I recently used accelerometer ADXL335 in my academical project, this sensor gives analog output. I interfaced this sensor with microcontroller PIC18F458 using ADC module. It was successfully interfaced and shown correct readings. But later I noticed in datasheet of controller that "The maximum recommended impedance for analog sources is 2.5 kΩ" and datasheet of ADXL335 shows that its output impedance is 32kΩ. I have not used any signal conditioning circuit between sensor and controller. I am wondering how I am getting correct output even if source impedance requirement for ADC module of controller is not satisfied.
The source's output impedance (and capacitance) will affect the settling time in most ADCs.
An ADC can (depending on the type) do a good measurement of higher-impedance sources if you give it more time (to charge the sampling capacitor); the more time you give it for a measurement, the higher the source's output impedance can be. Adding a small capacitor will also let the ADC work properly with higher impedances.
The recommendation is as it may be, but doesn't mean measurements will fail if done properly, taking the source's impedance into account.
So, my answer would be that it is working properly because you give the ADC plenty of time for a measurement, and/or you have added a capacitor on its input.
Here's probably what made it work fine: -
The text says that
capacitors must be added and presuming that you have, then the source impedance required by the ADC will become more relaxed. Pretty much most ADCs of this type can tolerate several tens of kohm source impedance if the capacitors fitted are circa 100 nF. There should be details in the MCU data sheet that explain this.
If you haven't used capacitors then you should.
Regards the input leakage current of the MCU, the maximum value is 500 nA and this will create an offset error (due to the 32 kohm source resistance) of 16 mV. If you can live with this (it affects all readings a constant amount) then that's fine. If you find that this DC error is problematic then you might need to buffer the signal with an op-amp. At normal operating temperatures the leakage will be about one-tenth of 500 nA so, is a 1.6 mV DC offset a big deal to you?