In general what do you do when some kind of data is missing from some data that follows exponential distribution. For example, say 3 observations are made by an instrument where x1=5, x2=3, but for x3 the instrument can not give a specific answer because it cant measure past 10. So the only info we have about x3 is that x3>10. How can we come up with a likelihood function of lambda for this data (which is exponential distributed)? do we just use x3=10, or what?