Modeling an IC as a black box

Thread Starter

Avshi

Joined Jan 16, 2015
48
Hello,
I'm trying to model an IC (DRV8801a-q1), but the internal components/structure are unknown. Is it possible to do something similar to the black box approach and use information from the data sheet to create a sort of model that has exhibits behavior only based on the data sheet?
 

Papabravo

Joined Feb 24, 2006
21,225
It depends on the IC. Many spice models for discrete components cannot be completed just by looking at the datasheet. That is why most models come from the manufacturers since they have the ability to characterize various parameters of the models which are ordinarily not useful to circuit designers.
 

Alec_t

Joined Sep 17, 2013
14,313
You can certainly create a model based only on the datasheet, but its accuracy will always be unknown because datasheets don't necessarily include all the nuances of ciruit behaviour (e.g. the temperature dependence of certain parameters may not be fully specced). Nevertheless the model can be useful; providing its limitations are allowed for.
 

WBahn

Joined Mar 31, 2012
30,056
The first thing you have to decide is what behaviors are important for your model to mimic and how well does it need to mimic them. That will go a long way to determining how you create your model.

Do you have the ability to make ABM (Analog Behavioral Models)?

A couple decades ago I was using Spice models from TI for some of their opamps and I was interested in total power consumption. But their models didn't model total power consumption at all and were only interested in getting good fidelity at the I/O pins. As a result, the current drawn from the power pins was in the thousand ampere range (when, in reality, it was a couple hundred microamps).
 

Thread Starter

Avshi

Joined Jan 16, 2015
48
The essential functions are the able to control the speed and direction of a dc motor, in which case I was thinking I could emulate a full bridge, but it would be nice to try to emulate as many functions as possible. Unfortunately no, my experience with modeling is very basic (rc circuits).
 

WBahn

Joined Mar 31, 2012
30,056
An IC is an electrical component that has inputs and outputs. The behavior is basically if the input is this I want the output to be that. It also needs to take into account the effect it has on the circuitry it is connected to. For instance, if I apply 5V to the input it will sink 1mA of current and if I connect the output to a 100Ω load and it is supposed to output 4V it will actually only output 3.7V (and deliver 37mA to the load). There are lots of behaviors that might need to be modeled, or it is possible that just a few simple ones are all that are important. But YOU have to decide what is and what is not important.
 

Thread Starter

Avshi

Joined Jan 16, 2015
48
Okay, I know what I should include. How should I proceed? Should I create a circuit that does something similar?
 

WBahn

Joined Mar 31, 2012
30,056
You should create a circuit (since ABM isn't an option) that behaves "close enough" to the actual circuit, meaning that its behaviors that you have identified as important are the same as the actual circuit to within the level of accuracy that you need. The circuit that you create to accomplish that depends entirely one what the behaviors are that you are trying to model and how accurately you need to model them.
 

Alec_t

Joined Sep 17, 2013
14,313
Fig 7.2 of the datasheet for the DRV8801a-q1 IC shows the functions which you could model using discrete components and simple logic ICs. Functions such as undervoltage and temperature sensing may or may not be important to you.
 
Top