Diode Turn-On Threshold for 3-LVL-ANPC necessary? (Circuit Model Options)

Question

Hello,

I would like to investigate different modulation strategies for a 3-LVL-ANPC-SiC topology under different operating points in terms of overall efficiency and power dissipation (temperature) per switch.

For my switches (Wolfspeed C3M0021120K), I use the thermal description provided by the manufacturer, one for the MOSFET and one for the intrinsic body diode. I have compiled both into a masked Switch submodel. Since I use a 3-LVL-ANPC topology, I have 18 switch submodels accordingly.

I now have the option in PLECS in the Simulation Parameters under Circuit Model Options to define a value other than 0 for the diode turn on threshold (Vth)

• Is it advisable to define a value Vth other than 0 here for my topology? 
• In reality, the diode turn on threshold is a non-linear value, i.e. temperature-dependent, among other things. Based on the data sheet, I have two different values for Vth for the temperatures 25°C and 175°C. Which of these values should I specify (if this is necessary for my topology)? 

I would really appreciate an answer. I have attached pictures to illustrate the problem.

Kind regards 

THK

Answer 1

Please this note from our user manual regarding the Diode Turn-On Threshold solver setting:

<<

Note  The Diode Turn-On Threshold is not equivalent to the voltage drop across a device when it is conducting. The turn-on threshold only delays the instant when a device turns on. The voltage drop across a device is solely determined by the forward voltage and/or on-resistance specified in the device parameters.

>>

This is a global setting for the model, and is not related to the Vf value of any given diode in your model.

By the way, are you intentionally including the variable resistor block in your model in this case? This will create a (likely solvable) algebraic loop in your model and therefore we don't recommend this approach unless such effect is really something of interest for your particular study. Further, if this is at all related to another question you asked about steady-state convergence, I do assume that such loop solving at each simulation step certainly would slow things down.

Comments

Hello Krls,

thank you very much for your answer.

• In the device parameters, I can normally only specify a fixed value Vf for the forward voltage of the diode, but not R_on_Diode=f(Tj, I_SD), right?
• Unless I would install a variable resistor in series to the diode and control it via a 2D look-up table which uses Tj_diode, I_diode as inputs. But this approach would lead to unsolvable algebraic loops again, wouldn't it?

I guess I have understood the meaning and the effect of Vth on the diode behavior in theory. Unfortunately, I don't see any effect in my circuit if I set Vth=4.5V vs. Vth=0V. The Diode current, diode forward voltage and time of conduction of the diode are exactly the same despite different Vth. In addition, the error messages that occur due to algebraic loops that cannot be solved at certain operating points remain despite Vth≠0V.

 

With regard to the variable resistors. Yes, I use these on purpose to map the temperature and current-dependent R_DS_on. I got the data for this from the PLECS thermal descriptions from Wolfspeed. Since I am investigating the influence of different control strategies under different operating points on the power dissipation and junction temperature of each Switch in the 3-LVL-ANPC topology, I assume that this R_DS_on(I_DS, Tj) is essential for my work in order to better represent "reality".

• Also, you mentioned in your answer that the algebraic loop is likely solvable. Could you please tell me what steps are necessary to achieve this?

Since you asked me in another of my questions, I will there gladly upload my PLECS model for you.

 

Kind regards 

THK

---

Yes, with the standard diode you can only specify the Vf and the Ron (fixed, non-temperature dependent). If you are doing steady-state simulations then a fixed Ron is fully appropriate and just needs to be tuned to the proper value.

I see my colleague answered your other questions and I think the main message there is that having multiple algebraic loops in a single model is not straightforward for the solver, so you might consider only applying this dynamic (temperature-dependent) Ron approach to one of the switches at a time. I'd then be more confident in saying that the algebraic loop is likely solvable.