Voltage dividers and noise

[gregcope]

Hi all

On another post Felix suggested a 0.1u cap on a voltage divider to help reduce noise?

If I am planning a 27m/3m low power voltage divider to measure 30vdc (on all the time) do I need a cap?  If so any references I can easily learn from?

Thanks

Greg

[TomWS]

Hi all

On another post Felix suggested a 0.1u cap on a voltage divider to help reduce noise?

If I am planning a 27m/3m low power voltage divider to measure 30vdc (on all the time) do I need a cap?  If so any references I can easily learn from?

Thanks

Greg

Definitely need a cap.  0.1 is perfect.  Make sure it is a low leakage type.   There were some references cited in the thread related to weathershield power consumption IIRC. 

BTW, the cap is useful for reducing noise, but absolutely ESSENTIAL for maintaining the voltage at the analog input pin during the ADC conversion cycle.  The analog mux will switch a capacitive load onto the line that will cause a momentary load on the analog input pin.  This load can either be offset by a low resistance source or a capacitor large enough to prevent a >1 LSB voltage change on the input.  A 0.01 uF is probably the minimum you can use.  Anything larger than 0.1uF is overkill and may cause problems due to leakage.

Tom

[joelucid]

absolutely ESSENTIAL for maintaining the voltage at the analog input pin during the ADC conversion cycle

Are you sure? The datasheet has this to offer:

The ADC contains a Sample and Hold circuit which ensures that the input voltage to the ADC is held at a
constant level during conversion. A block diagram of the ADC is shown in Figure 24-1 on page 238.

I have never used a cap on a voltage divider even though I use extremely large resistors (many megaohm). My measurements are neither noisy nor in any other way suspect as far as I can tell.

Only thing I've found to be necessary is to discard the first measurement after switching to MUX to give the divider time to load the S/H circuit.

[TomWS]

Only thing I've found to be necessary is to discard the first measurement after switching to MUX to give the divider time to load the S/H circuit.

Oh yeah, the thing I failed to mention is that you can skip the cap if you do two back to back reads with the same MUX setting...

Um and the reason is that you need a cap if you want an accurate read on the first one.

Tom