"Real" SPICE model for sync buck chip anywhere?

[Joerg]

Hello Folks,

Just to kick the tires of an idea I need a sync buck IC SPICE model.
Simple enough so the sims don't take forever but not fully behavioral
like the LTC ones are.

The ideal model would look like this: The FB comparator, current sense,
bootstrap, output drivers and such are all fully transistor-level
modeled. Mainly because I'll need to "reach in". The housekeeping stuff
and oscillator can be behavioral. This is because full transistor-level
models make sims very sluggish.

Anyone know a good model for that?

 

[Joerg]

Write your own ?

I might (have to) do that. I have designed semi-discrete switch mode
converters, meaning sans commercial PWM chip. But they contain logic
chips used in an unorthodox fashion for which there are no models. So
I'd have to make models for those. Not exactly a fun job.

I presume you're using an LT device? ...

It doesn't matter. It's just a theoretical experiment, not for a client
or something.

... Where are you going to get device-level models for the output?

Some models (older Unitrode chips etc.) are 100% device-level so that
would work. Except that even the timing stuff is device-level and then
sim runs take forever. Because upon each time step it has to also slosh
through 15-20 transistors in the RS flip flop, and another few hundred
in housekeeping circuits. Big waste. Those sims made the office temp
rise by several degrees. To my surprise one of our Labradors likes that
and curls up by the PC like a giant cat.

I do that sort of thing... for a fee ;-)

I know

Could pay in buckeye balls. But they are dwindling fast ...

 

[Joerg]

Let's narrow it down... you simply want a "buck core model" that you
add pass device and flyback diode?? Synchronized?

Not synchronized but synchronous. Meaning a bottom FET instead of the
diode. I think I am just going to start wit the LTC3810. It requires
external FETs, one can get at the internally regulated voltage and I
know its behoir because I've used it in real life. So I'd (hopefully) be
able to concoct some sort of spoiler up front of the FB pin if needed.

 

[Joerg]

Not synchronized but synchronous. Meaning a bottom FET instead of the
diode. I think I am just going to start wit the LTC3810. It requires
external FETs, one can get at the internally regulated voltage and I
know its behoir because I've used it in real life. So I'd (hopefully) be
able to concoct some sort of spoiler up front of the FB pin if needed.

Whoops, I meant behavior. Maybe the 9 volts of the Belgian style ale
yesterday was too much

The fun starts at 2:30min.

 

[Joerg]

Jim Thompson wrote: [snip]

Let's narrow it down... you simply want a "buck core model" that you
add pass device and flyback diode?? Synchronized?

Not synchronized but synchronous. Meaning a bottom FET instead of the
diode. I think I am just going to start wit the LTC3810.

[snip]

Encrypted :-(

They all are :-(

You don't like LT's model ?

Not really but one has to use what's there. It's largely behavioral. But
I can crochet some stuff around it to make it less so.