Finding dips in column of data

[MattyM]

Apologies if this has been answered elsewhere but I'm struggling to find a solution.

I have a table of data for a series of battery's that have had their voltages measured over a number of weeks. Due to the testing I've been doing, there are dips in this voltage roughly every two hours and I'm looking to produce a graph of these dips. I've put an example of one of those dips below.

0.082535	3.5	3.54	3.52	3.45
0.083229	3.5	3.54	3.52	3.45
0.083924	3.5	3.43	3.42	3.35
0.084618	3.5	3.52	3.51	3.44
0.085313	3.5	3.53	3.51	3.45

Unfortunately due to the nature of the tests, the dips are not exactly every two hours which prevents my original plan to just extract cell data at every nth entry. My next idea was to write a formula that compares each cell to the cells above and below it, and extract any data that is less than both values by at least 0.05. However as shown above, one of the batteries took a while to begin to show a dip which means the data here won't be extracted properly.

What I think I need to do is to check each row of cells, and extract the entire row if one or more entries are less than the entries above or below by 0.05 or more. However my excel experience is limited and I'm struggling with the execution of this.
Any help or alternative ideas are greatly appreciated.

 

[Kaper]

The first column is obvoiusly flat at exactly 3.5. Was it measured with the same equipment and the same settings as next columns?

Do you expect phenomenon appearing in different columns at different moments?

If yes I'd analyse each column separately and then for first column there was nothing and in next 3 coulmns we see local minimum near

02:00:51

If the phenomenon shall occur at the same taime in all columns - prepare helper column with averages of all 4 columns and test it for local minima.
The concept of checking if the value is smaller than previous and smaller than next one by some "threshold" value to avoid small fluctuations influence on results is probably good one (or at least worth checking).

One more idea:
if you can propose some parametrized function to fit changes in voltage with time you could use Solver tool to optimize these parameters to obtain best fit (may be with least squares method).