Yes, this subject becomes even more important when you are sampling from an analog multiplexer. Then the multiplexer output must settle before the sample is taken, which must also settle. :-)
Another similar important note is at the ADC reference. Like the input itself, some ADCs draw small surge currents from the reference, so a well thought out capacitance at the reference and a low impedance reference voltage source may reduce complications (and confusion) when figuring out settling time.
I think that is where experience comes into play, Jon - and why it is so important for junior engineers to learn from senior engineers. I am forever grateful for the engineers that took the time to teach me real world things that you don't necessarily learn (or remember) from school - including settling times!
Thanks, Nancy. Often we overlook details that later cause problems. I know of several cases in which not accounting for settline time caused erroneous measurements. Thankfully the people involved discovered the problem.
Great article, Jon – thanks so much for sharing on a sometimes overlooked but extremely important part of taking accurate measurements. Settling times should be taken into account and I think how that is best handled depends on the circuitry, what is being measured and what is being used to take the measurements. On test sets we would keep cables as short as possible and in the old days we would simple do that through software by trial and error to come up with a reliable software delay (scary I know!) However, with the systems we were working on we were mostly taking DC measurements that were extremely repeatable so it worked out okay. I really appreciate the variety of solutions you offered – with the complexity of today's designs they offer great solutions and makes me want to go experiment!