LTspice is a difficult tool to make useful

Kind of strange thing for an inductor company to do.

There are tons of LT Spice manuals and tutorials online. The built-in HELP isn't very good.

Yeah...its a disaster waiting to happen.

Many model makers just don't think about the convergence properties of the models. Spice has a pretty good function capability that lets one write continuous models to do pretty much everything.

Using an if() else() construct in an analog model is bizarre to say the least.

-- Kevin Aylward

"Charlie Brown frequently tries to tell Lucy that her crazy theories are false, and when he finally succeeds, Lucy would make an insensitive remark about the way he looks. Charlie Brown's stomach hurts when Lucy tries to teach her theories to Linus.

Charlie Brown often visits Lucy's psychiatric booth for help, but always gets useless advice (such as "Snap out of it." or "The insecurities people have can lead to colds and other illnesses")."

They are not an inductor company. I think they make now almost everything. (switches, many capacitor types, resitor, contacts, DCDC modules)

There is something you can't understand. Wurth makes many things in local language. .-) There FAEs are a little bit different compared to other company. And I thing there boss/owner is very traditional.

Olaf

Here's another example of how difficult it can be to use LTspice. One of m y problems is pasting in third party models. Most recently I had a sub cir cuit of a transistor and didn't want to redraw the durn symbol. So I cntl- right clicked one on my schematic and opened the symbol. I edited all the fields appropriately to link it to the sub circuit and saved it. Then I ad ded it to my schematic and it worked fine. But I noticed it had a U number instead of a Q. Repeating the process I saw the InstName change when I ed ited the model file field. Worse, there is no way in the symbol editor to edit the InstName attribute!!!

This is the sort of insane behavior that makes it so hard to get work done instead of fighting LTspice.

Is there really no way to link a symbol to a model file (using an intrinsic model) other than a .lib/.inc statement on the drawing? This clutters up the drawing and makes it more difficult to swap out devices to see how they perform. You not only have to change the part, you have to change the lib rary model link. Why would the link to the model file not be in the symbol like a sub circuit model?

Well.... In SuperSpice you don't really have to do anything for common components to attach an existing symbol to a model/subckt.

Drag drop the model file to the main window or paste it into an existing model file in the model browser.

Select one of the built in symbols and place it on the schematic. There are two versions of transistor symbols, and lots of different transistor symbols. One for connecting to .models, the other for connecting to .subckt.

Place the relevant symbol, double click on it and use the dialog box to navigate to the model that you want to attach it to.

Done.

-- Kevin Aylward

all

en

ol

ne

sic

the

re

kt.

Sounds very interesting especially how you simulate digital. Are there any convergence issues with digital and analog mixed? I have the impression t hat abrupt changes in a signal are what cause problems in LTspice. In the present design I have some analog which holds the digital in reset. Once t he digital reset is released the simulation rate slows from seconds per sec ond to ms or even us per second. Is there much of an existing digital libr ary or would I need to build my digital chips... again?

Do third party models play well on this simulator? I have some comparator models and even some intrinsic transistors that seem to cause the digital s imulation to bog.

Right now LTspice has been a real time sink with all the convergence issues and model issues. I'd say fully half the models won't work properly, eith er not converging or running very slow in my design. Maybe it's my digital stuff causing the problem and any given transistor model is just the prove rbial straw?

I might give it a try. Any support forums?

Well... yes.... there is real mixed-mode engine in the XSpice bit. There however...is that I haven't spent much effort in giving it support, so the models are a bit thin.

There are some examples that shows how it all works.

StateMachine.sss

uses the feature shown here:

and a mixed-mode one here:

PLLBehaviouralDigital.sss

If they have if() else() in them, no. If it has the usual stuff, mostly.

You muight be better replacing the comparator model with one based on tanh().

This will be down to $hit model design LTSpice itself has very good convergence properties, usually...

If you set stuff up in SS, I might be able to figure out what the bad bits of the model are.

Only me.

There are over 100 examples...

-- Kevin Aylward

Am 28.05.2020 um 23:51 schrieb Ricketty C: > Right now LTspice has been a real time sink with all the convergence issues and model issues. I'd say fully half the models won't work properly, either not converging or running >very slow in my design.

Hello Rick, Have you ever thought while 90% of the engineers use LTspice and most of them are very happy with it? It's because they come with the right attitude and learned how to efficiently used it. You always look for problems due to lack of knowing the features of LTspice. When you look to other SPICE-simulators, you will quickly discover that the grass there is not as green as you hoped.

When you have a third party transistor with a subcircuit, simply place a npn or pnp. Then CTRL-right-mouse-click on it and change the Prefix-attribute from QN or QP to X. Additionally change the value of its Value-attribute to the name of the subcircuit. Finally place a SPICE-directive into the schematic to tell LTspice the name of the model file. .lib name_of_modelfile

When use this procedure, your design can be run on every PC without any additional effort. Nothing is hidden. Everybody who knows SPICE will understand what's simulated with this schematic circuit.

Helmut

I just plunk the .asy and .cir or .mod files in the local folder and LT finds them somehow. No need to mess with libraries.

Weird things happen if some are not there!

Does this work?

If I had his collection of sick race and sex problems, I sure wouldn't continually post them in public. Especially in an electronic design forum.

Am 30.05.2020 um 16:45 schrieb snipped-for-privacy@highlandsniptechnology.com:

Hello John,

The example R_time.asc works as intended. R=time

The example with the OPA547 works too. You could directly make the value OPA547 visible in the schematic. Either enable it in the schematic with right-mouse-click and mark visible or you set its visibilty already in the symbol with the symbol editor.

Helmut

I don't have to "look" for problems. The problems are there if I look or n ot.

That may be... or it may not be. I'm going to give SS a try before I do an other design in LTspice.

Why do you say the .lib statement is needed? When a vendor gives me an int rinsic model I've found that I can make it a sub circuit and not have to cl utter up the schematic with .lib statements. Are you saying I don't need t o create a unique symbol for this component if I add the .lib statement? T his would be the first time I recall anyone saying I did not need to create a unique symbol. Maybe I missed that. Once it is used in a design it can be copied. If I don't need to create a symbol, that means I need to do th is same editing to the symbol every time it is used in a new design, right?

Ok, that is yet another way to remember for creating new components.

Good luck.

Of course John Larkin does continue to advertise the fact that he's the kind of gullible twit who falls for climate change denial propaganda.

I would 100%, any day of the week hand a collection of JL's and John Doe's posts along with mine to a representative collection of "normal Americans" and let them vote on who the "drug-crazed" one is. Hell you can send 'em to my mother.

Is she an electrical engineer?

Are you?

Thank heavens no.

I would have preferred to be a classical guitarist but compared to the music industry the engineering field is a far kinder place