Author Topic: Ultra fast fuse - are there such devices? And an IGBT question  (Read 8856 times)

0 Members and 1 Guest are viewing this topic.

Offline daqqTopic starter

  • Super Contributor
  • ***
  • Posts: 2301
  • Country: sk
    • My site
Ultra fast fuse - are there such devices? And an IGBT question
« on: February 23, 2017, 08:43:49 am »
Hi guys,

Is there any fuse that could blow in 10-40us at a current of about 1500Amps? The normal operation current would be about 500Amps, but short (2-3 us long), with a repetition rate of around 400 Hz. I'm looking for a fuse that would have a defined behavior for such a situation.

Also, what kind of fault should I suspect when an IGBT got stuck open for an extended period of time? This led to its destruction and is currently completely shorted out. Is there any way an IGBT might get stuck in one state?

The gate driver seems OK and I do not think it gave it a longer pulse.

Best regards,

David
« Last Edit: February 23, 2017, 08:46:21 am by daqq »
Believe it or not, pointy haired people do exist!
+++Divide By Cucumber Error. Please Reinstall Universe And Reboot +++
 

Offline capt bullshot

  • Super Contributor
  • ***
  • Posts: 3033
  • Country: de
    • Mostly useless stuff, but nice to have: wunderkis.de
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #1 on: February 23, 2017, 08:58:19 am »
There's no fuse that is able to protect an IGBT from failing short caused by any kind of failure or mistreat. The IGBT will fail and the fuse will stand long enough for the IGBT to fail.
The fuses sole purpose is to limit the collateral damage in case of IGBT failure (like explosion and starting fire)

Protecting the IGBT can be done by using a gate driver circuit that does incorporate desaturation detection or any other means of fast enough (< 10us) overcurrent detection and immediate turn-off of the IGBT. That won't protect from overvoltages which still may cause the IGBT to fail. Transient overvoltages are a real heavy concern when dealing with high power electronics since they can easily reach double the nominal supply voltage it the layout of the current path incorporates stray inductance.

Disclaimer: I'm not a power electronics specialist, but have many of them as colleagues and often have the possibility to watch failure tests and discussion of how to minimize the collateral damage from power electronics failures

Sorry, missed the original point somewhat: I do not know of a fuse acting this speed, but this is lack of my knowledge, not lack of this kind of fuse maybe existing somewhere in the world. I'm very sure this fuse may cost more than the IGBT if it existed. Did you check the major fuse manufactureres like SIBA
« Last Edit: February 23, 2017, 09:04:53 am by capt bullshot »
Safety devices hinder evolution
 
The following users thanked this post: daqq

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 8110
  • Country: fi
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #2 on: February 23, 2017, 09:12:12 am »
Use active protection in the gate driver. While first developing, scope for voltage ringing, cranking up the DC voltage level gradually, if you see spikes over about 1.2-1.3*DC voltage, something is wrong. Good thing is, you can do this at lower DC link voltage so that you won't blow too many devices in this stage even if you have layout issues. And once you have verified your layout+bypass caps+snubbers this way, the level of ringing is not likely to change too much afterwards.
 
The following users thanked this post: daqq

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #3 on: February 23, 2017, 12:10:09 pm »
...Is there any fuse that could blow in 10-40us at a current of about 1500Amps? The normal operation current would be about 500Amps, but short (2-3 us long), with a repetition rate of around 400 Hz. I'm looking for a fuse that would have a defined behavior for such a situation.

No such thing as a fuse that will blow in 20-40us at just 3x rated current; perhaps at around 100x rated current you'll find fuses are fast enough.  :scared:

That pulse width of 2-3us is a bit narrow for most IGBT modules, though, especially for NPT types which trade a lower switching speed against the ability to withstand dead shorts (usually for 6-10us) by self-limiting the current through them. Any halfway decent overcurrent protection circuit ought to be able to fully protect an NPT IGBT from harm; protecting speedier/higher-gain PT IGBTs and, of course, MOSFETs, is a much more difficult proposition (usually filed under: "the transistor bravely gave its life to protect the fuse").

But the pulse repetition rate is so low I can't think the narrow pulse width is a real problem... unless you thought you could push 500A through, say, a 100-300A IGBT module because the effective duty cycle was so low.

Also, what kind of fault should I suspect when an IGBT got stuck open for an extended period of time? This led to its destruction and is currently completely shorted out. Is there any way an IGBT might get stuck in one state?

I can't think of any failure mechanism for an IGBT which might be caused by it staying open for too long; plenty of circuit/functional failures, of course, but nothing that should damage the IGBT. Well, unless the IGBT is used to dump the charge stored in a capacitor bank that is charged by a constant current source with a compliance voltage that exceeds the IGBT's rating, but that would be a pretty obvious mistake.
 
The following users thanked this post: daqq

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21609
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #4 on: February 23, 2017, 01:20:33 pm »
Is there any fuse that could blow in 10-40us at a current of about 1500Amps? The normal operation current would be about 500Amps, but short (2-3 us long), with a repetition rate of around 400 Hz. I'm looking for a fuse that would have a defined behavior for such a situation.

Sure!  I can get you one for, [counts on fingers and thumbs, thinks for a moment], about $2000 at quantity 10, lead time 20 weeks. :)

Quote
Also, what kind of fault should I suspect when an IGBT got stuck open for an extended period of time?

Open?  So Vge <= 0V?  That'd be a victim of overvoltage.  IGBTs aren't rated for avalanche, and avalanching a transistor isn't healthy, at the best of times, anyway.

Or maybe excessive dV/dt.

Quote
This led to its destruction and is currently completely shorted out. Is there any way an IGBT might get stuck in one state?

Shorting it out certainly gets it stuck in one state... ;D

Quote
The gate driver seems OK and I do not think it gave it a longer pulse.

Mind that when an IGBT fails, it typically fails as a three-way short.  The gate driver probably got blasted with a few hundred volts, current limited by the gate resistance.  It may be okay, or it may malfunction (e.g., bad rise or fall time, excessive heat dissipation, temperature sensitivity), or it may be completely toast.

No such thing as a fuse that will blow in 20-40us at just 3x rated current; perhaps at around 100x rated current you'll find fuses are fast enough.  :scared:

The fastest (conventional passive) fuses in the world are "semiconductor fuses", so called because they can manage to protect some semiconductors, sometimes.  In practice, the only semiconductors robust enough to survive are diodes and SCRs (and I suppose IGCTs), and they really should be considered "suspect" after sinking fault current.

Semiconductor fuses are often used with IGBTs, but not for electrical purposes: the quick breaking action (within a few hundred microseconds -- ten times longer than it takes for the IGBT die to begin turning into magic smoke) minimizes arc flash and shrapnel.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: daqq

Offline daqqTopic starter

  • Super Contributor
  • ***
  • Posts: 2301
  • Country: sk
    • My site
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #5 on: February 23, 2017, 02:11:20 pm »
Hi guys,

Thanks for the answers.
Quote
There's no fuse that is able to protect an IGBT from failing short caused by any kind of failure or mistreat. The IGBT will fail and the fuse will stand long enough for the IGBT to fail.
The fuses sole purpose is to limit the collateral damage in case of IGBT failure (like explosion and starting fire)
Damn, I thought as much :(

Quote
Protecting the IGBT can be done by using a gate driver circuit that does incorporate desaturation detection or any other means of fast enough (< 10us) overcurrent detection and immediate turn-off of the IGBT. That won't protect from overvoltages which still may cause the IGBT to fail. Transient overvoltages are a real heavy concern when dealing with high power electronics since they can easily reach double the nominal supply voltage it the layout of the current path incorporates stray inductance.
I thought I had the driver logic set up in such a way that would limit the pulse length to 2us, the driver has some desaturation protection logic.
As to the transients, well, I've got RC snubbers, a diode clamp both internal in the IGBT and a set of external ones.

Quote
I can't think of any failure mechanism for an IGBT which might be caused by it staying open for too long; plenty of circuit/functional failures, of course, but nothing that should damage the IGBT.
I've been reading a bit on the latch up possibility on an IGBT - uncommon but possible?

Quote
Well, unless the IGBT is used to dump the charge stored in a capacitor bank that is charged by a constant current source with a compliance voltage that exceeds the IGBT's rating, but that would be a pretty obvious mistake.
There was a capacitor bank involved, but it was charged by a current limited power supply at 320V DC. As such the only way that a high voltage could occur in the device would be by means of transient events. The transistor in question is rated up to 1200V ( this fellow: http://www.infineon.com/dgdl/Infineon-FF900R12IP4-DS-v02_04-en_de.pdf?fileId=db3a30431f848401011febb861df3edd ), it was being driven by the 2ED300C17 driver ( http://www.infineon.com/dgdl/Infineon-2ED300C17_S_ST-DS-v04_03-en.pdf?fileId=db3a304412b407950112b429b99741da ).


Quote
No such thing as a fuse that will blow in 20-40us at just 3x rated current; perhaps at around 100x rated current you'll find fuses are fast enough.
After a few hours of searching I have not found any fuse that is specified for times lower than 1ms :-(

Quote
Sure!  I can get you one for, [counts on fingers and thumbs, thinks for a moment], about $2000 at quantity 10, lead time 20 weeks.
You know, I'm tempted : :P

Quote
Mind that when an IGBT fails, it typically fails as a three-way short.  The gate driver probably got blasted with a few hundred volts, current limited by the gate resistance.  It may be okay, or it may malfunction (e.g., bad rise or fall time, excessive heat dissipation, temperature sensitivity), or it may be completely toast.
The driver wasn't happy about it, particularly since it seemed to want to feed the gate which was shorted to the emmiter and collector. Fortunately, the IGBT driver is a two channel one with both of them completely isolated. I tested the driver, it seems to be working. That said, I will replace it, just for developement purposes it seems to serve for now.

Quote
The fastest (conventional passive) fuses in the world are "semiconductor fuses", so called because they can manage to protect some semiconductors, sometimes.
Thanks! I didn't now about these, I'll look into them.
Believe it or not, pointy haired people do exist!
+++Divide By Cucumber Error. Please Reinstall Universe And Reboot +++
 

Offline capt bullshot

  • Super Contributor
  • ***
  • Posts: 3033
  • Country: de
    • Mostly useless stuff, but nice to have: wunderkis.de
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #6 on: February 23, 2017, 02:59:47 pm »
From my knowledge, driving an IGBT of the size you mention (I've seen many of these modules in many colours and shapes :-) with a gate pulse width of 2us only is simply too short for the IGBT to completely turn on / turn off.
The power guys here emphasize that there is a minimum on time when using IGBTs, and the PWM generator must have a protection means against too short on time (also applies to off time). I don't know the exact figures, but turning on a IGBT for 2us only is too short. Doing that repetively will destroy the IGBT.

Longer pulses will be less critical, I believe. 2us is too short for any usual protection mechanism. If you need to generate such short pulses, in my opinion you'll have to use MOSFETs
Safety devices hinder evolution
 
The following users thanked this post: daqq

Offline daqqTopic starter

  • Super Contributor
  • ***
  • Posts: 2301
  • Country: sk
    • My site
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #7 on: February 23, 2017, 03:53:22 pm »
Quote
From my knowledge, driving an IGBT of the size you mention (I've seen many of these modules in many colours and shapes :-) with a gate pulse width of 2us only is simply too short for the IGBT to completely turn on / turn off.
Hello,

This is news to me, I've never noticed this mentioned anywhere  :o So, if an IGBT is turned on and then turned off before some time it will not be happy about that? How can I deduce the minimal time for this occurrence?

But are you sure about this? Even adding up all of the times that are in the datasheet ( http://www.infineon.com/dgdl/Infineon-FF900R12IP4-DS-v02_04-en_de.pdf?fileId=db3a30431f848401011febb861df3edd ) yields a lower time than 2us. And I've had my setup running like that for half an hour and nothing happened. Also, looking at the pulses during normal operation it seemed that the transistor got fully opened well before it started closing.

The pulses are meant to be 2us long and spaced 2.5ms apart - as such the transistor should have plenty of time to turn down.
Quote
If you need to generate such short pulses, in my opinion you'll have to use MOSFETs
I was hoping to avoid MOSFETs for this - for those kinds of currents/voltages they are not exactly common place. Also, the very high turn on speed might create all kinds of problems  :(
Believe it or not, pointy haired people do exist!
+++Divide By Cucumber Error. Please Reinstall Universe And Reboot +++
 

Offline capt bullshot

  • Super Contributor
  • ***
  • Posts: 3033
  • Country: de
    • Mostly useless stuff, but nice to have: wunderkis.de
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #8 on: February 23, 2017, 04:23:51 pm »
As I said, I don't have any exact figures about this, just the estimates I've got from watching other peoples work. For example: in the most general application of these modules (generating pwm for driving inverters or motors) minimum turn on/off times are a real concern, they have to be longer than the delay of the protection circuit which otherwise wouldn't work. There are some special cases when the protection circuit would not detect a short circuit, and the pwm generator would issue a small duty cycle (caused by control loop operation). This situation leads to IGBT failures, I was told. It's not totally your situation, but somehow the like.

What I know is: data sheets don't tell the full truth. Sometimes there are application notes available, going more into depth, sometimes not. IGBT manufactures do a lot of optimization for the main purpose of the IGBT, different from generation to generation. They don't tell you all the implications of the optimizations. I'd consider your use case a special case that is not necessarily covered by the usual use cases. If you want to avoid MOSFETs, I'd advise to check this use case with a Infineon FAE.
Safety devices hinder evolution
 
The following users thanked this post: daqq

Offline Marco

  • Super Contributor
  • ***
  • Posts: 6694
  • Country: nl
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #9 on: February 23, 2017, 05:32:52 pm »
From my knowledge, driving an IGBT of the size you mention (I've seen many of these modules in many colours and shapes :-) with a gate pulse width of 2us only is simply too short for the IGBT to completely turn on / turn off.

Could this be because of the free wheeling diode when used in a bridge? (See page 11.) The snap off behaviour of silicon diodes is on-time dependent (Russians engineers called this drift step recovery, they used it for high voltage pulse generators).

If that could be a problem an IGBT without an internal FWD, or an internal SIC FWD, could be a solution.
« Last Edit: February 23, 2017, 06:16:45 pm by Marco »
 
The following users thanked this post: daqq

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #10 on: February 23, 2017, 08:39:23 pm »
Quote
From my knowledge, driving an IGBT of the size you mention (I've seen many of these modules in many colours and shapes :-) with a gate pulse width of 2us only is simply too short for the IGBT to completely turn on / turn off.
Hello,

This is news to me, I've never noticed this mentioned anywhere  :o So, if an IGBT is turned on and then turned off before some time it will not be happy about that? How can I deduce the minimal time for this occurrence?



Your application is a bit unusual in that you running the IGBT at an extremely low effective duty cycle (less than 0.1%), but generally speaking, you want the minimum on time to be at least twice as long as the sum of all the switching time components (ie - on delay, rise time, off delay, fall time). Your IGBT is fairly fast for 1200V NPT technology (gotta love those Infineon dice), but even still it has a total transition time (on + off) of 1.57us at 125C, so the minimum pulse width should be >3us. With a 2us pulse the IGBT is only fully on for less than 500ns; the rest of the time it is acting more like a resistor than a switch!

If you need to deliver precise pulse widths at this time scale you need to be using MOSFETs. Also, don't look at the voltage waveform from the IGBT, measure the current instead. I suspect it will look more like a triangle than a pulse.

One other thing - how are you actually driving the IGBT gate? Is it, by any chance, a bootstrap/charge-pump high side driver IC?

 
The following users thanked this post: daqq

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16548
  • Country: us
  • DavidH
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #11 on: February 24, 2017, 04:51:52 am »
Quote
If you need to generate such short pulses, in my opinion you'll have to use MOSFETs

I was hoping to avoid MOSFETs for this - for those kinds of currents/voltages they are not exactly common place. Also, the very high turn on speed might create all kinds of problems  :(

The failure you describe reminds me a lot of what happens to bipolar transistors and MOSFET when their secondary breakdown rating is exceeded.

This might be an ideal application for a high voltage bipolar transistor cascode switched by a low voltage MOSFET except that the IGBT module is much better for simplicity and power density.

Long ago I know I saw a base switched power bipolar switch circuit which used a MOSFET in series with the emitter for a super fast overload shutoff but I do not remember why they did it that way and I suspect it would not work with an IGBT because there is no place for stored charge to go.
 
The following users thanked this post: daqq

Offline daqqTopic starter

  • Super Contributor
  • ***
  • Posts: 2301
  • Country: sk
    • My site
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #12 on: February 24, 2017, 08:15:09 am »
Damn... :( IGBTs seemed ideal for this on paper.
Quote
One other thing - how are you actually driving the IGBT gate? Is it, by any chance, a bootstrap/charge-pump high side driver IC?
No, I'm using an Infineon 2ED300C17 driver ( http://www.infineon.com/dgdl/Infineon-2ED300C17_S_ST-DS-v04_03-en.pdf?fileId=db3a304412b407950112b429b99741da ). It has two completely isolated +/-15V push pull output stages.

Quote
Open?  So Vge <= 0V?  That'd be a victim of overvoltage.  IGBTs aren't rated for avalanche, and avalanching a transistor isn't healthy, at the best of times, anyway.
To the best of my knowledge there was no overvoltage present - the was a 640uF capacitor bank charged to +320V and the transistor is rated to 1200V. There were also snubbers (RC, resistor plus diode and the built in free wheeling diode of the transistor), so neither a positive nor negative overvoltage should have been present.


So, since IGBTs are a no go, I'll have to go with a MOSFET. Would you say that paralleling say, 6 of these: http://www.farnell.com/datasheets/2173895.pdf would work? The load is a pulse transformer, with a high voltage output.
Believe it or not, pointy haired people do exist!
+++Divide By Cucumber Error. Please Reinstall Universe And Reboot +++
 

Offline capt bullshot

  • Super Contributor
  • ***
  • Posts: 3033
  • Country: de
    • Mostly useless stuff, but nice to have: wunderkis.de
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #13 on: February 24, 2017, 08:54:15 am »
If you go for cost optimization, I'd say you might get away with 3 of the Ixys Mosfets ... just do the calculations right and do extensive testing.
SOA diagram looks good for this purpose.
Inductive kickback from the transformer easily kills any semiconductor device driving it, so be very careful selecting the catch diode.
Safety devices hinder evolution
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 37664
  • Country: au
    • EEVblog
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #14 on: February 24, 2017, 10:39:34 am »
Is there any fuse that could blow in 10-40us at a current of about 1500Amps? The normal operation current would be about 500Amps, but short (2-3 us long), with a repetition rate of around 400 Hz. I'm looking for a fuse that would have a defined behavior for such a situation.

Not that fast at only 3 times the rated current, you need a different solution.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21609
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #15 on: February 24, 2017, 11:58:51 pm »
As I said, I don't have any exact figures about this, just the estimates I've got from watching other peoples work. For example: in the most general application of these modules (generating pwm for driving inverters or motors) minimum turn on/off times are a real concern, they have to be longer than the delay of the protection circuit which otherwise wouldn't work. There are some special cases when the protection circuit would not detect a short circuit, and the pwm generator would issue a small duty cycle (caused by control loop operation). This situation leads to IGBT failures, I was told. It's not totally your situation, but somehow the like.

There's no longer any need to speak in generalisms, or rules of thumb*.  OP posted a link to the datasheet.  We can talk in specifics now.

Question: how would it be necessary for the on-time to be longer than the protection circuit?

I mean, isn't that the point?  The protection is supposed to turn it off as soon as a fault has been detected.  Sooner is always better, right?

If there's not even enough time for a fault to be detected, then HELL YEAH IT'S SAFE TO FIRE! :o

dV/dt, peak V, and TJ all need to be limited carefully, but the turn-on phase is fine.


*I hate rules-of-thumb with a particular ire.  1. They're almost always wrong, anyway.  2. NO ONE using them, understands where they come from, or 3. how to prove or disprove them.

Using rules of thumb shows ones ignorance of the subject!

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: daqq

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21609
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #16 on: February 25, 2017, 12:01:39 am »
To the best of my knowledge there was no overvoltage present - the was a 640uF capacitor bank charged to +320V and the transistor is rated to 1200V. There were also snubbers (RC, resistor plus diode and the built in free wheeling diode of the transistor), so neither a positive nor negative overvoltage should have been present.

Could you please draw us a schematic already, and take a picture of the layout?



Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #17 on: February 25, 2017, 02:18:00 pm »
...
Question: how would it be necessary for the on-time to be longer than the protection circuit?

I mean, isn't that the point?  The protection is supposed to turn it off as soon as a fault has been detected.  Sooner is always better, right?

If you use Vce as the proxy for current then you have to delay measuring it until the IGBT is fully saturated, which at a minimum needs to be longer than the sum of the IGBT delay and rise times, plus the overshoot/ringing times (the latter being dependent not only on various strays, but also on the circuit conditions; e.g. - delay/rise times change with temperature while overshoot/ringing amplitude changes with current). Furthermore, faster response time automatically means higher frequency response and that means a higher sensitivity to noise, EMI, etc... So, no, sooner is not always better!

Even if the actual current through the IGBT is measured- rather than its Vce - you'll still need to window out or integrate away the overshoot (or ringing) from charging up stray capacitances; the former enforces a minimum on time/maximum frequency while the latter slows down response time in general (ie - if inductor saturation occurs later in the cycle it will take longer to detect it).

(Hence, btw, why pretty much all overcurrent comparator functions in SMPS controller ICs window out the first 50-100ns of the current sense signal.)
« Last Edit: February 25, 2017, 02:38:23 pm by MagicSmoker »
 
The following users thanked this post: daqq

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #18 on: February 25, 2017, 02:36:26 pm »
I confirm that the only kinds of semi-conductors you can protect against over-current by a fuse are diodes, scr's and triacs.
For these semi-conductors, manufacturer specify the I²dt of the chip.
You need to use ultra-fast fuses with a I²dt at least 30% lower to protect the semi-conductor.
It works very well....I have projected a lot of thyristorized DC drives and industrial battery chargers protected by UF fuses and it worked well when needed.
http://www.ferrazfuses.com/

There is no way to protect other semi-conductors by UF fuses....
NB:if possible, always place the fuse in the AC side, not in the DC side.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21609
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #19 on: February 25, 2017, 05:22:48 pm »
If you use Vce as the proxy for current then you have to delay measuring it until the IGBT is fully saturated, which at a minimum needs to be longer than the sum of the IGBT delay and rise times, plus the overshoot/ringing times (the latter being dependent not only on various strays, but also on the circuit conditions; e.g. - delay/rise times change with temperature while overshoot/ringing amplitude changes with current). Furthermore, faster response time automatically means higher frequency response and that means a higher sensitivity to noise, EMI, etc... So, no, sooner is not always better!

Yeah, but, that's all about the desat circuit being able to make a confident decision -- and for the design of that circuit, this is spot on -- but that's a separate circuit, not the IGBT die itself.  While the IGBT is turning/turned on, it's sitting there heating up.  Two microseconds of on-time is strictly greater than the heating of one microsecond, and so on.

If the on-time is too short for a desat to decide, then that simply means you don't need it, and it can't help you anyway.  That'd be like walking into a store, and asking an attendant where an item is, while you are currently en route to the very item that you already know is there (or not).  It doesn't matter! ;)

(You could, however, perhaps, design a pipelined desat circuit: a circuit that measures the output waveform amplitude, and determines if it reached saturation for enough of the pulse width.  Based on this information, the control circuit could decide if the next pulse is safe to fire, or if it should stop instead.)

Tim
« Last Edit: February 25, 2017, 05:25:17 pm by T3sl4co1l »
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline daqqTopic starter

  • Super Contributor
  • ***
  • Posts: 2301
  • Country: sk
    • My site
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #20 on: February 28, 2017, 06:11:46 pm »
My apologies for the delay  :(
To the best of my knowledge there was no overvoltage present - the was a 640uF capacitor bank charged to +320V and the transistor is rated to 1200V. There were also snubbers (RC, resistor plus diode and the built in free wheeling diode of the transistor), so neither a positive nor negative overvoltage should have been present.

Could you please draw us a schematic already, and take a picture of the layout?



Tim
My apologies, schematic of the setup that blew is here. There is a snubber board consisting of a 4x RC snubbers and 4x flyback diodes with a resistor (to dissipate the kickback more quickly).


Now, I've been doing some measurements and when the other side of the transformer is unloaded (can happen in the real situation) there's an oscillation whose peaks reach up to 1.5x of the capacitor banks DC value. That said, at +320V this should still not be a problem, seeing as that's still less than half of the transistors voltage rating.

Quote
If you go for cost optimization, I'd say you might get away with 3 of the Ixys Mosfets ... just do the calculations right and do extensive testing.
SOA diagram looks good for this purpose.
Inductive kickback from the transformer easily kills any semiconductor device driving it, so be very careful selecting the catch diode.
Thanks, I'll think about that option - the IGBT just seemed wonderfully compact and overspecced for the situation.

While we're at it, is there some kind of weird arcane rule of thumb that the minimal pulse time of a MOSFET should not be less than something?
Believe it or not, pointy haired people do exist!
+++Divide By Cucumber Error. Please Reinstall Universe And Reboot +++
 

Offline capt bullshot

  • Super Contributor
  • ***
  • Posts: 3033
  • Country: de
    • Mostly useless stuff, but nice to have: wunderkis.de
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #21 on: February 28, 2017, 09:07:39 pm »
As I see your schematic (I did imagine something like that, but would have put the switch into the low side) - what about a short circuit on the secondary side? This will transform to a non-ideal kind of short circuit on the primary side and surely exceed the nominal current (but limited by the stray inductance of the transformer). For this case, the maximum current of the MOSFETs should be somewhat greater than the expected short circuit current. The SOA diagram tells that the MOSFETs should survive this for 2us pulses as long as you don't exceed the max. pulse current of each MOSFET. So maybe more than three of them might be appropriate.

In case of open load or some kind resonance / oscillation, the full on-state current would go into the diodes and 1.5R resistors, thats roughly calculated 190V (as you observed). Calculate this for the maximum expected current on the primary side in case of any failure, this will be the voltage that the MOSFETs will have to stand. I'd go for smaller resistor values, there's plenty of time for the magnetic field to dissipate until the next pulse.

To generate clean pulses, the turn-on duration should be sufficiently longer than the sum of rise and fall time of the MOSFET, but I do not know of a failure mechanism caused by too short pulses.

If you want fast switching action, you should not overspec your switches (smaller switches often will be faster)
Safety devices hinder evolution
 
The following users thanked this post: daqq

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21609
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #22 on: February 28, 2017, 09:31:16 pm »
"Well, there's yer problem" -- maybe.

Without a picture, I don't know how distant that "snubber board" is, but if it's more than a few inches, your problem is the loop from capacitor to transistor to ground.  With added resistance, and stray inductance, transistor peak voltage can easily become excessive.

Move the diodes (which are handling the load flyback energy) as close to the transistor and caps as possible.  Or, perhaps, slow the gate drive so it spends more time turning off?

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: daqq

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 8110
  • Country: fi
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #23 on: March 02, 2017, 07:01:19 pm »
Although a schematic is not a layout description, that schematic screams a layout problem, IMO, so you should post some photos of the setup.

Low-inductance path between the freewheeling diode, now located on "snubber board" (?), the IGBT, and the input capacitor is paramount.

That's why they produce co-packed half-bridge IGBT modules with the freewheeling diode (and thus, the third therminal). This allows you to directly mount a low-ESL, medium ESR DC link cap, sometimes called DC link snubber cap, to the screw terminals, and use the third terminal as the output. Any RC snubbers over either the switching IGBT, or the freewheeling diode (as you have them now), or both, can be directly mounted on those terminals, too.
 
The following users thanked this post: daqq

Offline daqqTopic starter

  • Super Contributor
  • ***
  • Posts: 2301
  • Country: sk
    • My site
Re: Ultra fast fuse - are there such devices? And an IGBT question
« Reply #24 on: March 07, 2017, 01:03:06 pm »
So, a bit of an update:

Unfortunately I can't take a photo of the setup :(

In the normal operation (when the transformer is loaded) there is nothing strange going on and it performed well at nominal load with no problems.

We have traced the blowing up problem to the situation when the transformer is not loaded - as such all of the energy that gets put into the transformer starts wobbling about, the transformer (and everything) starts acting as an LC resonator and all sorts of nasty happens.

The peak voltage on the transformer (tested at DC on the caps of +60V) becomes around +80-90V. This then goes through the diode that is parallel to the transistor back to the capacitor bank. Apparently, it's a pretty hard source, since based on a current measurement we are getting hundreds of amps from that backpulse. If this scales with the voltage, the current on the backpulse could have been thousands of amps for the 320V.

So, at the moment it seems that the best I can do is get a bunch of really fast diodes, put them in parallel with the main FWD diode of the IGBT but in a physically shorter path.


Also, good news on the fuse front! An ET20 ( http://uk.farnell.com/bussmann-by-eaton/20et/fuse-hrc-high-speed-20a/dp/1202331 ) managed to work properly when the load on the transformer was present and blow when the transformer was unloaded. So, the transistor survived another high voltage experiment that would have blown him :-BROKE . I'm going to do a little teardown to see what makes it break :)

Best regards,

David
Believe it or not, pointy haired people do exist!
+++Divide By Cucumber Error. Please Reinstall Universe And Reboot +++
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf