LTspice heads up

LTspice heads up...

Latest LTspice sync release over-wrote many text-based subcircuits with encrypted versions.

So back up all the text-based ones before they disappear. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson
Loading thread data ...

Hasn't that always been the case? I recall that you should always back up LTspice device models.

--

Rick
Reply to
rickman

I save all of my older LTspice download EXEs so that I can always go back. Sometimes there is a bug that you can't deal with so that's another good reason to keep old installations around.... Then can check the operation of a newer vs. older LTspice as well for speed and/or accuracy or whatever.

boB K7IQ

Reply to
boB

Does Ver4 Pspice run encrypted subcircuits?

Will LTspice begin to handle encrypted models that currently demand Pspice Rev9 or higher?

RL

Reply to
legg

The problem is that every simulator has its own version of encryption... LTspice encryptions won't run on PSpice and vice versa.

My defense (offense ?:-) is to write models that are simulator-independent. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

I should have added... the in-house Spice-modelers are usually PhD's... aka village idiots, when it comes to modeling... just look at the trash that comes out of Microchip.

So I will amuse myself by writing models from their datasheets that outperform their models (and they have the device-level netlists).

Bunch-a-dummies. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

Can't see how that would make much sense. Linear Tech sells ICs, not LTpice simulator software.

I'd assumed that previous unencrypted LTspice models worked on Pspice...

RL

Reply to
legg

It wouldn't surprise me if the models they give to customers are written by people in sales/marketing who have no access to the device level netlists, just like the datasheets are often written by people who had nothing to do with the circuit design. I was sometimes surprised and quite amused by the block diagrams that appeared in the datasheets of chips that I designed circuitry for.

The spice modellers who work on the models USED in-house were extremely competent, but those models were totally separate from anything a customer would ever get to see. After all if the models used in design verification were wrong, that would result in new mask sets and massive costs and delays, whereas spice models for customers are just a tick-box exercise.

It is sad that chip companies have detailed and accurate models of their parts for design validation, but then assume that customers are happy with any old behavioural model that misses out most of the subtleties of the behaviour. Any other chip company can easily get complete reverse-engineered netlists of competitor parts, so the secrecy is pointless.

Chris

Reply to
Chris Jones

They are paranoid about other manufacturers learning the innards of their parts via the spice models published?

I think there is a clause in the license about LTSpice not being legal to use by their competitors or for chip design.

--

John Devereux
Reply to
John Devereux

I think Mike is trying to hide the sleights-of-hand he uses to speed up simulations.

Also ponder the following statement from _SCAD4_Manual_, particularly consider why there are two netlist "parsers":

"Also interesting is which solver is used. LTspice contains two complete versions of SPICE. One is called the normal solver and the other is called the alternate solver. The alternate solver uses a different sparse matrix package with reduced roundoff error. Typically the alternate solver will simulate at half the speed of the normal solver but with one thousand times more internal accuracy. This can be a useful diagnostic to have available. There is no .option to specify which solver is used, the choice must be made before the netlist is parsed because the two solvers use different parsers."

The normal solver hangs and churns forever on many Berkeley-Spice-compliant models.

Wonder why ?>:-}

Since there's no .OUT file generated by LTspice, one can't compare and see what models are used for "normal" versus "alternate". My suspicion is that "normal" substitutes over-simplified behavioral models to enhance the speed (like idealized diodes, etc) and barfs when its linearizations methods encounter an unusual model.

I discovered this when I ran a device with an S-shaped transfer curve thru LTspice. LTspice "normal" solver completely linearized it, then hung half-way thru the simulation. It matches real-world and PSpice if you select the "alternate" solver.

So I've recommended to my clients that use LTspice to always set "alternate" solver... slower by about a factor of two, but accurate.

Yes. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

Sorry, I meant version4 of Pspice.

The question is whether encrypted LTmodels will (still?) work on the same old and easily available versions of Pspice, or if they, like other vendors' encrypted models, need V9 or higher.

RL

Reply to
legg

That's too far back for me to remember ;-) The earliest version I still have documentation for is PSpice v6.3. I presently use PSpice v15.7, which is 9 years old... that's when Cadence stopped really issuing updates... just cosmetic changes at $1800 a pop, plus I had to undo the INI file munges they did to castrate features :-(

I've not tried an LTspice-encrypted model on PSpice. IIRC it doesn't work. PSpice-encrypted models only work on PSpice. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

I just tried an encrypted LTspice model (LT3971) on PSpice... nope, can't read it. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

I disagree totally. LT sells LTspice and models (read: support). They charge a pretty penny for it, too.

Reply to
krw

Then I think they're just shooting themselves in the foot.

If they're going to encrypt, they should remain compatible with the SW packages that the non-encrypted versions ran in. Other semi vendors may encrypt,but it's to run in a 'common' vehicle.

Perhaps LT has figured out that there are now more LTspice users than there are Pspice users? A pretty bold move, if so.

This would make LTspice the common vehicle, but one that other semi vendors may tend to continue to avoid. I'm pretty sure I've seen ~some vendors models that were specifically listed as LTspice compatible - but off-hand can only see EPC's (GaN), without serious searching.

Just issuing non-encrypted pspice models would be enough for everyone to get along.....we already know that behavioral models don't reveal crap about a chip's internal characteristics.

RL

Reply to
legg

Can you be a bit more specific about that statement?

No-one at LT has ever denied me support with ICs - and new models seem to be issued fairly frequently, if not exactly on-demand.

Cadence doesn't seem to consider model function to be it's responsibility...and they are not historically much use with issues in their own GUI, even after taking your money.

RL

Reply to
legg

Their parts cost at least 2x the competition. They justify it with the tools (support). It's great for small customers but I can't afford them.

Exactly. No one at TI has refused to sit down with me and tell me what I did wrong. ;-) On demand.

Well, models aren't Cadence responsibility.

Reply to
krw

It seems to me a really stupid marketing model to make your Spice models not work on any simulator but LTspice !??

But it may provide me with an opportunity to write cross-platform models and charge for it >:-} ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

Most of them don't own the foundry device models, so they're probably contractually prevented from publishing the real innards.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Most Spice users wouldn't be happy with device-level models because they would run more slowly.

The real problem for users is that the behavioral models created by most chip vendors are real crap and have annoying convergence issues. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.