LM3429: PSPICE for TI built-in library model appears to have problems

Mark, 

Sorry that I could not understand the circuit you attached. Are you going to design a boost circuit, buck converter or other? What is the purpose for ROV1, ROV2 in the circuit? 

You may refer to LM3429 PSPICE simulation circuit for design. If buck - boost is needed, you need to change OV feedback following design guide in datasheet. 

The topology that I am using is a buck-boost (input higher of lower than output setting) and I am essentially copying the model from that datasheet that you pasted into your response. ROV1 and ROV2 are stubs for after I get basic simulation working. I have tried an over-ride voltage above or below the threshold of "OVP" but startup fails after an initial pulse. As I did with the LM3409 model, I am running the .TRAN simulation so that I have to stop the initial calculation of bias point; the difference is that the LM3409 simulation ran fine.

I have a suspicion that the simulation model is broken or there is some "trick" to using it.  And, I suspect this because the "NC" pin needs to be connected as shown in my initial drawing, or the simulator refuses to run. (The "NC" pin is, of course, a "no connect" pin.)

Have you used the LM3429 model ?

Mark, 

I tried your circuit with a single PNP for Vo sensing mentioned in datasheet. It shows that the model works for me.

I looked carefully at your simulation traces, and realized that the model started working at about 27 milliseconds.  Based my experience with the LM3409 model, I was only simulating to 200 microseconds and getting activity from essentially time=zero (and quick runtimes).  So I re-ran the simulation a few times starting with about 70 ms to make sure I did not miss anything.  As you see, in my plots, the LM3429 model starts-up for me at exactly 4.0 milliseconds. This is very strange that there is such a disparity between your start-up time and mine.  To save time, I am now running the simulations for 7 milliseconds.SimPlot-SepTraces_0-7ms.pdf

So yes, the model sort-of works. But the LM3429 model should start-up within a few microseconds.

Lastly, your schematic gave me a clue about the NC (no connect) pin.  I had tried to put the schematic "X" indicating a non-connected pin, but PSPICE would not tolerate that, so I had to put two very high value resistors between Vin and Ground.  That made PSPICE happy, but it was messy.  I tried your solution of just attaching a short wire stub to the NC pin, and, voila, PSPICE was happy.  This behavior is clearly not a problem with the model but a problem related to either the schematic capture or the module that extracts the circuit from the schematic.