LMR36015-Q1: LMR36015AQRNXRQ1

Hello

Try to increase your input voltage to 12V and see if it will start regulating.

Thanks

Hi Tried Simulating by increasing the voltage. Again I am getting output zero. I have used different model that I got from TI website.

Find below snap from where I have downloaded Pspice files

Hi Tried Simulating by increasing the voltage. Again I am getting output zero. I have used different model that I got from TI website.

Find below snap from where I have downloaded Pspice files

Hello

You might try setting the fastss = 1.

Thanks

Hi Frank

Still I am not getting output.

Hello

I will assign this to someone who can help with your issue.

Thanks

Hi Ajay -

Unfortunately it seems like that the model for that particular variant (LMR36015A) isn't working correctly - we'll try to fix this ASAP, sorry. If you're flexible on your switching frequency, LMR36015B (1MHz) and LMR36015FSC-Q1 (spread spectrum at 2.1MHz) are quite similar and I just checked that the models work. 

Hi Ajay --

I did some more digging and it seems to work with after changing the maximum step size in the transient analysis to under ~40ns (see attached image). Let us know if you face issues after trying that.

Hi Sekhar,

I also observed the same regarding step size. Now Simulation is running fine.

But what could be the possible reason for this step size.

Thanks

Hi Ajay - glad to hear it's working. I'm not super well-versed with spice, but what I think is happening is that the dynamic step size inner loop in the transient analysis algorithm is making the step size too large in the beginning and skipping over some of the fast digital edges (those are linearly modeled in spice), and in future timesteps a convergence to zero is still valid so it goes with it. 40ns is ~1/20th the period of the internal frequency so I think slowing the simulation down to a max of that ensures that it catches the fast edges, and then it dynamically lowers the step size to ensure accuracy. Any bigger than 40ns and the simulation was still wrong.

As to why the other models worked fine, perhaps those failed to converge in later timesteps so spice backtracked and reduced the step size or something along those lines.

Again I'm not very sure about the cause so if anyone finds flaws in this reasoning please point them out.