Circuit design software and components...

[Occularis]

Does anyone else here use NI circuit design suite or Multisim?
I use multisim and Ultiboard, and they're supposedly Pspice compatible as far as the simulation models go, but I cant find any reasonable way of importing models, which is a big annoyance, since that basically means that I cant use, for instance the awesome LT1074CT step-down IC in any simulations, and that sucks
I have downloaded Linear Techs' Pspice database, but it's in a .lib format, and importing databases to multisim doesnt support that file suffix at all . -It only supports .prz-files...

Linear Tech:
" These models are designed for compatibility with LTspice (SwitcherCAD), PSPICE, and PSPICE-compatible simulation programs. They may require a syntax conversion for use with other simulation programs."

Ideas? (Yea, i dont really understand that line where it says "syntax conversion")

[The_Doctor]

The file LT1074.sub is a binary, but its associated symbol file LT1074.asy is plain text as follows:

Version 4
SymbolType CELL
RECTANGLE Normal -128 -144 128 144
TEXT 0 0 Center 0 LT
WINDOW 0 0 -96 Center 0
WINDOW 3 0 80 Center 0
SYMATTR Value LT1074
SYMATTR Value2 LT1074
SYMATTR Prefix X
SYMATTR SpiceModel LT1074.sub
SYMATTR Description 5A Step-Down Switching Regulator
PIN 0 -144 TOP 8
PINATTR PinName Vin
PINATTR SpiceOrder 1
PIN 128 -64 RIGHT 8
PINATTR PinName SW
PINATTR SpiceOrder 2
PIN -128 64 LEFT 8
PINATTR PinName Ilim
PINATTR SpiceOrder 3
PIN 0 144 BOTTOM 8
PINATTR PinName GND
PINATTR SpiceOrder 4
PIN 128 0 RIGHT 8
PINATTR PinName FB
PINATTR SpiceOrder 5
PIN 128 64 RIGHT 8
PINATTR PinName Vc
PINATTR SpiceOrder 6
PIN -128 -64 LEFT 8
PINATTR PinName _SHDN
PINATTR SpiceOrder 7

I've added some external Pspice format parts to LTspice (CA3240, a laser diode, few others), so if a Pspice program can handle that binary file, then going the other way might be possible, only node and subcircuit names might need changing, if anything. If they can't handle the binary, all bets might be off because there's a good chance that LT made it binary to prevent easy adapting or editing.

The downloadable zipped .lib file is a plain text from which you can extract and make separate subcircuit (.sub) files, but the LT1074 isn't in it, so I think they're deliberately keeping that one encrypted, or at least compressed in some way. I don't know enough to try to decode it.

EDIT:
I've seen suggestions that LTspice is deliberately made to lock people into it, but that might not be a bad thing if you're doing analog design, it's apparently faster (not tested other tools myself though), and one thing about people trying to coerce others into their field is they make it easy to get in, if not to get out. As it's also free, the best idea is to get it anyway, but keep your other tools too. That way you'll soon figure out how much interoperability you get. Judging by what I've seen of LTspice and the LT1213 op-amp, it is probably wise to use more than one tool to test a model where possible. Spice is meant to show where a circuit might fail between concept and build, but with the LT1213 I was shown a circuit that failed, yet logically should not have, and the real build worked. The only thing I could have tried other than the build would have been another modelling tool, so serious Spice users probably use several.

[The_Doctor]

Interesting post here:
http://www.electronicspoint.com/deco...s-t114523.html

I'm not sure what legal situation applies if you're decoding one, but it appears it can be done. If the license prohibits use of the LTspice program, that wouldn't matter if the aim is to use another tool. You'd probably be safe if you used it yourself in another tool and did not distibute it, and even if you did maybe you could claim legal exemption under the DMCA 'interoperability' clause if you kept the decoded text file intact, complete with copyright notices, etc.

Edit:
Another tack is to use the LTspice ability to make an encrypted file from a very simple text string to deduce how it encodes them. The output is a text file with hexadecimal ASCII, but if it's the same method of encryption used in their true binary files you can reverse the process. As you'd be reversing your own logical analysis and not examining anything other than text output you're legally entitled to see and use, this would be safe especially if you kept the results to yourself. I doubt there's any enforcible case against plugging small strings of model data into LTspice to see what what encrypted strings result.

[Occularis]

I guess I could custom create the simulation model, using the "Create component wizard" and the info from the datasheets, it's just that it takes half an eternity.
The most annoying thing is the part where it says "Pspice-compatible" without being able to import Pspice-databases .
There shouldnt be any need for decoding.

Edit: Damnit. Cant get proper values working without specific model data reports. Mailed Linear Tech asking for specifics.

[The_Doctor]

That might not work unless you can characterise every transistor in it. I'm not sure if even LT went that far but they were able to try to match model performance with real performance.

If you have a tool that will get data from a data sheet and come up with realistic complex models, you have something amazing. The only (free) tool I found was made by Orcad, it was old, and limited to diodes only. It was enough for my laser diode model (I only need basic electrical performance modelled), but for anything more I'd have liked Intusoft's 'SpiceMod' but while that also was free once in DOS form, it's strictly NOT free now, and is a Windows program, and is very expensive. (And if anyone's got a copy of the old freeware DOS version, please can I have a copy?)

(Tiny rant here: This is what I meant when I told Liteglow he was lucky to have such easy access for machining rings for making laminar flow jets. Some of us really do NOT have ready access to places to do this stuff, or any other thing so easily found in an academic location. Some of us have to get by with our own very limited money, buy all our own tools, we can't access private university data bases, hell, even Steve Roberts commented on this when he was not working in a job with such accesses for a while. Private individuals in the UK can not just expect to find someone willing and able to make them stuff for free in government funded labs and workshops. We're lucky if we can bend our will to finding demos of software that let us do things that would otherwise cost £700 a seat! End rant.)

Actually, while I agree with you that the incompatibility of importing Pspice models to LTspice was annoying, (I know, you're likely going the other way, but I guess it cuts both ways..), I found the only difficulty (of those I could actually handle) was things like node order in the subcircuit declaration line, file names, node counts, the sort of stuff that mainly derives from differences in user conventions. That hurdle was a fast lesson in modelling so I didn't mind that much.

EDIT:
Just seen your own edit... Try asking them for a plain text model of that IC. They might let you have it and ask you not to distribute. I doubt they want to be totally obstructive, they just want to induce people to use their stuff, so this is in their interest.

[andythemechanic]

I have downloaded Linear Techs' Pspice database, but it's in a .lib format, and importing databases to multisim doesnt support that file suffix at all . -It only supports .prz-files...

It is the same with Orcad Pspice. It uses the .cir suffix but the file syntax is the same. I would suggest opening one of your .prz files with an editor and look if this are really spice circuit files or something different. If they are the same just rename the files.
I guess you still have to build the symbol for display in the schematic after importing the spice circuit.
Good Luck!

Andreas

[Occularis]

It's really easy to create components in multisim if you can get the correct component data, it's just time-consuming...

(Link to an example in a tutorial: http://zone.ni.com/devzone/cda/tut/p/id/3173 )

Creating a buck-converter that works in simulations is no biggie, since there's several drop-down menus for each component family in tha component creator, problem is that in the end all you get is a generic component unless you alter the values that makes it the specific piece that you want to simulate. The component footprint can either be copied from a component already in the database with a similar layout, and later alter it to suit your needs, add/subtract pins and so on, or you can make a new one from scratch in either of the programs (multisim/ultiboard).


Generic buck converter data report example: (the info I need from LinearT)

.subckt buck 24 11 99
* 24 11 99
* input duty cycle output
R11 11 0 1MEG
R12 12 0 1MEG
R14 14 0 1MEG
R13 13 0 1MEG
BGLD 0 15 i=v(100)
VSEND 15 16 DC 0
R16 16 0 1
C15 15 0 3.183e-006
B103 103 0 v=v(11)*v(12)
R103 103 0 1MEG
B101 101 0 v=v(22)+v(11)
R101 101 0 1MEG
B 100 0 v=v(13)*v(103)*v(101)
R100 100 0 1MEG
BEL 17 0 v=v(110)
VSENL 17 18 DC 0
L 18 19 5.000e-006
R19 19 0 5.000e-003
B104 104 0 v=v(22)*v(14)
R104 104 0 1MEG
B110 110 0 v=v(103)-v(104)
R110 110 0 1MEG
BFIL 0 10 i=i(VSEND)+i(VSENL)
R10 10 0 1
BEX20 20 0 v=v(12)/v(14)
BE21 21 0 v=v(20)*v(11)
R21 21 22 1G
BFVSENL 0 22 i=i(VSENL)
D1 22 23 DIDEAL
BE23 23 0 v=1-v(11)
D2 0 22 DIDEAL
BX2 300 0 v=v(11)/v(101)
BG24 24 0 i=v(10)*v(300)
BG99 0 99 i=v(10)
BE12 12 0 v=v(24)-v(99)
VG 13 0 DC 1.000e+000
CL 99 90 1.000e-003
RC 90 0 1.000e-005
BE14 14 0 v=v(99)
.MODEL DIDEAL D (n=1M)
.ends

It is the same with Orcad Pspice. It uses the .cir suffix but the file syntax is the same. I would suggest opening one of your .prz files with an editor and look if this are really spice circuit files or something different. If they are the same just rename the files.
I guess you still have to build the symbol for display in the schematic after importing the spice circuit.
Good Luck!

Andreas

-Yea, for importing single component data its' .cir for me as well, databases = .prz