SPICE model for SN74AC14

[elki]

Basically, the subject I was able to find the official SPICE model for SN74AC14 inverter https://www.ti.com/product/SN74AC14-Q1, but somehow the model has only four pins available for the assignment: one input, one output, positive voltage supply and the ground. When trying to use it in KiCAD, I see that there is no way to assign all IC pins in this limited behavioral model. Is there a version with all pins encoded in the routine or maybe there is some other way to do it? Thanks!

[tggzzz]

Basically, the subject I was able to find the official SPICE model for SN74AC14 inverter https://www.ti.com/product/SN74AC14-Q1, but somehow the model has only four pins available for the assignment: one input, one output, positive voltage supply and the ground. When trying to use it in KiCAD, I see that there is no way to assign all IC pins in this limited behavioral model. Is there a version with all pins encoded in the routine or maybe there is some other way to do it? Thanks!

Firstly, it would be sensible to define what aspect of its operation you are trying to simulate, e.g. threshold, ground bounce, output drive, etc.

Secondly, the model you mention is simple text so you can modify it to contain more gates.

[elki]

Thanks a lot, I will implement the missing gates directly.

[Zero999]

Going from your first post, your model is for a single gate only. This is probably because it's possible to get single gate packages i.e. the 74AC1G14, so it makes sense to generate a model for that.

The 74AC14 is exactly the same, just with six gates, so it's a non-issue.

[T3sl4co1l]

Ya, even if there were no single-gate packages, a single-gate model is most sensible because there's no point in simulating a half-dozen gates per instance when you might only need exactly one of them.

Make a multi-part component.  You cannot make a block (full IC) with single models, well, unless it allows assigning multiple instances (that would probably be a mess), or you craft a "package" or "wrapper" model yourself (which is certainly doable).

Maybe KiCAD doesn't allow the mismatch regardless, and it's a feature they still need to add.  I don't know.  All other tools I've used (which is only a couple, to be fair) have worked as above.

Even if you can't make a multi-part component with models associated to parts, you can still make a single-part component and model, and simulate that.  You don't have to have a simulation perfectly in sync with an EDA design (SCH+PCB), it's just supporting information.  Having them together is mainly a matter of convenience and confidence, particularly in large designs; assuming one is able to effectively simulate a given design at scale, of course.

Tim

[elki]

Thank you very much for all the useful feedback! I ended up expanding the single-gate model file to multiple gates. I agree that for the simulation purposes one can just use multiple instances of the single-gate version.

[tggzzz]

Thank you very much for all the useful feedback! I ended up expanding the single-gate model file to multiple gates. I agree that for the simulation purposes one can just use multiple instances of the single-gate version.

Not necessarily. If all outputs simultaneously switch heavy loads, the package parasitics (especially ground lead inductance) can cause problems. Not an issue with single gate devices, of courae.

[elki]

You are right!

[Benta]

You're using KiCAD wrong.
The 74AC14 (actually 74HC14) symbol contains 7 elements: 6 inverters and one power unit (A...G).
You'll need to use at least one inverter (eg, A) and the power unit (G) in your schematic.
Edit the properties of the "A" unit and add the Spice model.
Proceed from there.

EDIT: good practice is to add all seven elements to the schematic (just click 7 times), push units B...F to one corner of the sheet and ground the inputs. Add "no connect" X-es to the outputs. Now they're pacified and harmless.

[elki]

Thanks, Benta!

[Benta]

Wow!
Three contributors with impressive post counts chiming in on a subject they obviously know absolutely nothing about.
Interesting.

[T3sl4co1l]

Wow!
Three contributors with impressive post counts chiming in on a subject they obviously know absolutely nothing about.
Interesting.

Four?

For my part, I'll take that as a compliment. Indicating ones' state of knowledge is a rare sight on the all-too-confident internet.

If nothing else, there's always the old, you got what you paid for.

Tim

[Benta]

[Zero999]

Wow!
Three contributors with impressive post counts chiming in on a subject they obviously know absolutely nothing about.
Interesting.

How does your post positively contribute to the discourse? Perhaps you should not post anything at all, unless it will be helpful, otherwise you just come across as being nasty or trolling.

Thank you very much for all the useful feedback! I ended up expanding the single-gate model file to multiple gates. I agree that for the simulation purposes one can just use multiple instances of the single-gate version.

Not necessarily. If all outputs simultaneously switch heavy loads, the package parasitics (especially ground lead inductance) can cause problems. Not an issue with single gate devices, of courae.

That's true, but the SPICE models I've seen don't model parasitics. Most are just behavioural. At best you get the propagation delays and output resistance.

[tggzzz]

Thank you very much for all the useful feedback! I ended up expanding the single-gate model file to multiple gates. I agree that for the simulation purposes one can just use multiple instances of the single-gate version.

Not necessarily. If all outputs simultaneously switch heavy loads, the package parasitics (especially ground lead inductance) can cause problems. Not an issue with single gate devices, of courae.

That's true, but the SPICE models I've seen don't model parasitics. Most are just behavioural. At best you get the propagation delays and output resistance.

Have you looked at IBIS models?

The entire reason they exist is to model package parasitics and i/o parameters so their effect can be simulated.

[T3sl4co1l]

Somewhere somehow, I've seen a package model like, here's lead inductance and resistance (probably resistance measured at frequency, not a general model thereof), and all the coupling factors between them, fairly round, but still.

...Oh, I think I found it!  It's bits of I guess "include" data, not a SUBCKT proper, in TI's HSPICE models -- plain text.  Looks like the section could easily be "packaged" as a .SUBCKT.  For example, find scaj006.zip.  The *.pkg seem to be of this format.  (Don't worry about the *.inc, unless you want to try and decrypt it.  Which, FREELIB looks disturbingly easy to crack, it seems to be a permutation space initialized every line or few; but that still leaves a lot of degrees of freedom between columns.)

Tim