Using TI's encrypted MOSFET model in LTSpice- error: too few nodes

Dear Texas Instruments,

the fact that the models are encrypted is RIDICULOUS, but you don't see me laughing! Am I supposed to trust that the simulation model actually correspond to the datasheet? How do I change the model parameters to account for temperature dependence ?!  I assume your model is only correct at the ideal temperature of 25C.

 

Regards,

Gert

Gert,

Thanks for providing your opinion. Encryption is a necessary evil at this point in time. You may consider it ridiculous, but you may want to ask a different question as to why it exists in the first place. Encryption is protection for the model provider as to ANY intellectual property that is contained in the model. It is also protection to keep someone from changing the model without having the proper training or understanding of how the model was made.

Do I expect you to trust that the simulation model accurately represents the datasheet? Yes, but I also expect you to trust your engineering judgement when you see results that do not make sense. The best model in the world placed in a poor implementation looks poor. An an Engineer, you should trust your engineering judgement moreso than a linear equation solver. Simulation does not know what you are trying to do. It is trying to solve a matrix based on a bunch of rules and limits. Modeling Engineers strive to provide the most accurate product models that simulate quickly and represent measured data for a given product and its design.

Temperature simulation is a very complicated issue. Yes, the nominal PSpice models are correct at room temperature and no additional temperature is added to the simulations to approximate temperature behavior. The vast majority of the users I am familiar with would not be able to modify the model (or its parameters) properly and get the actual temperature behavior. If it was a very simple task, it would already be incorporated into the modeling process.

I am not trying to change your opinion about encryption or our models in general.

Britt

Britt,

Thank you for the reply. Are you selling the physical device or the model? Thus, IMO, the disclosure of intellectual property in the model, from which the physical device can hardly be reverse engineered, is what ought to be the necessary evil. The reason of protecting the user of the model by encryption is a fallacy and severely limits the capabilities of those users who do know what they are doing. I do hope that the time will come when encryption will no longer be necessary and here is why:

I already have a simulation model of a much larger system in LTSPICE. Do you expect me to convert my model to another simulator? No Sir. I would really have liked to use the CSD19531Q5A Mosfet from TI. It really has an amazing Miller Charge characteristic, unlike I have ever seen. From an end user perspective I gain trust in a model when I can simulate it. And better yet, when I can adjust its parameters and perform a sensitivity analysis, because your right I do have to use my own judgement, but that judgement is based on quantitative results.

I am sorry to say that without that trust I will have to choose another manufacturer (and I am not saying that as a petty trump card).

Regards,
Gert

Hi Gert,

I can understand your frustration with the encryption of models. It does limit the usage of the models to the simulator in which they are encrypted. However the models do contain sensitive IP information in many cases which can be reverse engineered so we can not always release them unencrypted on the web site.

However, if you do need the unencrypted model, we can provide that to you individually. In this case, we will need an NDA with you and your company. I would recommend that you contact your local TI AFA for this. Let me know if you need any help getting this information.

Thanks for considering TI Parts in your application.

Best regards,