SD21x spice model?

[Klaus Bahner]

Hi,

Does anyone have spice models for the SD210/SD214 DMOS Fets?
I've found the models from Linear Systems, but they produce clearly
wrong results at least when used in OrCAD PSPICE. There doesn't seem to
be anything wrong with the syntax, hence I tend to believe there is
something wrong with the model parameters. I've contacted Linear Systems
a while ago, but of course no answer ...

Regards,
Klaus

 

[Helmut Sennewald]

Hi,

Does anyone have spice models for the SD210/SD214 DMOS Fets?
I've found the models from Linear Systems, but they produce clearly wrong results at least when used in OrCAD PSPICE.

Hello Klaus,

Every Spice model is an approximation for typical parameters of a device.
What's not correctly simulated?

Best regards,
Helmut

 

[Winfield Hill]

Klaus Bahner wrote...

Does anyone have spice models for the SD210/SD214 DMOS Fets?
I've found the models from Linear Systems, but they produce clearly
wrong results at least when used in OrCAD PSPICE. There doesn't seem
to be anything wrong with the syntax, hence I tend to believe there
is something wrong with the model parameters. I've contacted Linear
Systems a while ago, but of course no answer ...

You should be checking with Vishay (was Siliconix).

My Intusoft spice has models for sd210 through sd215.
Without looking at them all, I see the capacitance varies
between types, which is not supported by the datasheets.

..MODEL SD210DE NMOS Level=1 CBD=4.65P CBS=5.58P CGBO=20.6N
+ CGDO=2N CGSO=2.4N GAMMA=.992 IS=3.25E-13 KP=7.15M
+ LAMBDA=22.5M MJ=.46 PB=.8 PHI=.75 RD=12 RS=15 VTO=1.15

..MODEL SD215DE NMOS Level=1 CBD=8.08P CBS=9.7P CGBO=20.6N
+ CGDO=2N CGSO=2.4N GAMMA=.992 IS=3.25E-13 KP=7.2M
+ LAMBDA=22.5M MJ=.46 PB=.8 PHI=.75 RD=11 RS=13 VTO=1.1

I have not evaluated these models. In my experience,
if you want to use a MOSFET model for linear work,
you have to first vet it against bench measurements,
and be prepared to make substantial changes.

 

[Klaus Bahner]

Hello Klaus,

Every Spice model is an approximation for typical parameters of a device.
What's not correctly simulated?

Well, it's not just an approximation thing. Even applying 20V as
gate-source voltage, doesn't switch the Fet completely on. In this case
the model just creates a drain current of about 600uA (@Vds ~13V)
corresponding to an on resistance in the order of 20k, although it
should be only 45 Ohms max.

Wouldn't call this just the usual difference between modelled and real
circuit


Cheers,
Klaus

 

[Jim Thompson]

Well, it's not just an approximation thing. Even applying 20V as
gate-source voltage, doesn't switch the Fet completely on. In this case
the model just creates a drain current of about 600uA (@Vds ~13V)
corresponding to an on resistance in the order of 20k, although it
should be only 45 Ohms max.

Wouldn't call this just the usual difference between modelled and real
circuit


Cheers,
Klaus

From where can I download the models?

...Jim Thompson

 

[Klaus Bahner]

From where can I download the models?

...Jim Thompson

http://www.linearsystems.com/download/alldmosspice.zip

To be specific, from the zip archive I've used the file LIS_PSPICE.LIB.

Klaus

 

[Jim Thompson]

Well, it's not just an approximation thing. Even applying 20V as
gate-source voltage, doesn't switch the Fet completely on. In this case
the model just creates a drain current of about 600uA (@Vds ~13V)
corresponding to an on resistance in the order of 20k, although it
should be only 45 Ohms max.

Wouldn't call this just the usual difference between modelled and real
circuit


Cheers,
Klaus

I duplicated your result.

L=0.5u W=889u is NOT a large device.

What/where are the specs/data-sheet for the real device?

I note that the library specifies VGSmax = +2.5V

...Jim Thompson

 

[Helmut Sennewald]

Well, it's not just an approximation thing. Even applying 20V as gate-source voltage, doesn't switch the Fet completely on. In
this case the model just creates a drain current of about 600uA (@Vds ~13V) corresponding to an on resistance in the order of 20k,
although it should be only 45 Ohms max.

Wouldn't call this just the usual difference between modelled and real circuit


Cheers,
Klaus

Hello Klaus,

You have to specify the size W and L depending on what FET you want.
Then you will get for the SD210DE about Id=11.9mA @ Vgs=2.5V, Vds=5V.
The Rds_on is only 10 Ohm with Vgs=10V, but the model is
only specified for Vgs up to 2.5V.

Best regards,
Helmut



***** MODEL NAMES: *****
***** SD210DE: DMOS device model W=889um L=0.5um T=27C. *****
***** BIAS RANGE: VDSmax=20V VGSmax=2.5V VBSmax=-2V *****
***** *****
***** SD214DE: DMOS device model W=889um L=0.5um T=27C. *****
***** BIAS RANGE: VDSmax=12V VGSmax=2.5V VBSmax=-2V *****
***** *****
***** SD5400: DMOS device model W=889um L=0.5um T=27C. *****
***** BIAS RANGE: VDSmax=20V VGSmax=3.5V VBSmax=-2V *****
***** *****
***** SD5400CY: DMOS device model W=889um L=0.5um T=27C. *****
***** BIAS RANGE: VDSmax=12V VGSmax=3.5V VBSmax=-2V *****
***** *****
***** SST211: DMOS device model W=889um L=0.5um T=27C. *****
***** BIAS RANGE: VDSmax=30V VGSmax=2.5V VBSmax=-2V *****
***** *****
***** SST213: DMOS device model W=889um L=0.5um T=27C. *****
***** BIAS RANGE: VDSmax=10V VGSmax=3.5V VBSmax=-2V *****
***** *****
***** SST215: DMOS device model W=889um L=0.5um T=27C. *****
***** BIAS RANGE: VDSmax=20V VGSmax=3.5V VBSmax=-2V *****
***** *****

 

[Jim Thompson]

I duplicated your result.

L=0.5u W=889u is NOT a large device.

What/where are the specs/data-sheet for the real device?

I note that the library specifies VGSmax = +2.5V

...Jim Thompson

OK, I found the problem. The bare model is not SIZED.

You need to create a part such that its template is...

MN1 %d %g %s %b SD210 L=0.5u W=889u <<<<

I then get ID = 12mA at VGS = +2.5V, VDS = +2V

And the slope near the origin is 52.6 ohms.

...Jim Thompson

 

[Jim Thompson]

Klaus Bahner wrote...

You should be checking with Vishay (was Siliconix).

My Intusoft spice has models for sd210 through sd215.
Without looking at them all, I see the capacitance varies
between types, which is not supported by the datasheets.

.MODEL SD210DE NMOS Level=1 CBD=4.65P CBS=5.58P CGBO=20.6N
+ CGDO=2N CGSO=2.4N GAMMA=.992 IS=3.25E-13 KP=7.15M
+ LAMBDA=22.5M MJ=.46 PB=.8 PHI=.75 RD=12 RS=15 VTO=1.15

.MODEL SD215DE NMOS Level=1 CBD=8.08P CBS=9.7P CGBO=20.6N
+ CGDO=2N CGSO=2.4N GAMMA=.992 IS=3.25E-13 KP=7.2M
+ LAMBDA=22.5M MJ=.46 PB=.8 PHI=.75 RD=11 RS=13 VTO=1.1

I have not evaluated these models. In my experience,
if you want to use a MOSFET model for linear work,
you have to first vet it against bench measurements,
and be prepared to make substantial changes.

Those models are Level = 1, which suck the big lemon.

Fortunately Klaus cites models that use PSpice Level = 7, same as
HSpice Level = 49, and don't require the "Win-fix" ;-)

...Jim Thompson

 

[Helmut Sennewald]

Those models are Level = 1, which suck the big lemon.

Fortunately Klaus cites models that use PSpice Level = 7, same as
HSpice Level = 49, and don't require the "Win-fix" ;-)

Hello Jim,

The model given by Win has given a "correct" Rds_on of about
40 Ohms for Vgs=10V. So this model is much better suited for
switched applications than the model from Linear Systems.

Best regards,
Helmut

 

[Klaus Bahner]

Thanks Jim and Helmut,

I will try that. (Hope I can figure out how to make the necessary
changes in OrCAD's model editor ...)

Klaus

 

[Jim Thompson]

Hello Jim,

The model given by Win has given a "correct" Rds_on of about
40 Ohms for Vgs=10V. So this model is much better suited for
switched applications than the model from Linear Systems.

Best regards,
Helmut

I haven't seen the data sheet, but the library says VGSmax = +2.5V,
NOT +10V. I simulated at VGS = +2.5V, thus my slightly higher
predicted slope.

As for a Level = 1 model being "better suited" than a Level = 7,
ROTFLMAO ;-) Even Win admits the Level = 1 model needs a patch.

Saying a Level = 1 model is "better suited", is like saying a Model-T
can outrun my Q45 ;-)

...Jim Thompson

 

[Klaus Bahner]

You should be checking with Vishay (was Siliconix).

I did this already, but couldn't find any models on Vishay/Siliconix' site.
By the way, it seems as if Linear Systems has acquired the DMOS line
from Vishay/Siliconix. Although Vishay still lists these devices on
their homepage, they seem not to make them any longer (Haven't yet got
an authorative answer from my distributor, though). Then there is a
company called Calogic, which also offers them, but it could be that
these devices will be harder to get.

Are there any alternatives out there, i.e. other small signal, high
speed, low leakage, lateral MOSFETs without gate protection diodes?

Klaus

 

[Klaus Bahner]

I haven't seen the data sheet, but the library says VGSmax = +2.5V,

The datasheet says VGSmax +/- 40V! Hence, in general the 2.5V of the
model seem to be a limitation, although in my case it's fine, since I
anyway want to use the smallest possible gate voltage swing to minimize
charge injection.

Klaus

 

[Joerg]

Hello Jim,

As for a Level = 1 model being "better suited" than a Level = 7,
ROTFLMAO ;-) Even Win admits the Level = 1 model needs a patch.

That's usually when I trudge over to the lab and fire one up. In analog
apps that include fast switching the "not so known" effects of
surrounding pulse transformers and stuff like that are hard to simulate.

Saying a Level = 1 model is "better suited", is like saying a Model-T
can outrun my Q45 ;-)

Put that 1100HP engine in there plus some Formula-1 rear tires and it
might just do that ;-)

 

[Joerg]

Hello Klaus,

Are there any alternatives out there, i.e. other small signal, high
speed, low leakage, lateral MOSFETs without gate protection diodes?

Don't know what you want to do with them but check out TV tuner FETs
such as the BF998. Many of the do have diodes but the capacitances and low.

SD stuff has become harder to buy and some of it like my favorite SD5400
arrays have become expensive so I am not using them anymore.

 

[Jim Thompson]

Hello Jim,


That's usually when I trudge over to the lab and fire one up. In analog
apps that include fast switching the "not so known" effects of
surrounding pulse transformers and stuff like that are hard to simulate.



Put that 1100HP engine in there plus some Formula-1 rear tires and it
might just do that ;-)

The machine shop operator was very cordial and bought the engine back.
Duane had been one of his best customers and had paid for the engine
even though he knew he was near death.

...Jim Thompson

 

[Joerg]

Hello Jim,

The machine shop operator was very cordial and bought the engine back.
Duane had been one of his best customers and had paid for the engine
even though he knew he was near death.

That was very nice of him. If they don't part out the engine maybe they
could let you know where it went so you could see it in action.