Scaled analog value?


I’m using analogs and I have a strange behavior:
If Vref = Vcc = 3.3V, the value reaches 1023 at 2.16V.
This seems to be about a 2/3 ratio regarding the theoritical 1023 for 3.3V!

And amazing:
If Vref = not connected, 1023 is reached for 1.75V!
If Vref = GND, 1023 is reached for 1.28V!

I’ve two parts and they have the same behavior…

Any idea?

(X-CTU reports: XBP24 - XBEE PRO 802.15.4 - firmware version 10E6)


My only thought is to ask whether you have a decoupling capacitor fitted between the analog input and VREF. The need for it is mentioned in footnote 1 on page 8 of the 1xEx manual, and it’s quite easy to miss. The capacitor should be in the range 0.01 to 0.1 microfarads.

Hmm. Now that I’ve looked at it again, it actually says VREFL so maybe the capacitor should be connected between the analog pin and ground. Does anyone know whether it matters?

The source has a very low impedance and has decoupling.
Nevertheless, I don’t see how this would do with a 2/3 ratio.
Returned value is very stable (no noise).

About where to put a decoupling cap: IMHO to put a cap between signal and Vref is a good way, it makes a kind of common mode rejection. But it depends on where the signal is referenced - on Vref or on Gnd (Vref- is not available: this VSSad reference in the manuals is meaningless).


Nobody has ideas?
Should I assume that my parts have some trouble?


You could wait a little to see whether there are any other responses, but in the absence of a helpful resopnse here I believe the kind folks at Digi will be happy for you to raise a support request directly.

My own XBees behave sanely in this respect. Getting the same behaviour on two parts would seem to indicate a circuit problem in general.

Oh! Wait a minute. On the lines to which they apply, pullup resistors are enabled by default. That includes all the analog inputs except number 5. Does input 5 behave differently, and/or what value do you have set for the PR parameter?