Modelling SPICE load by AC I-V data

[hopeful1]

Hi there,

I am trying to simulate an AC circuit with a real-world AC-DC power supply as the load. The input (AC side) I-V curve (rms current vs rms voltage relationship) of the supply has been obtained by physical measurement with a resistive load. I want the load in my SPICE model to show the same I-V behaviour as my real-world supply.

Using a non-linear resistor (with I-V data as input) does not do the work, seeing as the resistor evaluates the instantaneous voltage and not the RMS value. The same goes for voltage-controlled current sources.

I would be grateful if anyone have suggestions to how this might be solved. I am using B2Spice for my simulations, but suggestions for other SPICE software will be gladly appreciated as well. Please let me know if my problem statement is unclear.

 

[Harald Kapp]

I don't think that your approach using rms measurements to model the load is going to work.
Any SPICE based simulator operates oninstantaneous values. However, rms values are based on integrating instantaneous values over time. You can know the rms value of e.g. a voltage only after having measured and integrated it for some time. Therefore you know the rms value only after the time interval has elapsed. But for simulation you would need that value already at the beginning of the interval. That is a contradiction in itself.

How exact has the model of your load to be? Can you approximate it by a simple circuit? What do you mean by this:

a real-world AC-DC power supply as the load

Is this the input circuit of a power supply (rectifier, capacitor, etc.)? You can model this with a handful of components in SPICE.

 

[hopeful1]

Thank you for your reply, Harald. Your thoughts on the rms contradiction makes very much sense. I reckon that I will need to approximate it by a simple circuit.

Yes, the load should represent the input circuit of a power supply. I have tried to model the load with a rectifier and a capacitor, but I find it hard to fit these with my empirical data. Can the following curve be (approximately) reproduced with such a model by altering the values of components?

 

[Harald Kapp]

Can the following curve be (approximately) reproduced with such a model by altering the values of components?

That depends on the model circuit you use.

I think what you see here can be modeled mathematically simply as I=P/U where P is the power consumption of your load.

But I think the problem is buried deeper: You heve't described what exactly you want to simulate.
Are you interested in the behaviour of the input circuit of the power supply? Then you need to model this circuit as good as required to give you the answers you seek.
Or do you want to simulate the behaviour of another circuit (e.g. an electronic fuse) in reaction to the power supply input connected to this circuit? Then you need to know which effects are going to influence this circuit and you need to build a model of the power supply input that can generate these effects. You could look to a behavioural source with an appropriate function. You wil have to test the model of your input under the different operating conditions that exist in the circuit's environment. You may even need different models for different effects, e.g. for modeling rms behaviour r for modeling instantaneous behaviour.