00:00:00
hi I'm samanyakov this presentation is
00:00:03
entitled intuitive explanation of
00:00:05
silicon carbide thermal impedance safe
00:00:08
operating area and empty spy simulation
00:00:11
there is a disclaimer here that the
00:00:14
devices and simulation tools shown in
00:00:16
this presentation are for educational
00:00:19
purposes and only there is no
00:00:22
endorsement or recommendation implied
00:00:25
the issue that I'm addressing in this
00:00:28
presentation is the thermal problem of a
00:00:33
transistor and in particular The
00:00:35
Junction temperature
00:00:37
so here is an equivalent circuit a
00:00:39
thermal equivalent circuit in which the
00:00:42
thermal behavior is emulated by an
00:00:45
electrical circuit in this case the
00:00:48
current Source represent the power which
00:00:50
is pumped into the transistor that is
00:00:53
the voltage and current through it and
00:00:56
I'm primarily interested in this
00:00:58
presentation in the linear mode
00:01:01
inductive not the switching mode the
00:01:04
transistor is completely on and there is
00:01:06
the earlier Zone but rather when you
00:01:08
have a voltage across the transistor any
00:01:11
current passing through it so this will
00:01:14
represent the power dissipated within
00:01:17
the unit these are the thermal
00:01:20
resistances
00:01:22
this is the junction temperature is
00:01:25
represented by the voltage here
00:01:27
case temperature is the voltage here and
00:01:31
this is the thermal resistance between
00:01:33
Junction and case also I'm showing here
00:01:37
the thermal resistance between the case
00:01:40
and ambient
00:01:41
and then we have the ambient temperature
00:01:44
represented by a clamped by a voltage
00:01:48
source now these capacitors represent
00:01:52
thermal capacitances and that is the
00:01:55
amount of heat that is absorbed by the
00:01:58
mass you need it for transient in DC you
00:02:02
don't need it that is for steady state
00:02:04
you don't need it because you just
00:02:06
consider the resistance but if you have
00:02:10
a pulse coming in then temperature
00:02:13
starts rising and some of the heat is
00:02:15
being absorbed by the mass so this
00:02:18
capacitors represent the
00:02:21
terminal capacitances within the area of
00:02:24
the dye and the structure as a matter of
00:02:27
fact there is a need for a distributed
00:02:30
presentation because the heat is
00:02:33
propagating throughout the die you might
00:02:36
say it's like a three-dimensional
00:02:38
thermal transmission line and therefore
00:02:42
the best way is of course to use this
00:02:45
the leather type of a presentation
00:02:48
although some cases you can get by by a
00:02:51
simpler structure so with this
00:02:55
equivalent circuit you can probe into
00:02:57
the issue of the junction temperature in
00:03:01
the case that you have a current with
00:03:03
some background current and there is a
00:03:05
pulse and then some more current and the
00:03:08
pulse or any other profile of a car
00:03:11
manufacturers are giving in the data
00:03:14
sheet information regarding the
00:03:16
transient behavior of the junction
00:03:19
temperature and one of them is they save
00:03:22
operating area plot which is shown here
00:03:25
and let me go over it very quickly the
00:03:28
y-axis is current the drain current the
00:03:31
x-axis is the drain to Source voltage
00:03:33
this device is a 1200 volt device
00:03:38
silicon carbide 25 million quite a hefty
00:03:43
transistor and then we have boundaries
00:03:45
here this
00:03:47
right boundary is actually the maximum
00:03:50
voltage that you can expose the
00:03:53
transistor to 1200 volt and then there
00:03:57
is a upper boundary which is the maximum
00:04:00
current and then this boundary here is
00:04:03
actually the RDS on because if the
00:04:06
voltage is low you cannot go above
00:04:10
certain currents which are limited by
00:04:12
the resistance of the S1 okay so in fact
00:04:16
each point here represents a ratio
00:04:19
between voltage and current to give you
00:04:22
the audio song so this is this boundary
00:04:25
here and then we have these curves here
00:04:28
and these are four pulses so if I have a
00:04:31
pulse of say 100 microseconds
00:04:35
in order to be in the safe
00:04:38
Junction temperature you have to be
00:04:40
below this line meaning that the
00:04:42
boundary here let's say take for example
00:04:45
if the voltage of the transistor is 100
00:04:48
volt then
00:04:51
you can have a current of the pulse of
00:04:55
the 100 microsecond of about I'd say 50
00:05:00
M okay so as long as you are below this
00:05:05
line you are okay if you are about this
00:05:08
line you are supposed to
00:05:09
violate this maximum Junction
00:05:12
temperature unfortunately Cree does not
00:05:15
give you what is the maximum Junction
00:05:20
temperature for these curves
00:05:22
well the maximum Junction temperature
00:05:24
for the transistor of 3 wolf speed is
00:05:29
150 degree so you might have thought
00:05:33
that this is 150 degrees well we'll
00:05:36
check it later on on the other hand some
00:05:38
windows like St of course also give this
00:05:41
safe operating area plot in their data
00:05:44
sheet and then this particular unit that
00:05:47
they have the silicon carbide has
00:05:50
specified the maximum operating
00:05:53
temperature of 200 Centigrade as opposed
00:05:58
to 150 for world speed I I really don't
00:06:01
understand the very big difference in
00:06:04
any event
00:06:05
regarding our discussion here they are
00:06:08
showing the maximum temperature for
00:06:11
these plots that they're saying okay
00:06:13
these plots are for a maximum of 200
00:06:18
degrees and of course we have the same
00:06:20
situation here we've got different
00:06:22
pulses of different width however this
00:06:26
safe operating area is is good for the
00:06:29
boundary but does not give you
00:06:32
information of say suppose you have a
00:06:35
100 microsecond pulse okay and let's say
00:06:39
they
00:06:40
voltage across the transistor is 100
00:06:43
volt what is the junction temperature
00:06:45
well you don't get it from here what you
00:06:48
get here is that you can go up to a
00:06:51
current of this value but what about
00:06:54
this case this current you just don't
00:06:58
have this information here you can get
00:07:01
it from this plot which is the thermal
00:07:04
impedance
00:07:05
and what it is it sort of takes into
00:07:10
account the fact that if you have pulses
00:07:13
the temperature are not of course
00:07:15
stabilized but they are sort of a
00:07:18
transient temperature and this will help
00:07:21
you to find out what is the maximum
00:07:23
temperature due to pulses so
00:07:27
let's just go over it very quickly in
00:07:30
general and then I'll probe into it a
00:07:32
little bit more in depth what we have
00:07:35
here this is the thermal impedance that
00:07:38
is the thermal resistance during
00:07:41
transient this is the meaning of this
00:07:43
thing this is the pulse
00:07:46
width okay so we have one microsecond up
00:07:49
to 100 milliseconds and here I'm
00:07:52
concentrating now on one
00:07:55
curve here which is the Single part this
00:07:58
is the curve okay it goes up and then it
00:08:00
sort of clamps to this value which is
00:08:04
the DC thermal resistance between
00:08:07
Junction and K so this is what is given
00:08:10
is the data sheet as The Junction to
00:08:13
case thermal resistance in this
00:08:15
particular case for this particular
00:08:17
transistor it's 0.24 degrees per watt
00:08:22
okay but if the pulses are short if the
00:08:26
pulses are short not long passes for
00:08:30
example 100 microsecond where at this
00:08:34
point and at this point we have a
00:08:36
certain value okay so this would be like
00:08:39
uh 20 10 to the 10 to the minus
00:08:43
3.03 okay so this will be the value here
00:08:47
what does it mean it means that if the
00:08:51
pulse is short then the thermal
00:08:56
resistance that you have to take into
00:08:58
account is not this wonderful steady
00:09:00
state but rather they want for this
00:09:03
particular case of the width of the
00:09:06
pulse
00:09:08
and then we're talking here about a
00:09:12
single path okay so this is this curve
00:09:15
here we'll talk about these later on so
00:09:17
let's have a look what is the meaning of
00:09:19
this single pulse line here or curve
00:09:23
here okay so let's assume that we have a
00:09:26
pulse of a power injected into the
00:09:30
transistor okay this is the pause
00:09:32
hundred microsecond width and there's a
00:09:35
second height for the power
00:09:38
now obviously the temperature will start
00:09:40
going up
00:09:42
if this pulse is much longer we are in
00:09:47
the continuous steady state situation
00:09:49
and then of course they
00:09:53
temperature of the junction or Junction
00:09:55
to case will be the power times the
00:09:58
thermal resistance between Junction and
00:10:01
K however if the pulse is short
00:10:04
temperature will rise only to this point
00:10:07
okay so obviously uh just the power
00:10:10
times the steady state thermal
00:10:13
resistance is incorrect it has to be
00:10:15
shorter this is exactly this line I have
00:10:18
to sort of Mark this as the thermal
00:10:22
resistance Junction to case for pulse
00:10:24
okay so this pulse is 10 microsecond and
00:10:29
for 10 microsecond the value of this
00:10:32
thermal resistance pulse is seven Milli
00:10:36
degree pair what as opposed to
00:10:40
240 Milli degree
00:10:43
Centigrade per Watts so it's a much much
00:10:46
lower one
00:10:48
thermal resistance because it only got
00:10:51
to this point so when you multiply the
00:10:54
power by this value from this curve
00:10:57
you'll get the right Junction
00:11:00
temperature
00:11:01
if the pulse is wider then you'll get to
00:11:06
a higher point and obviously you have to
00:11:09
look at the corresponding pulse width
00:11:13
which is 100
00:11:14
micro second it'll be here and this will
00:11:18
be 21 and this will be this point here
00:11:21
so this is the meaning of this plot
00:11:24
which is very neat for a single part
00:11:26
however if you have a train of pulses
00:11:30
that is you have some frequency with
00:11:34
pulses of certain width then you have to
00:11:37
go to this area here and let me go over
00:11:40
it quickly here in this case we have the
00:11:44
pulses coming in now obviously for this
00:11:48
first one we got up to here but if this
00:11:51
train persists eventually we'll get to a
00:11:54
higher temperature
00:11:56
and this temperature now depends on the
00:11:59
average power as well as the duty cycle
00:12:02
because of course the if the due to
00:12:04
cycle is large then it will go to a
00:12:07
higher value okay so here are the plots
00:12:10
now that allow you to estimate this
00:12:14
temperature for the train of pulses
00:12:17
now the weight is specified here is this
00:12:22
is still the width of the pulse this is
00:12:25
this width here and this curves now are
00:12:28
for the duty cycle okay so for any given
00:12:32
duty cycle you have to look for the
00:12:35
corresponding curve here so say if it is
00:12:38
10 microsecond pulse with the duty cycle
00:12:43
of let's say
00:12:46
0.01 then this will be this value
00:12:48
however
00:12:50
if you have
00:12:52
a current of another profile not a
00:12:55
single pulse and not a train of pulses
00:12:58
which are starting from zero you cannot
00:13:01
get the junction temperature problem
00:13:04
plots that I've shown not the safe
00:13:07
operating area and not the thermal
00:13:10
impedance okay for this you need this
00:13:14
equivalent circuit or this
00:13:16
presentation which will give you for any
00:13:21
profile The Junction temperature okay
00:13:23
you cannot get it from here and you
00:13:27
cannot get it from here this is just the
00:13:30
limit
00:13:31
for a single pulse
00:13:33
and here this is the you get a
00:13:36
temperature for a train of pulses not
00:13:40
any arbitrary current profile so you
00:13:43
need
00:13:44
this equivalent circuit
00:13:46
some vendors will give it as a separate
00:13:49
unit that you can sort of during
00:13:52
simulation find out what is the
00:13:55
instantaneous power and you get this
00:13:57
value or
00:14:00
some vendors and we see more and more
00:14:04
vendors are giving
00:14:07
a model of a transistor that includes
00:14:10
this equivalent circuit so here is an
00:14:15
example this is just one example as I've
00:14:17
said other
00:14:19
manufacturers are providing similar
00:14:21
models and this is for Wall Street the
00:14:25
free and here I am showing a transistor
00:14:29
this is this one this third 25 million
00:14:33
1200 volt voltage
00:14:36
and you just put it in as a regular
00:14:40
model of in this case entity spice and
00:14:44
the difference is that now you're
00:14:46
getting two outputs additional outputs
00:14:50
one is the case temperature in one of
00:14:53
the junction temperature these are
00:14:55
coming
00:14:57
out of here
00:14:59
now in order to examine the linear case
00:15:03
linear operational mode
00:15:06
I've set up this circuit to be a current
00:15:10
source and this is done by having this
00:15:12
resistor here now if I have an input
00:15:15
pulse say of 10 volt
00:15:17
and if this transfer starts to conduct
00:15:21
and then it will lock at the threshold
00:15:26
voltage between gated force and let's
00:15:29
say that the threshold voltage is a say
00:15:31
4 volt then the remaining 10 minus 4 6
00:15:35
voltage on this resistor being one ohm
00:15:39
you'll have a 6 amp passing through the
00:15:43
current while the voltage is of course
00:15:46
the Thousand volt here which is the
00:15:48
voltage source minus the voltage here
00:15:50
which is six volt in this particular
00:15:52
cave you can neglect it or take it into
00:15:54
account okay so this is just a tool to
00:15:58
see how well if this model operates
00:16:02
so in this case I'm feeling a pulses
00:16:06
here of 10 volt to the gate
00:16:10
in the width of the pulse is 10
00:16:13
microseconds and the period is 100
00:16:17
microseconds meaning that we have a duty
00:16:20
cycle of 10 to 100.1 okay
00:16:25
due to cycle does not need to be entered
00:16:28
here I'm just taking it as a reference
00:16:31
for comparison later on
00:16:33
so in this case I've clamped this case
00:16:38
to 25 degree which is like representing
00:16:43
the ambient temperature exactly as I put
00:16:46
it here
00:16:48
this is the terminal of the case I don't
00:16:52
have a thermal resistance between case
00:16:55
and ambient because I'm assuming as in
00:16:58
most data sheets that the case is kept
00:17:03
constant clamped to 25 degrees so this
00:17:07
resistance is zero okay so this goes
00:17:11
directly to this clamping temperature
00:17:15
of course if I would have had a heatsink
00:17:17
and I would have known the thermal
00:17:20
resistance of the heatsink I could have
00:17:22
put here a resistor okay but in this
00:17:25
case I'm just examining parameters
00:17:28
against the data sheet and most of the
00:17:30
parameters in the data chips are given
00:17:33
for a case temperature of 25 degrees or
00:17:36
some other and here is what I'm getting
00:17:38
and you can see we see first of all the
00:17:41
junction temperature this is the output
00:17:43
of the junction obviously I'm starting
00:17:44
from a low temperature 25 doesn't show
00:17:47
exactly what this is 25 and then we go
00:17:50
up
00:17:51
and here I'm already at a steady state
00:17:56
here I'm showing the drain current it's
00:18:01
about six a little bit actually lower
00:18:03
than that would fit in a minute
00:18:06
and then I have here the gate to Source
00:18:09
voltage with threshold voltage and here
00:18:12
I have it in expanded scale in this
00:18:15
region here I see that the temperature
00:18:18
went up to about I don't know 170 we'll
00:18:22
see it later on and then the currents
00:18:25
are about 6 amp and this is the gate to
00:18:28
Source voltage we see here some Peak
00:18:33
here due to probably capacitances and
00:18:36
and then we have this value here we'll
00:18:40
see it later on the exact value okay so
00:18:43
this is a typical run I don't have to
00:18:45
add any information all the thermal
00:18:47
information is within the model of the
00:18:50
transistor okay so here are some
00:18:55
other plots with some more information
00:18:57
for same run again Junction temperature
00:19:01
here it is now I've measured it to be
00:19:04
175 degrees Centigrade
00:19:08
here I'm showing the power
00:19:11
the power is measured by looking at the
00:19:15
current here because this current here
00:19:18
is this current which is coming from the
00:19:22
current search with represent the power
00:19:25
and its steady state of course the
00:19:28
current through the capacitor is zero so
00:19:30
we see here actually the power coming
00:19:33
from here to the output okay so that's
00:19:36
the current through
00:19:38
voltage source V4 and here it is this is
00:19:43
at this point here this region here it's
00:19:46
550. now the current now I've measured
00:19:50
it to be 5.5 amp the current here if
00:19:54
this is the current that passes through
00:19:57
I4 this is this current here so this is
00:20:00
5.5 now we can do a sort of a sanity
00:20:04
check here the voltage is 1000 volt the
00:20:07
current is 5.5
00:20:10
the duty cycle is 0.1 so the average
00:20:14
power is 550 Watts this is what I see
00:20:18
here 551 so everything seems to be
00:20:22
okay in terms of the inner parameters of
00:20:27
this model now I can check it against
00:20:30
this thermaline pins okay check it here
00:20:35
so I found it from the simulation that
00:20:38
the temperature should have gone up here
00:20:40
it is 175.
00:20:43
so we can get this number for this
00:20:46
particular case which is a simple case
00:20:48
because this is a train of pulses
00:20:51
which is compatible with this plot now
00:20:55
we have a 0.1 well I've covered it but
00:21:00
this is a duty cycle of 0.1 we have 10
00:21:03
microseconds so the thermal resistance
00:21:07
or thermal impedance is 0.03
00:21:11
okay or 30 Milli degree Centigrade per
00:21:15
watt now the power we already know is
00:21:19
5.51 kilowatt this is the peak power
00:21:22
which we are using here you are not
00:21:24
using here the average you are using
00:21:26
here the pick
00:21:27
so we have the peak power of 5.5
00:21:30
kilowatt and we multiply this 5.5
00:21:34
kilowatt by 30 this Milli degree
00:21:39
Centigrade per watt 0.03 and I'm getting
00:21:44
165 now you have to add to it 25 degrees
00:21:49
because this is just the junction to
00:21:52
case so I have to add the 25 degree I
00:21:55
will get the 190 in the simulation I got
00:21:59
175 well that's not that bad well could
00:22:04
have been better but it's pretty good
00:22:06
let's have a look now at this safe
00:22:09
operating area plot
00:22:11
and see where we are at as compared to
00:22:15
the model okay so I have a case here
00:22:18
which I marked here of 1000 volt
00:22:24
and I see that I'm getting to the Limit
00:22:28
here
00:22:30
with a current of 18 M okay this is 10
00:22:34
this will be 20. so it's about 18 amp
00:22:39
okay so I now can can simulate this case
00:22:44
with the model
00:22:46
and in this case here is the pulse okay
00:22:50
the single pulse
00:22:52
start from 25 degrees going up to uh
00:22:56
to 106.
00:22:59
this is a 10 microsecond pulse this is
00:23:02
the case that I'm looking at
00:23:04
and this is 106 while the maximum
00:23:09
Junction temperature
00:23:11
specified by the
00:23:14
manufacturer is 150
00:23:17
it's a long way it's a very long way
00:23:20
okay so I don't understand why the limit
00:23:23
here is 106 only
00:23:26
so just to see if maybe this is a wrong
00:23:32
point or whatever I've tried another
00:23:35
point in this case it's 100 volt okay so
00:23:39
I've adjusted the voltage so to be 100
00:23:41
volt and the current to be 50 amp
00:23:46
with a pulse width of 100 microsecond
00:23:51
and here is what I'm getting in this
00:23:53
case this is the temperature of the
00:23:56
pulse this is the current 50 amp and you
00:24:00
see that I'm getting again 109 so
00:24:03
obviously the limit
00:24:07
here is probably 100 volt this is a
00:24:11
little bit too low okay so it's sort of
00:24:13
underestimates it's a big big safety
00:24:15
factor and for a 150 degree device
00:24:20
setting the limit here to 100 it's a
00:24:24
little bit too much so this brings me to
00:24:27
the end of this presentation I hope you
00:24:30
found it of interest and perhaps it will
00:24:32
be useful to you in the future thank you
00:24:35
very much