SMPS switching losses in LTspice ..accurate?

[Flyback]

Hello,
LTspice calculates a switching loss of 1.07W in the FET of the buckboost smps that I have uploaded here. "Buckboost SMPS _switching loss.asc".
How accurate is this to the real circuit? I mean, the switching Mosfet capacitances, especially Cgd, are voltage dependent, and this would need modelling to make the switching losses accurate. How accurate is it in LTspice?

 

[alec_t]

How accurate is it in LTspice?

You should be asking Linear Technology. They know about the proprietary way the sums are done.

 

[Flyback]

Yes I put a qu on their yahoo groups forum...but they say it depends on the model of the fet...so I told them I am using the model of the fet provided with the ltspice simulator package. There has been no conclusive answer about whether linear.com's provided mosfet models accurately model the cds capacitance of the fet and its variability with changing voltage.

 

[alec_t]

Perhaps you could model Cds yourself, by adding a varactor between drain and source?

 

[ronsimpson]

In spice there are so many things not entered.
PCB inductance and capacitance on high speed traces like the Gate and Drain on the MOSFET. The real world current fall time can be very different than you thought. A fast diode on long traces will ring.
The inductor (transformer) will not be a simple coil.
The insulator on the power transistor to heat sing will change things. (capacitance)

 

[crutschow]

Offhand I would think the MOSFET models are adequate, including parasitic capacitances, to give a reasonably accurate calculation of the transistor switching losses, at least sufficiently adequate for most engineering design requirements. Note that the gate capacitance has little effect on the MOSFET dissipation itself. The dissipation from that occurs in the gate driver circuitry (including any added gate resistor).

 

[Flyback]

I am not suggesting that the mosfet models in the provided mosfet models from linear.com have no modelled device capacitances. I am wondering if these device capacitances, such as the cgd capacitance, vary with voltage applied to it?
The cgd capacitance has a large bearing on switching losses.

 

[crutschow]

I am not suggesting that the mosfet models in the provided mosfet models from linear.com have no modelled device capacitances. I am wondering if these device capacitances, such as the cgd capacitance, vary with voltage applied to it?
The cgd capacitance has a large bearing on switching losses.

Why do you think it significantly varies with voltage? It's not like junction reverse capacitance, which does have a large voltage coeficient.

 

[Flyback]

See page 4, RHS of the following which states how voltage dependent cgd and cds are...
**broken link removed**

 

[alec_t]

According to the LTS 'Help' on capacitor modelling:

There is also a general nonlinear capacitor available. Instead of specifying the capacitance, one writes an expression for the charge.........
This device is useful for rapidly evaluating the behavior of a new a hypothetical charge model for, e.g., a transistor.

Can you make use of that to model Cds and Cgd explicitly?

 

[Mosaic]

LTSpice does not model circuit layout, pcb trace inductance, crosstalk and so on. You need to 'model' for these parameters by adding resistance, inductance & capacitance as parasitics based on what you expect to see in the real world.
I have designed circuits that work 'out of the box' and circuits which oscillate on the breadboard because of parasitics.

Sometimes you do your best in Spice but there is no exchange for getting out the breadboard or soldering iron with a good scope / logic analyzer & DMM.

 

[tvtech]

Spice is OK..

Not good enough for real world though....
Especially when dealing with high voltages/currents....

Regards,
tvtech

 

[killivolt]

I don't know if what I found on youtube will apply to your buckboost SMPS; it's above my head. So, I apologize now.

This guy builds a Flyback by hand and then does (Analysis of a Flyback part 1-8) on youtube a warning ahead of time "Boring" but he did some modeling in LTspice of the Transformer and some pre-calculations for external components. After doing a simulation he felt he was close enough. He then put into the real world with the external components created in LTspice and took final measurements; it performed very close to the Simulation.

Someone else may find this interesting as well; I hope.