pspice vs ngspice

[David Logan]

Hello. I asked this question in an electronics mailing list that I am
subscribed to, but I guess the moderator went on vacation or something.
It was never sent out. So I will post my question here. It's a good
forum anyway, since it deals with some sort of pspice issue (either an
error on my part or an error in the software, I don't know which.)

So, here is the question:

Hey all! I am struggling my way through my introductory book. I have to
spend 30 minutes on every one out of three problems figuring out where I
went wrong But enough whining about it. I have a question about
pspice for Windows (student version.)

On this one in particular, I was having trouble getting the right
answer, so I poked it into Linux/ngspice, and got an answer different
than me, and different than the book. So I poked it into pspice for
Windows (student version) and got different answers still.

This is where I learned about current sources, arrows and +/- signs with
regards to ngspice and pspice. After I fixed my circuit in ngspice, it
gives me answers that correspond with the book.

I just worked the problem by hand again, and got answers that also
correspond with the book.

Thus, my multi-question mail has now dropped to one or two questions:

1) Is pspice wrong? It must be, I think. But I'd like somebody to verify.
2) If so, did I do something wrong? Or is there an error in their
simulation code?

Since it's a pain in the butt to post all of the details in an email, I
have set up a web location to review:

http://www.loganshouse.net/problem

original_problem.jpg
- The page with the original problem. It's problem 3.15
pspice.jpg
- screen shot of the schematic, and the voltages/currents
ngspice.jpg
- screen shot of the schematic in ngspice

The ngspice output files

 

[Helmut Sennewald]

Hello. I asked this question in an electronics mailing list that I am
subscribed to, but I guess the moderator went on vacation or something.
It was never sent out. So I will post my question here. It's a good
forum anyway, since it deals with some sort of pspice issue (either an
error on my part or an error in the software, I don't know which.)

So, here is the question:

Hey all! I am struggling my way through my introductory book. I have to
spend 30 minutes on every one out of three problems figuring out where I
went wrong But enough whining about it. I have a question about
pspice for Windows (student version.)

On this one in particular, I was having trouble getting the right
answer, so I poked it into Linux/ngspice, and got an answer different
than me, and different than the book. So I poked it into pspice for
Windows (student version) and got different answers still.

Hello David,
you made an obvious mistake with the values of your resistors.
A value has to be a number and not a mathematical expression.
You tried with values like 1/3, 1/5 and 1/6.
This can't work in SPICE. PSICE has treated all your resistors
as 1 Ohm resistors.
Why have you given NGSPICE a chance with with 0.333333, 0.2 and 0.166666?
That's not fair. Give NGSPICE the same bad try with 1/3, 1/5 and 1/6.
I assume you will get an error message.

This is where I learned about current sources, arrows and +/- signs with
regards to ngspice and pspice. After I fixed my circuit in ngspice, it
gives me answers that correspond with the book.

Be aware that the sign of the reported current through resistors
may depend on the orientation. Resistors have also a "polarity"
in some SPICE programs.

I just worked the problem by hand again, and got answers that also
correspond with the book.

Thus, my multi-question mail has now dropped to one or two questions:

1) Is pspice wrong?
No!

It must be, I think. But I'd like somebody to verify.

See my explanantion above.

2) If so, did I do something wrong? Or is there an error in their
simulation code?

Enter numbers like 0.333333, 0.2 and 0.166667 for the resistor values.

Since it's a pain in the butt to post all of the details in an email, I
have set up a web location to review:

Thanks for the good explanation and the screenshots. They really
helped to understand what you have tried to simulate.


You should try LTspice. It is similar in features to PSPICE, but it's
totally free and unlimited regarding circuit size.

LTSPICE can be downloaded from here:
www.linear.com SwitcherCAD = LTspice

The user group for LTspice:
http://groups.yahoo.com/group/LTspice/


Best Regards,
Helmut

http://www.loganshouse.net/problem

original_problem.jpg
- The page with the original problem. It's problem 3.15
pspice.jpg
- screen shot of the schematic, and the voltages/currents
ngspice.jpg
- screen shot of the schematic in ngspice

The ngspice output files

------------ LTSPICE schematic -------------

Schematic for LTspice
Copy the following text into a file named op_3_15.asc .


Version 4
SHEET 1 880 680
WIRE 80 96 80 -32
WIRE 80 160 80 96
WIRE 80 304 80 240
WIRE 144 96 80 96
WIRE 256 -32 80 -32
WIRE 288 96 224 96
WIRE 288 160 288 96
WIRE 288 304 80 304
WIRE 288 304 288 240
WIRE 288 336 288 304
WIRE 352 96 288 96
WIRE 496 -32 336 -32
WIRE 496 96 432 96
WIRE 496 96 496 -32
WIRE 496 160 496 96
WIRE 496 304 288 304
WIRE 496 304 496 240
FLAG 288 336 0
SYMBOL res 64 144 R0
SYMATTR InstName R1
SYMATTR Value 0.166666667
SYMBOL res 272 144 R0
SYMATTR InstName R2
SYMATTR Value 0.2
SYMBOL current 496 240 R180
WINDOW 0 -52 77 Left 0
WINDOW 3 -47 6 Left 0
SYMATTR InstName I1
SYMATTR Value 4
SYMBOL res 336 112 R270
WINDOW 0 32 56 VTop 0
WINDOW 3 0 56 VBottom 0
SYMATTR InstName R3
SYMATTR Value 0.3333333333
SYMBOL voltage 128 96 R270
WINDOW 0 32 56 VTop 0
WINDOW 3 -32 56 VBottom 0
SYMATTR InstName V1
SYMATTR Value 10
SYMBOL current 256 -32 R270
WINDOW 0 32 40 VTop 0
WINDOW 3 -32 40 VBottom 0
SYMATTR InstName I2
SYMATTR Value 2
TEXT 72 -112 Left 0 !.OP

 

[David Logan]

Thanks for the info! The reason why I gave ngspice a chance and not
pspice is quite simple. ngspice said "um, I don't know what these values
are. Type something intelligent in." pspice was happy to take what I
entered. Thus, I assumed pspice knew how to handle them.

I'll keep that in mind for the future. Thanks!

David Logan

 

[Jim Thompson]

Thanks for the info! The reason why I gave ngspice a chance and not
pspice is quite simple. ngspice said "um, I don't know what these values
are. Type something intelligent in." pspice was happy to take what I
entered. Thus, I assumed pspice knew how to handle them.

I'll keep that in mind for the future. Thanks!

David Logan

[snip]

PSpice treats "/" as additional or appended text, UNLESS written as
{1/5}, then computation is done

...Jim Thompson

 

[Helmut Sennewald]

Thanks for the info! The reason why I gave ngspice a chance and not
pspice is quite simple. ngspice said "um, I don't know what these values
are. Type something intelligent in." pspice was happy to take what I
entered. Thus, I assumed pspice knew how to handle them.

I'll keep that in mind for the future. Thanks!

David Logan

[snip]

PSpice treats "/" as additional or appended text, UNLESS written as
{1/5}, then computation is done

...Jim Thompson

Hello Jim,
thanks for this tip. It works in LTspice as well.
Expressions in curly braces {} are correctly evaluated in PSPICE and
LTspice.

Best Regards,
Helmut

 

[Stuart Brorson]

Hi Helmut --

[ . . . . snip!. . . .]
: You should try LTspice. It is similar in features to PSPICE, but it's
: totally free and unlimited regarding circuit size.

: LTSPICE can be downloaded from here:
: www.linear.com SwitcherCAD = LTspice

: The user group for LTspice:
: http://groups.yahoo.com/group/LTspice/

I use and love LTSpice on my Windoze box at work, so I have no
argument with your LTSpice advertisement. Howver, it does merit
pointing out that ngspice is *also* totally free for download, and
unlimited w.r.t. circuit size. Moreover, in contrast to LTSpice,
ngspice is open-source, meaning that students can see the code, play
with it, and modify it if they want.

Finally, the OP apparently is running gEDA on Linux (look at his
screen shots), and generating SPICE netlists with it for ngspice.
That may explain why he is running ngspice, since it is Linux native.
Yes, LTSpice does run on Linux under wine, but it's not a Linux
native program.

It's great to see the gEDA stuff being used by students!

Stuart

 

[David Logan]

That's absolutely right. I'm running Linux, which is why I'm using geda
(which doesn't work so well), gschem, gnetlist and ngspice.

I am also running pspice for windows, student version, because:
1) Primary reason--my book uses pspice, and shows me how to use
electronic modelling software via pspice.

2) The linux software is seriously lacking in documentation, so I figure
as I learn how to use pspice, I will simultaneously learn how to use the
Linux tools.

3) To compare pspice and ngspice for accuracy, to make sure that I get
the same results from both. Since the book will have pspice examples and
answers, this is possible.

But my ultimate goal is be an electronics expert using Linux tools.

David Logan

 

[Stuart Brorson]

: That's absolutely right. I'm running Linux, which is why I'm using geda
: (which doesn't work so well), gschem, gnetlist and ngspice.

It's good to see that you are using gEDA, bad to hear that it's not
working so well.

Since I'm one of the folks working on it, perhaps you could help us
out a little bit. When you say "[it] doesn't work so well", please
list two or three specific things which you think don't work well.
Specific bug reports provide us with things to work on; therefore I'd
be interested in hearing about the top two or three complaints you
have about gEDA.

[. . . . snip . . . .]

: 2) The linux software is seriously lacking in documentation, so I figure
: as I learn how to use pspice, I will simultaneously learn how to use the
: Linux tools.

There is some truth to this, but have you looked in your
${GEDA_INSTALL_DIR}/share/doc/geda-doc directory? There's lots of
stuff there. Also, for gEDA/SPICE specific info you can look here:

http://www.brorson.com/gEDA/SPICE/t1.html

For ngspice commands, many people use this reference:

http://newton.ex.ac.uk/teaching/CDHW/Electronics2/userguide/

Finally, posting questions on the geda-user mailing list is a good
way to get help if you have problems with the software. The mailing
lists are here:

http://www.geda.seul.org/mailinglist/index.html

: But my ultimate goal is be an electronics expert using Linux tools.

Welcome abord!

Stuart

 

[David Logan]

Thanks for all of that. I will review the links you gave. In all
fairness, I suspect part of the problem I consider the documentation
insufficient is because I don't yet know how to use electronics
modelling software in general yet, and Linux software package
documentation is usually lacking, especially for complicated software.

I have a question with before I give my replies. The question is: When
you say you are interested in hearing about gEDA, does that mean just
gEDA, or does also mean gschem, gnetlist and ngspice?

gEDA itself is crashing on me when I try to do certain things, so I quit
using gEDA, and was just using the individual tools. If you are
interested in specifics (regardless of the answer to the above
question), show me the forum to post the bugs and the forum to post the
whines, and I will be happy to do so.

By the way, I have been a professional programmer for about 15 years,
just so you know a bit of my background.

David Logan

 

[Stuart Brorson]

: Thanks for all of that. I will review the links you gave. In all
: fairness, I suspect part of the problem I consider the documentation
: insufficient is because I don't yet know how to use electronics
: modelling software in general yet, and Linux software package
: documentation is usually lacking, especially for complicated software.

: I have a question with before I give my replies. The question is: When
: you say you are interested in hearing about gEDA, does that mean just
: gEDA, or does also mean gschem, gnetlist and ngspice?

Aaah, I see part of the problem already.

gEDA = The gEDA project. This includes gschem, gnetlist, ngspice, and
so on.

geda = The gEDA project's project manager. An unfortunate choice for
it's name, but we're saddled with it now. A better name might have
been "gmanager" or something like that. Oh well.

: gEDA itself is crashing on me when I try to do certain things, so I quit
: using gEDA, and was just using the individual tools.

You mean the project manager "geda" crashed on you, right? I don't
use the project manager at all & instead use the individual tools.
Geda (the project manager) is old, and hasn't been updated recently.
It might have some bugs.

The core tools "gschem", "gnetlist", and "gattrib" are under constant
developement, as are several of the other smaller utilities. They
tend to be bug-free, and very functional (except for gattrib, which
awaits some features related to net and pin attributes).

: If you are
: interested in specifics (regardless of the answer to the above
: question), show me the forum to post the bugs and the forum to post the
: whines, and I will be happy to do so.

Mailing lists:

http://www.geda.seul.org/mailinglist/index.html

Bug reports:

http://www.geda.seul.org/developer.html

(click through to "jitterbug")

: By the way, I have been a professional programmer for about 15 years,
: just so you know a bit of my background.

Cool! Any interest in hacking gEDA?

Stuart

 

[David Logan]

: Thanks for all of that. I will review the links you gave. In all
: fairness, I suspect part of the problem I consider the documentation
: insufficient is because I don't yet know how to use electronics
: modelling software in general yet, and Linux software package
: documentation is usually lacking, especially for complicated software.

: I have a question with before I give my replies. The question is: When
: you say you are interested in hearing about gEDA, does that mean just
: gEDA, or does also mean gschem, gnetlist and ngspice?

Aaah, I see part of the problem already.

gEDA = The gEDA project. This includes gschem, gnetlist, ngspice, and
so on.

geda = The gEDA project's project manager. An unfortunate choice for
it's name, but we're saddled with it now. A better name might have
been "gmanager" or something like that. Oh well.

: gEDA itself is crashing on me when I try to do certain things, so I quit
: using gEDA, and was just using the individual tools.

You mean the project manager "geda" crashed on you, right? I don't

Yes, the project manager crashed. It is unfortunate that a cohesive
environment doesn't exist like in pspice. One written with Qt would
probably be good.

use the project manager at all & instead use the individual tools.
Geda (the project manager) is old, and hasn't been updated recently.
It might have some bugs.

The core tools "gschem", "gnetlist", and "gattrib" are under constant
developement, as are several of the other smaller utilities. They
tend to be bug-free, and very functional (except for gattrib, which
awaits some features related to net and pin attributes).

: If you are
: interested in specifics (regardless of the answer to the above
: question), show me the forum to post the bugs and the forum to post the
: whines, and I will be happy to do so.

Mailing lists:

http://www.geda.seul.org/mailinglist/index.html

Bug reports:

http://www.geda.seul.org/developer.html

(click through to "jitterbug")

: By the way, I have been a professional programmer for about 15 years,
: just so you know a bit of my background.

Cool! Any interest in hacking gEDA?

I always have an interest. It's the time I don't have

Stuart

David Logan