cathode the diode blocks and the capacitor is being discharged to a certain voltage level (say Vclamp) through Rclamp.
When the voltage rises again on the next pulse the diode becomes forward biased again, but only when the voltage on the anode is higher than Vclamp + forward voltage drop of the diode.
That means, as long as the diode does not conduct the voltage rises with the same slope as if there where no snubber at all, but as soon as the diode conducts, the energy that would cause the peak is diverted to the capacitor. The voltage peak is being "cut". The capacitor is only slightly being charged and discharged every cycle, in your case it could be maintained at a maximum level of e.g. 450V.

2nd image: rate of rise

Main difference: for this snubber to limit the rate of rise of the voltage on the MOSFET, the capacitor needs to be discharged every cycle.
When the voltage rises, the diode immediatly conducts and charges the capacitor (slows down the rising slope). When the MOSFET conducts again the diode blocks and the capacitor is being completely discharged through the resistor.

Note that the snubber's correct behaviour depends on the right component values.

Also it is not absolutely necessary to connect the snubber to GND, it could be a higher fixed voltage level if this helps to reduce the resistor's dissipation.

If it's not clear yet, don't hesitate to ask again.
 

Thread Starter

kandilx

Joined Oct 21, 2011
69
well today i had better results after many trialsthe back emf goes from above 40v to 16v(form a supply of 12 v) between the drain and source where this last one was done using a parallel capacitor of 3300uF for the small motor and a snubber circuit using 684nF and 5 ohm 2w resistance in parallel with each other connected to a diode
while being at the uni a physics Dr. tell me to use "filter" or series resistance and parallel capacitor i did remember that when i was looking for the snubber in the electronic shop the seller thought i need a filter he bring one as a sample to show it to me it was this one in the lin k
https://www.distrelec.com/ishopWebF...1&node=DC-22966&artView=true&productNr=112108

well it is writen that it is used for ac and i can see some inductance in the circuit which i think is bad so is there any chance that filters may be used instead of snubber?
 

praondevou

Joined Jul 9, 2011
2,942
The link you posted leads to a differential and common mode noise line filter. Has nothing to do with what you are trying to achieve.

"a parallel capacitor of 3300uF for the small motor". I hope you mean a capacitor from cathode of the freewheeling diode to source of the MOSFET and not anywhere else. ;)
2.5 Ohm and 680nF looks like a rate-of rise snubber right? Do the resistors get hot? You need to consider that with the bigger motor and higher voltage dissipation will increase.

I personally prefer that if you do any modifications or have new ideas and want some advice about them, that you make a drawing, it can be a simple handmade one, on a piece of paper, take a picture.
I don't know if anyone else is still following this thread, but I think it really helps to avoid confusions while explaining circuits with words.
 

shortbus

Joined Sep 30, 2009
10,045
I personally prefer that if you do any modifications or have new ideas and want some advice about them, that you make a drawing, it can be a simple handmade one, on a piece of paper, take a picture.
I don't know if anyone else is still following this thread, but I think it really helps to avoid confusions while explaining circuits with words.
I agree. When the OP is asked about something and doesn't answer (what diode is being used) its hard to tell whats really being done. If the circuit is still on a bread board with full length component leads, its hard to tell what is a real problem or just a 'layout' problem. The long leads are "antennas" to signals and points of possible problems.
 

praondevou

Joined Jul 9, 2011
2,942
I agree. When the OP is asked about something and doesn't answer (what diode is being used) its hard to tell whats really being done. If the circuit is still on a bread board with full length component leads, its hard to tell what is a real problem or just a 'layout' problem. The long leads are "antennas" to signals and points of possible problems.
Nice , someone else is following this. I really would like to have this setup in front of me to see what's going on.
And yes Shortbus you're right, I also asked for a picture of the current layout a few pages ago. :rolleyes:
 

strantor

Joined Oct 3, 2010
6,782
I'm still following (loosely) but haven't commented lately. I'm saving my sweet nuggetts of misinformation for when they're needed most.

I agree; it's a courtesy (and almost always a necessity) to post schematics; otherwise the people who are trying to help you have to devote far more time than should be needed into decifering exactly what is going on. a picture is worth a thousand words. And how many times has this diode been asked about? and still no answer? That gets old after a while.
 

Thread Starter

kandilx

Joined Oct 21, 2011
69
well i'am sorry not put any pictures or photos i hope those in the attachment are enough
i did many trials with different diodes and i didn't see any difference that's why i didn't state the number but any way
schematic 1 (the left one)
first i start using diode D1 1n4004,then BY 399 , and last was 1n5404
and did try many capacitor C1 684, 474, 224 , best results was using 684(or 680nF)
with a resistance R1 =5 ohm 2w the peak was less than 16v
note: C2 the same value 100uF

2 (the one on the right of the paper)

a series resistance R2 was connected (4ohm 4w)
the pulses were 12 volt with very small time(less than 0.5uS) peak of 20v
 

Attachments

Thread Starter

kandilx

Joined Oct 21, 2011
69
The link you posted leads to a differential and common mode noise line filter. Has nothing to do with what you are trying to achieve.

"a parallel capacitor of 3300uF for the small motor". I hope you mean a capacitor from cathode of the freewheeling diode to source of the MOSFET and not anywhere else. ;)
2.5 Ohm and 680nF looks like a rate-of rise snubber right? Do the resistors get hot? You need to consider that with the bigger motor and higher voltage dissipation will increase.
.
resistor R1 didn't get hot at all only R2 in the second schematic
 

praondevou

Joined Jul 9, 2011
2,942
1. What is the value of C3? Did you add this one or is it the 220nF the motor came with. I do not hope that it is the 3300uF. That would be wrong. That's a short circuit when turning on the MOSFET.
2. Well, I didn't see the waveform on C1 but it looks like you're using values that discharge C1 completely each cycle, therefore it would be a rate of rise snubber. If in picture 1 you change the snubber for just a freewheeling diode then what happens?
3. The 1N4004 has a typical reverse recovery time of 30us, so if there was a problem with spurious MOSFET turn-on's after you turned it off the BY399 would be a much better choice since it has a much shorter Trr.
4. Forget R2
5. C2 location is ok (in the schematic at least, couldn't see exactly how connections where made on the breadboard), I would however prefer a bigger capacitance value

I would be interested to see the Vds in picture 1 if you leave everything as it is and replace the snubber by just the freewheeling diode BY399

Then, since it's the MOSFET you want to protect I would also prefer to mount the snubber as near as possible to the MOSFET pins. This could be done by just changing the snubber configuration, i.e. connect it from drain to source (diode pointing upwards)

6. What is the decoupling capacitor value from pin 2 to 3 of the 2110? It looks like a small ceramic cap with let's say 100nF. There is a electrolytic capacitor missing to provide current to the gates.
7. The gate wires are not too long but bringing the driver near your MOSFETs will only help you. I indicated the current path to charge and discharge the gates. In a PCB layout that would not be the right thing to do.

I asked some other questions in the picture I attached.
 

Attachments

Thread Starter

kandilx

Joined Oct 21, 2011
69
1. What is the value of C3? Did you add this one or is it the 220nF the motor came with. I do not hope that it is the 3300uF. That would be wrong. That's a short circuit when turning on the MOSFET.
C3 is 3300uF why short circuit??!
2. Well, I didn't see the waveform on C1 but it looks like you're using values that discharge C1 completely each cycle, therefore it would be a rate of rise snubber. If in picture 1 you change the snubber for just a freewheeling diode then what happens?
if i remember right almost the same
3. The 1N4004 has a typical reverse recovery time of 30us, so if there was a problem with spurious MOSFET turn-on's after you turned it off the BY399 would be a much better choice since it has a much shorter Trr.
4. Forget R2
ok i did try both
5. C2 location is ok (in the schematic at least, couldn't see exactly how connections where made on the breadboard), I would however prefer a bigger capacitance value

I would be interested to see the Vds in picture 1 if you leave everything as it is and replace the snubber by just the freewheeling diode BY399

Then, since it's the MOSFET you want to protect I would also prefer to mount the snubber as near as possible to the MOSFET pins. This could be done by just changing the snubber configuration, i.e. connect it from drain to source (diode pointing upwards)
well C2 is the small one your pointing in the picture you uploaded
i will try to do it next time connecting only BY399 instead of the snubber and connecting the snubber between the drain and source

6. What is the decoupling capacitor value from pin 2 to 3 of the 2110? It looks like a small ceramic cap with let's say 100nF. There is a electrolytic capacitor missing to provide current to the gates.
its 104 capacitor yes 100nF , ok i have 2.2uF and 47uF which is better?
7. The gate wires are not too long but bringing the driver near your MOSFETs will only help you. I indicated the current path to charge and discharge the gates. In a PCB layout that would not be the right thing to do.

I asked some other questions in the picture I attached.
what you mean by "that" is it the long wires ?

the big capacitor you are indicating in the picture is the 3300uf polarized cap
 

praondevou

Joined Jul 9, 2011
2,942
C3 is 3300uF why short circuit??!
Imagine the MOSFET being OFF. C3 will be discharged through the motor winding (if it was charged). You turn the MOSFET on. A discharged capacitor acts like a short circuit. If you really connected it like in the schematic C3 is at the wrong place and you would have had a nice surprise when connecting this circuit to 260VDC.

well C2 is the small one your pointing in the picture.
You have only one cap because you use the same 12V to drive the motor and the driver circuit. You will need a capacitor at the power supply of the 2110. How much depends actually on the amount of gate charge you need. If you have 47uF at hand, use this.
So eventually you will have the capacitor on the 2110 driver power supply AND C2 (much higher voltage rating for the bigger motor) which has to be connected as you drew it in the schematic, from cathode of freewheeling diode to the source of the MOSFET. And I mean exactly "connected to these component pins and not anywhere else".

what you mean by "that" is it the long wires ?
Yes.
 
Last edited:

Thread Starter

kandilx

Joined Oct 21, 2011
69
i was still working praondevou
now i have good news and bad news the good is that i have just finished the assembly of the PCB and the bad news is that i'am entering the army after tomorrow so today i have to try may last try because the past week we were having our feast so the whole country was off including the uni and the stores but i was spending the vacation reading and preparing for today experiment and because friday is the day off in egypt i can't do this try in the university :S
here are the schematic(i used four mosfet parallel each 2 on one 10 ohm resistance) of the circuit i draw it on white board i hope it is clear the picture of the PCB does not include the 2200uF capacitor because i will go to buy it today
the snubber circuit configuration and calculations are all from this link
http://www.daycounter.com/Calculators/Snubbers/Snubber-Design-Calculator.phtml
i was wondering what is the snubber values precision because in some capacitors i did get bigger values and one more thing in the second calculator in the link there is a inductance value is needed i really didn't knew what is the inductance value so i put it as 1 uH
last i did bought a new motor instead of the old one this one is black and decker kg2300
2300w 3.6 ohm resistance
i will test the circuit in about 5 hours so any recommendations you suggest before try?
 

Attachments

praondevou

Joined Jul 9, 2011
2,942
I'm on a mobile device and at work so I can't really write a lot.

Just a few things:

1. Where is the freewheeling diode of the motor? I didn't see it in the schematic.
2. Do not use 12V zener diodes at the gates if your drivers power supply is 12V. Zener diodes have tolerances too so in worst case you will be drawing too much unnecessary current out of the driver, i.e. it will heat up. Use a higher value, e.g. 16 to 18V.
3. Why use a 275V transorb diode if the max Vds of the Mosfet is 500V? Use a higher value but lower than max Vds! They are not supposed to constantly dissipate power. They are protection devices for abnormal conditions.

In your current circuit the transorb most likely and the zeners possibly will overheat.

Most importantly: Start with a lower voltage, not 260V right away.
 

Thread Starter

kandilx

Joined Oct 21, 2011
69
I'm on a mobile device and at work so I can't really write a lot.

Just a few things:

1. Where is the freewheeling diode of the motor? I didn't see it in the schematic.
2. Do not use 12V zener diodes at the gates if your drivers power supply is 12V. Zener diodes have tolerances too so in worst case you will be drawing too much unnecessary current out of the driver, i.e. it will heat up. Use a higher value, e.g. 16 to 18V.
3. Why use a 275V transorb diode if the max Vds of the Mosfet is 500V? Use a higher value but lower than max Vds! They are not supposed to constantly dissipate power. They are protection devices for abnormal conditions.

In your current circuit the transorb most likely and the zeners possibly will overheat.

Most importantly: Start with a lower voltage, not 260V right away.
1 ithought that when i use the snubber thren the diode is not needed , i will connect it 
2i did use 4747 zener i think min is 15 v and max is 17v
3mm well i didnt think aboit this way but iwill start with 70v 
And btw if it get hot or the mosfet fail b4 the transorb fails is these cases will have any signifcant reason?
 

praondevou

Joined Jul 9, 2011
2,942
1 ithought that when i use the snubber thren the diode is not needed , i will connect it 
2i did use 4747 zener i think min is 15 v and max is 17v
3mm well i didnt think aboit this way but iwill start with 70v 
And btw if it get hot or the mosfet fail b4 the transorb fails is these cases will have any signifcant reason?
I know you said you will not have access to the oscilloscope. But IMO you would absolutely need one in order to correctly configure the snubber and to see if voltage peaks are low enough to avoid doing any damage to anything...

If the MOSFETs get hot and your turn-off/on times are ok, you may consider using a bigger heatsink. I remember I used once a heatsink of about 10cm x 20cm with a fullbridge (12 MOSFETs @ 30kHz) and forced ventilation through an attached air duct and the body temperature still went up to 65 degree celsuis (@25 ambient). Don't underestimate the usefulness of a good heatsink and a fan.
 
Top