finding peak or shoulder of a curve

[rpursley8]

When I did it, I first took the log of the data and then filtered it before I took the first derivative.  The peak seemed to be where I expected.

[unclebump]

Try this llb file. It uses limit testing with arrays and clusters.

[Telemaque]

First of all, thanks you two for taking the time from your day to help me with this; I hope I can do the same for someone else to repay the community.
rpursley8, your method, though it doesn't seem to help much with the nice peak case, looks to work well for the bad peak one!  For good peaks, I can just look for a negative-to-positive cross-over, and when the program finds no such cross-over, I can use try your method.  Thanks, I look forward to playing with it.
unclebump, I need to evaluate your suggestion more closely.  I am not sure how setting the limits would help me in my case as the amplitudes (counts per second) could vary greatly depending on how close my detector is to the source.