I believe I started a similar discussion quite some time ago, but can’t seem to locate it. In any case, I’m inquiring about the use of bypass capacitors to shunt noise away from inputs. Typically, I use a precision 10K resistor pulldown with either dry contacts or a 10K thermistor to 3.3 Vdc. This has worked very reliably in nearly all situations, but occasionally I will see what might be noise on an analog input. I do sample several times in order to determine the average value of an input.
My earlier PCB designs included a 0.1uF capacitor across each 10K resistor as a means to shunt noise to ground. For a reason I don’t recall, I removed these from later designs. Before I simply include these in new boards I need to order, I’m asking if this is a good technique and if this is a reasonable value for this purpose.
How fast does the input signal change? A cap that big will slow things down, especially when used with a thermistor (cap is effectively charged through the thermistor, hence has an RC time constant)
The actual input is relatively slow changing–air/water temperature in commercial HVAC. 0.1uF seems to be the “standard” for bypass capacitors from a novice perspective and they seemingly worked well on all my earlier boards, hence the choice. I’m a bit surprised by your characterization of “A cap that big…”. These babies are tiny (50v) and I wondered if they would be effective at all.
Big and small are always relative. The RC time constant that Hugo mentions is (10K * 0.1uF) = 1msec, for your case, so roughly speaking, this will be the fastest sample rate you can expect (or alternately, any noise lasting for much less than this will be filtered out). Of course, you need to make sure you have this in a “low-pass” filter configuration for your input, to keep the noise out.