pulsewidth not working with higher sampling rates?

I have a waveform like this:

And I would like to know the width of this spike at half height, a nice function to do so is pulsewidth (thank you Daniel M!!). Since the waveform is originally sampled at 200 kHz, I tried running pulsewidth using this code:

figure
pulsewidth(x,2e5);

This is the output:

There is no pulse width identified.

Now I resample the waveform at 50 kHz and run pulsewidth again using the following code:

x = resample(x,5e4,2e5);
figure
pulsewidth(x,5e4);

And here is my pulsewidth clearly identified:

Anyone know why this would not work at 200 kHz; but work at 50 kHz?

BTW I've also tried using a time-stamped vector for input instead Fs; but get the same exact output. Also tried adding the statelevels for max and min values on this trace.

Thank you in advance!!

Eric

[Daniel M]

I'm going to comment instead of answer, so the question can remain more visible/

I think it has to do with the statelevel detection being different in the two figures.

Check out my answer on your previous question, because I think it is relevant here too.

https://www.mathworks.com/matlabcentral/answers/489368-how-do-i-re-sample-a-continuous-function-but-maintain-shape#answer_400174

[Eric Kuebler]

I actually didn't notice that the function identified the post-peak portion as the lower state with the 200kHz waveform.

Perhaps the edge effects in the 50 kHz waveform has something to do with the function picking the pre-peak portion as lower state?

Sorry about posting two questions, I suppose I could have posted in the other question; but wasn't sure the answer would be overlapping.

Thank you!

[Daniel M]

Yes, probably because of the edge effects. Try to deal with those. My proposed solution also seems like it would be robust to the edge effects, but I can't test that.