Maker Pro
Maker Pro

Modeling JFET IDSS in SPICE

S

Stephen-I-am

Jan 1, 1970
0
I've been measuring IDSS on a batch of JFETS (On semi MMBFJ310) --
they are specified to run from 24 to 60 mA. (IDSS is the drain current
with the gate grounded and the drain-source voltage in this case at
10V).

Looking at the parameters available for JFETs in SPICE, there doesn't
seem to be an explicit IDSS available. Anyone know how to include it?
Thanks.

Stephen
 
J

Jim Thompson

Jan 1, 1970
0
I've been measuring IDSS on a batch of JFETS (On semi MMBFJ310) --
they are specified to run from 24 to 60 mA. (IDSS is the drain current
with the gate grounded and the drain-source voltage in this case at
10V).

Looking at the parameters available for JFETs in SPICE, there doesn't
seem to be an explicit IDSS available. Anyone know how to include it?
Thanks.

Stephen

JFET's don't specifically have an IDSS equivalent to that of CMOS,
since drain current at Vgs=0 is still quite voltage sensitive
(saturation region)....

Idrain = BETA·(1+LAMBDA·Vds)·(Vgs-VTO)^2

...Jim Thompson
 
H

Helmut Sennewald

Jan 1, 1970
0
Jim Thompson said:
JFET's don't specifically have an IDSS equivalent to that of CMOS,
since drain current at Vgs=0 is still quite voltage sensitive
(saturation region)....

Idrain = BETA·(1+LAMBDA·Vds)·(Vgs-VTO)^2

...Jim Thompson

Hello Stephen,
Idrain = BETA·(1+LAMBDA·Vds)·(Vgs-VTO)^2

You need at least the threshold voltage VTO too.

LAMBDA is used to model the slope of Ids versus Vds
in the saturated region. After you have fitted the
curves with BETA, VTO and LAMBDA, you will discover
that you have to make some compromise with these
parameters to get the best fit for gain(conducatnce),
Idss(Ids at 0V Vgs) and the threshold voltage.

Best regards,
Helmut
 
W

Winfield

Jan 1, 1970
0
Helmut said:
Jim Thompson wrote...



Hello Stephen,


You need at least the threshold voltage VTO too.

LAMBDA is used to model the slope of Ids versus Vds
in the saturated region. After you have fitted the
curves with BETA, VTO and LAMBDA, you will discover
that you have to make some compromise with these
parameters to get the best fit for gain(conducatnce),
Idss(Ids at 0V Vgs) and the threshold voltage.

I haven't found spice to be very useful when working with
typical JFET circuits. That's because critical production
JFET parameters vary over such a wide range that either
a) one is tricked into thinking he's got a good circuit,
thanks to his spot-on spice JFET, or b) the circuit has
been well designed not to be badly affected by the JFET's
wide range of parameters, in which case spice modeling
may not reveal much new information about the circuit.

For example, consider a JFET follower that's biased with
a current source in its source, the drain current can be
easily predicted without resorting to spice.
 
J

Jim Thompson

Jan 1, 1970
0
I haven't found spice to be very useful when working with
typical JFET circuits. That's because critical production
JFET parameters vary over such a wide range that either
a) one is tricked into thinking he's got a good circuit,
thanks to his spot-on spice JFET, or b) the circuit has
been well designed not to be badly affected by the JFET's
wide range of parameters, in which case spice modeling
may not reveal much new information about the circuit.

For example, consider a JFET follower that's biased with
a current source in its source, the drain current can be
easily predicted without resorting to spice.

I used to use rudimentary JFET's as "kick-starters" for low-current
bandgaps. But now that many processes have depletion mode MOSFET's,
they're no longer needed.

...Jim Thompson
 
J

Joerg

Jan 1, 1970
0
Jim said:
I used to use rudimentary JFET's as "kick-starters" for low-current
bandgaps. But now that many processes have depletion mode MOSFET's,


<drool>

Now I wish you could buy them as discretes. With discretes I mean
something in SOT23 or SC75 at under $0.05/1000. That would be the ideal
thing to get fuel cell converters and stuff kick-started. JFETs are ok
but not very steep and a bit wimpy. The BF862 looks nice but NXP,
single-sourced and pricey.
 
S

Stephen-I-am

Jan 1, 1970
0
JFET's don't specifically have an IDSS equivalent to that of CMOS,
since drain current at Vgs=0 is still quite voltage sensitive
(saturation region)....

Idrain = BETA·(1+LAMBDA·Vds)·(Vgs-VTO)^2
Thanks. By assuming lambda = 0 (it's that way in all of the J310
models I've seen) I can just scale beta to change IDSS for at least a
rough order model.

Stephen
 
J

Jim Thompson

Jan 1, 1970
0
Thanks. By assuming lambda = 0 (it's that way in all of the J310
models I've seen) I can just scale beta to change IDSS for at least a
rough order model.

Stephen

Yep. That'll work.

...Jim Thompson
 
W

Winfield

Jan 1, 1970
0
<drool>

Now I wish you could buy them as discretes. With discretes I mean
something in SOT23 or SC75 at under $0.05/1000. That would be the ideal
thing to get fuel cell converters and stuff kick-started. JFETs are ok
but not very steep and a bit wimpy. The BF862 looks nice but NXP,
single-sourced and pricey.

You can buy depletion MOSFETs as discretes. They're single
sourced, but from reliable suppliers, and with alternates
that can do the same simple job if one becomes hard to get.
They're easy on the pocketbook. An obsession with five-cent
parts is going to box you into a pretty tight corner.

As for cheap sot-23 JFETs, the bf862 has a rather large die,
meant for low-noise. Still, its reel price is only 16 cents.
Or use the smaller-die jellybean J108 for 12 to 13 cents in
sot-23 packages.
 
W

Winfield

Jan 1, 1970
0
Winfield said:
As for cheap sot-23 JFETs, the bf862 has a rather large die,
meant for low-noise. Still, its reel price is only 16 cents.
Or use the smaller-die jellybean J108 for 12 to 13 cents in
sot-23 packages.

The J112 is even cheaper - MMBFJ112 for 7.7 cents from Mouser.
Idss delivers 5mA, min. BTW, most of these jellybean J1xx
types are multiply sourced.
 
J

Joerg

Jan 1, 1970
0
Winfield said:
The J112 is even cheaper - MMBFJ112 for 7.7 cents from Mouser.
Idss delivers 5mA, min. BTW, most of these jellybean J1xx
types are multiply sourced.

Those can't really compete with the BF862 from a Vgs versus Id
perspective, take a look at figure 7:

http://www.nxp.com/acrobat_download/datasheets/BF862_3.pdf


I don't mind the occasional >5c part but then there has to be a darn
good reason why I need it. What I do try to avoid is single-source
because of all the grief I've seen with such situation. Some companies
are worse than others and there are some (such as Analog Devices) whom I
trust a whole lot more than others.
 
W

Winfield Hill

Jan 1, 1970
0
Joerg said:
Those can't really compete with the BF862 from a Vgs versus Id
perspective, take a look at figure 7:

http://www.nxp.com/acrobat_download/datasheets/BF862_3.pdf

I don't mind the occasional >5c part but then there has to be a
darn good reason why I need it. What I do try to avoid is single-
source because of all the grief I've seen with such situation.

Picky, picky - you want a large-die, high-current part for
a small-die price and second sources as well. Harrumph! The
large-die J105, J106, and J107, etc., parts compete nicely
with the bf862 for high-current capability. But they must
not be in very much demand, because, sot-23 versions aren't
stocked by the distributors I checked. Perhaps the 16-cent
bf862 isn't so bad, afetr all?

Fairchild's JFTJ105 uses the sot-223 power-tab package, which
shows just how serious those J105 parts are, Idss = 500mA,
min (eat your heart out bf862). Mouser stocks 'em, 50 cents
qty 2k. And worth every penny!

In closing, I'll just point out you can do a lot with 5mA,
and a 3-cent BJT can be used to beef that up if necessary.
 
W

Winfield

Jan 1, 1970
0
Winfield said:
The J112 is even cheaper - MMBFJ112 for 7.7 cents from Mouser.
Idss delivers 5mA, min. BTW, most of these jellybean J1xx
types are multiply sourced.

Just a note to complete the story, Newark has a nice range of
Fairchild's sot-23 mmbfJ1xx JFETs in stock.
 
T

Tim Williams

Jan 1, 1970
0
Winfield Hill said:
In closing, I'll just point out you can do a lot with 5mA,
and a 3-cent BJT can be used to beef that up if necessary.

Whoa, an...er...JGBT?!

Tim
 
W

Winfield

Jan 1, 1970
0
Tim said:
Winfield Hill wrote...


Whoa, an...er...JGBT?!

Yep! By any of several means...

,-----+----------
|--' |
,--->|--, |/ High-current self-biased current source
| '---|
| |\V
| | Rset
| '--/\/\--+--
'----------------------'



power source
---+--------------,
| |
| |/V Automatic startup, with disable
'--/\/\--+---| JFET current sink & PNP switch
| |\
| |
|--' '--/\/\--- startup
,--->|--, power
| | Rset
| +--/\/\--,
'-------| -------+--- gnd
|
disable -->--'

The disable signal can be activated after the desired power
pathway has been established.
 
J

Joerg

Jan 1, 1970
0
Winfield said:
Picky, picky - you want a large-die, high-current part for
a small-die price and second sources as well. Harrumph! The
large-die J105, J106, and J107, etc., parts compete nicely
with the bf862 for high-current capability. But they must
not be in very much demand, because, sot-23 versions aren't
stocked by the distributors I checked. Perhaps the 16-cent
bf862 isn't so bad, afetr all?

It's not just current, it's also how far it'll swing when you only have
500mV total to start with. But the BF862 is single-sourced and NXP, and
there I am bit careful right now. Q2/2007 financials were rather dismal
as I had expected, Q3 supposedly comes out Oct-30. Let's see. If dismal
again I'd think we could see chop-chop time pretty soon. Or in corporate
speak there might come a "right-sizing" of staff and product portfolio.

Fairchild's JFTJ105 uses the sot-223 power-tab package, which
shows just how serious those J105 parts are, Idss = 500mA,
min (eat your heart out bf862). Mouser stocks 'em, 50 cents
qty 2k. And worth every penny!

That's a pretty fat JFET. Doesn't look nearly as steep as the BF862
though but I bet one can buyilt a very nice high-IP3 preamp with that.

In closing, I'll just point out you can do a lot with 5mA,
and a 3-cent BJT can be used to beef that up if necessary.

One trick is to use a start-oscillator that then starts another (bigger)
start oscillator which then starts the real inverter. Reminds me of the
start procedure of the old Dornier-27 we used for parachuting. Only one
guy really mastered that without too much popping and banging, a pilot
with serious bush pilot skills.
 
R

Robert Latest

Jan 1, 1970
0
Joerg said:
It's not just current, it's also how far it'll swing when you only have
500mV total to start with. But the BF862 is single-sourced and NXP, and
there I am bit careful right now. Q2/2007 financials were rather dismal
as I had expected, Q3 supposedly comes out Oct-30. Let's see. If dismal
again I'd think we could see chop-chop time pretty soon. Or in corporate
speak there might come a "right-sizing" of staff and product portfolio.

Yeah, I wonder, too. I know they're upsizing the wafer size in their Hamburg
fab (discretes, I think) from 6" to 8" or something like that. And they're
hiring. They've called me twice, and it's something I think would be
interesting (fab engineering stuff), but I'm wary of a company that's run by
a financial consortium that neither knows nor cares about semiconductors,
engineering, employees or all the other things that make industry fun.
Providing good quality at a good price with qualified and motivated staff
on a long-term basis doesn't seem to be a concept that investors grasp.

robert
 
J

Joerg

Jan 1, 1970
0
Robert said:
Yeah, I wonder, too. I know they're upsizing the wafer size in their Hamburg
fab (discretes, I think) from 6" to 8" or something like that. And they're
hiring. They've called me twice, and it's something I think would be
interesting (fab engineering stuff), but I'm wary of a company that's run by
a financial consortium that neither knows nor cares about semiconductors,
engineering, employees or all the other things that make industry fun.
Providing good quality at a good price with qualified and motivated staff
on a long-term basis doesn't seem to be a concept that investors grasp.

Yep. A very long time ago right out of university I almost started with
them in Hamburg. The former Valvo plant, in the branch that made the NXA
CCD imaging chips. Somehow I had a hunch they might even lose that to
Japan, which they ultimately did. In hindsight I am glad I chose a
smaller US company. I would not have been a very good fit in a larger
corporation as I like efficiency and absolutely despise bureaucratic
hurdles.
 
Top