What about boxplots with one box per time?
This way you'll capture the whole distribution of the 10 values and you can still have a compact 2D representation of the phenomenon.
Example 3 in the documentation can give you an idea.
On the X-axis the time and on the Y-axis the number of faults.
EDIT 12:57 GMT 00 15 March 2011
I wouldn't store the faults in excel at all. I would use a matrix and store each sample as a column. The resulting matrix can be fed to boxplot as it is.