Forums

strange MOSFET failure mode

Started by Winfield Hill September 11, 2016
 The subject MOSFET is a C2M0280120D, a serious
 1.2kV SiC TO-247 MOSFET.  The scope pic shows
 driving it with a -3.6 to +16.5V gate pulse,
 100ns long in this case, and it's switching a
 50-ohm resistor from +200V, or 4A (part of my
 fast HV pulser project.)  I previously had
 it making a burst of five 15ns-long 250V 5A
 pulses, and everything was fine, perfect.

 But after fooling around a bit, it stopped
 making 15ns pulses.  Yes, it still turned on 
 in under 10ns, as programmed, but it turned
 off with a 50us time constant.  Scope traces:
https://www.dropbox.com/sh/w0cf8knzcx4t4fj/AABtw_zniLjK1You9fWt6slla?dl=0
 (Ignore the ringing, sloppy probe grounding.)

 Carefully replacing the MOSFET with a new one
 (from a different batch, as it happened), I
 started the signalling pulse generator, and
 observed the exact same failure mode on the
 very first pulse. 

 There are two issues, A. how did I damage
 the part, and B. what is its failure mode:
 able to turn on in <10ns, but with -3.5V
 on its gate, failing to turn off until
 mucho time, 100000ns, has past..

 For A. I suspect a supply-sequencing issue
 (I hadn't installed the delayed HV switch
 on the PCB), but as for B. I've never
 seen anything like it and haven't a clue.


-- 
 Thanks,
    - Win
On 11 Sep 2016 12:11:30 -0700, Winfield Hill
<hill@rowland.harvard.edu> wrote:

> > The subject MOSFET is a C2M0280120D, a serious > 1.2kV SiC TO-247 MOSFET. The scope pic shows > driving it with a -3.6 to +16.5V gate pulse, > 100ns long in this case, and it's switching a > 50-ohm resistor from +200V, or 4A (part of my > fast HV pulser project.) I previously had > it making a burst of five 15ns-long 250V 5A > pulses, and everything was fine, perfect. > > But after fooling around a bit, it stopped > making 15ns pulses. Yes, it still turned on > in under 10ns, as programmed, but it turned > off with a 50us time constant. Scope traces: >https://www.dropbox.com/sh/w0cf8knzcx4t4fj/AABtw_zniLjK1You9fWt6slla?dl=0 > (Ignore the ringing, sloppy probe grounding.) > > Carefully replacing the MOSFET with a new one > (from a different batch, as it happened), I > started the signalling pulse generator, and > observed the exact same failure mode on the > very first pulse. > > There are two issues, A. how did I damage > the part, and B. what is its failure mode: > able to turn on in <10ns, but with -3.5V > on its gate, failing to turn off until > mucho time, 100000ns, has past.. > > For A. I suspect a supply-sequencing issue > (I hadn't installed the delayed HV switch > on the PCB), but as for B. I've never > seen anything like it and haven't a clue.
dV/dT ? I see the spec sheet has parameters for Vds of 800v. Cheers
On 11 Sep 2016 12:11:30 -0700, Winfield Hill
<hill@rowland.harvard.edu> wrote:

> > The subject MOSFET is a C2M0280120D, a serious > 1.2kV SiC TO-247 MOSFET. The scope pic shows > driving it with a -3.6 to +16.5V gate pulse, > 100ns long in this case, and it's switching a > 50-ohm resistor from +200V, or 4A (part of my > fast HV pulser project.) I previously had > it making a burst of five 15ns-long 250V 5A > pulses, and everything was fine, perfect. > > But after fooling around a bit, it stopped > making 15ns pulses. Yes, it still turned on > in under 10ns, as programmed, but it turned > off with a 50us time constant. Scope traces: >https://www.dropbox.com/sh/w0cf8knzcx4t4fj/AABtw_zniLjK1You9fWt6slla?dl=0 > (Ignore the ringing, sloppy probe grounding.) > > Carefully replacing the MOSFET with a new one > (from a different batch, as it happened), I > started the signalling pulse generator, and > observed the exact same failure mode on the > very first pulse. > > There are two issues, A. how did I damage > the part, and B. what is its failure mode: > able to turn on in <10ns, but with -3.5V > on its gate, failing to turn off until > mucho time, 100000ns, has past.. > > For A. I suspect a supply-sequencing issue > (I hadn't installed the delayed HV switch > on the PCB), but as for B. I've never > seen anything like it and haven't a clue.
Are you sure the 50 ohm resistor is OK? -- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
On Sunday, September 11, 2016 at 3:11:42 PM UTC-4, Winfield Hill wrote:
> The subject MOSFET is a C2M0280120D, a serious > 1.2kV SiC TO-247 MOSFET. The scope pic shows > driving it with a -3.6 to +16.5V gate pulse, > 100ns long in this case, and it's switching a > 50-ohm resistor from +200V, or 4A (part of my > fast HV pulser project.) I previously had > it making a burst of five 15ns-long 250V 5A > pulses, and everything was fine, perfect. > > But after fooling around a bit, it stopped > making 15ns pulses. Yes, it still turned on > in under 10ns, as programmed, but it turned > off with a 50us time constant. Scope traces: > https://www.dropbox.com/sh/w0cf8knzcx4t4fj/AABtw_zniLjK1You9fWt6slla?dl=0 > (Ignore the ringing, sloppy probe grounding.) > > Carefully replacing the MOSFET with a new one > (from a different batch, as it happened), I > started the signalling pulse generator, and > observed the exact same failure mode on the > very first pulse. > > There are two issues, A. how did I damage > the part, and B. what is its failure mode: > able to turn on in <10ns, but with -3.5V > on its gate, failing to turn off until > mucho time, 100000ns, has past.. > > For A. I suspect a supply-sequencing issue > (I hadn't installed the delayed HV switch > on the PCB), but as for B. I've never > seen anything like it and haven't a clue. > > > -- > Thanks, > - Win
NASA has specifically characterized those CREE SiC as fairly high reliability. Looks like you've damaged the gate in some weird way so that the gate charge doesn't have a low impedance path for being removed at turn off. Maybe there are limitations on peak gate current you're not observing.
John Larkin wrote...
> > Winfield Hill wrote: > >> The subject MOSFET is a C2M0280120D, a serious >> 1.2kV SiC TO-247 MOSFET. The scope pic shows >> driving it with a -3.6 to +16.5V gate pulse, >> 100ns long in this case, and it's switching a >> 50-ohm resistor from +200V, or 4A (part of my >> fast HV pulser project.)
[ snip ]
>> For A. I suspect a supply-sequencing issue >> (I hadn't installed the delayed HV switch >> on the PCB), but as for B. I've never >> seen anything like it and haven't a clue. > > Are you sure the 50 ohm resistor is OK?
S##t, John, brilliant! Why didn't I think of that? I just measured it and the 50-ohm resistor is toast. Damn, that's the second expensive 100-watt resistor I've blown out on this project. Idea A could have caused that as well. Both MOSFETs are likely fine. -- Thanks, - Win
I was also going to ask if half the gate driver might've given up.  But load 
resistor, yeah.

Wait, so you're not testing into a termination (dummy load)?  :o

Tim

-- 
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Contract Design
Website: http://seventransistorlabs.com

"Winfield Hill" <hill@rowland.harvard.edu> wrote in message 
news:nr4a9202059@drn.newsguy.com...
> > The subject MOSFET is a C2M0280120D, a serious > 1.2kV SiC TO-247 MOSFET. The scope pic shows > driving it with a -3.6 to +16.5V gate pulse, > 100ns long in this case, and it's switching a > 50-ohm resistor from +200V, or 4A (part of my > fast HV pulser project.) I previously had > it making a burst of five 15ns-long 250V 5A > pulses, and everything was fine, perfect. > > But after fooling around a bit, it stopped > making 15ns pulses. Yes, it still turned on > in under 10ns, as programmed, but it turned > off with a 50us time constant. Scope traces: > https://www.dropbox.com/sh/w0cf8knzcx4t4fj/AABtw_zniLjK1You9fWt6slla?dl=0 > (Ignore the ringing, sloppy probe grounding.) > > Carefully replacing the MOSFET with a new one > (from a different batch, as it happened), I > started the signalling pulse generator, and > observed the exact same failure mode on the > very first pulse. > > There are two issues, A. how did I damage > the part, and B. what is its failure mode: > able to turn on in <10ns, but with -3.5V > on its gate, failing to turn off until > mucho time, 100000ns, has past.. > > For A. I suspect a supply-sequencing issue > (I hadn't installed the delayed HV switch > on the PCB), but as for B. I've never > seen anything like it and haven't a clue. > > > -- > Thanks, > - Win
I had a strange case about a year ago. A fet, standard 2n7002, would turnot on at 0V, and off at negative 3V ish

Turned out it was damaged by a leaking current from a year cap

I am not saying your case is the same, just that a damaged fet can act funny

Cheers

Klaus 


On 11 Sep 2016 14:10:06 -0700, Winfield Hill
<hill@rowland.harvard.edu> wrote:

>John Larkin wrote... >> >> Winfield Hill wrote: >> >>> The subject MOSFET is a C2M0280120D, a serious >>> 1.2kV SiC TO-247 MOSFET. The scope pic shows >>> driving it with a -3.6 to +16.5V gate pulse, >>> 100ns long in this case, and it's switching a >>> 50-ohm resistor from +200V, or 4A (part of my >>> fast HV pulser project.) > [ snip ] >>> For A. I suspect a supply-sequencing issue >>> (I hadn't installed the delayed HV switch >>> on the PCB), but as for B. I've never >>> seen anything like it and haven't a clue. >> >> Are you sure the 50 ohm resistor is OK? > > S##t, John, brilliant! Why didn't I think > of that? I just measured it and the 50-ohm > resistor is toast. Damn, that's the second > expensive 100-watt resistor I've blown out > on this project. Idea A could have caused > that as well. Both MOSFETs are likely fine.
What kind of resistor is (was) that? What's your average power dissipation? -- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
John Larkin wrote...
> >Winfield Hill wrote: >> John Larkin wrote... >>> Winfield Hill wrote: >>> >>>> The subject MOSFET is a C2M0280120D, a serious >>>> 1.2kV SiC TO-247 MOSFET. The scope pic shows >>>> driving it with a -3.6 to +16.5V gate pulse, >>>> 100ns long in this case, and it's switching a >>>> 50-ohm resistor from +200V, or 4A (part of my >>>> fast HV pulser project.) >> [ snip ] >>>> For A. I suspect a supply-sequencing issue >>>> (I hadn't installed the delayed HV switch >>>> on the PCB), but as for B. I've never >>>> seen anything like it and haven't a clue. >>> >>> Are you sure the 50 ohm resistor is OK? >> >> S##t, John, brilliant! Why didn't I think >> of that? I just measured it and the 50-ohm >> resistor is toast. Damn, that's the second >> expensive 100-watt resistor I've blown out >> on this project. Idea A could have caused >> that as well. Both MOSFETs are likely fine. > > What kind of resistor is (was) that? What's > your average power dissipation?
The resistor is a Caddock MP9100 in a TO-247 package, rated at 100 watts. It doesn't have much thermal mass; momentary overload rating is only x1.5 for 5 seconds. 20nH inductance gives my 50-ohm version a 400MHz bandwidth. I was dissipating 1.25kW five times, 100ns each; 0.6mJ per trial. The "fooling around" afterwards must have destroyed the resistor. My present circuit board provides for two 100W resistors, and it has a HV shutoff lasting 150ms after the LV is established, to prevent mishaps. I'm going to solder in those parts! A future version of the board has other safeguards. These mishaps are so easy and so deadly. -- Thanks, - Win
On 9/11/2016 8:25 PM, Winfield Hill wrote:
> John Larkin wrote... >> >> Winfield Hill wrote: >>> John Larkin wrote... >>>> Winfield Hill wrote: >>>> >>>>> The subject MOSFET is a C2M0280120D, a serious >>>>> 1.2kV SiC TO-247 MOSFET. The scope pic shows >>>>> driving it with a -3.6 to +16.5V gate pulse, >>>>> 100ns long in this case, and it's switching a >>>>> 50-ohm resistor from +200V, or 4A (part of my >>>>> fast HV pulser project.) >>> [ snip ] >>>>> For A. I suspect a supply-sequencing issue >>>>> (I hadn't installed the delayed HV switch >>>>> on the PCB), but as for B. I've never >>>>> seen anything like it and haven't a clue. >>>> >>>> Are you sure the 50 ohm resistor is OK? >>> >>> S##t, John, brilliant! Why didn't I think >>> of that? I just measured it and the 50-ohm >>> resistor is toast. Damn, that's the second >>> expensive 100-watt resistor I've blown out >>> on this project. Idea A could have caused >>> that as well. Both MOSFETs are likely fine. >> >> What kind of resistor is (was) that? What's >> your average power dissipation? > > The resistor is a Caddock MP9100 in a TO-247 > package, rated at 100 watts. It doesn't have > much thermal mass; momentary overload rating > is only x1.5 for 5 seconds. 20nH inductance > gives my 50-ohm version a 400MHz bandwidth. > > I was dissipating 1.25kW five times, 100ns > each; 0.6mJ per trial. The "fooling around" > afterwards must have destroyed the resistor. > > My present circuit board provides for two > 100W resistors, and it has a HV shutoff > lasting 150ms after the LV is established, > to prevent mishaps. I'm going to solder > in those parts! A future version of the > board has other safeguards. These mishaps > are so easy and so deadly.
Hi, Win - Is it possible that the internal construction can't take the 5A current pulse? The data sheet says it can survive 150W for 5 seconds which is about 1.7A in your application. We have no idea about the internals. Since the resistor appears to be open, I would suspect this failure mode. Call Caddock?