I recently used an accelerometer (ADXL335) in my academic project. This sensor has an analog output.

I interfaced this sensor with a microcontroller (PIC18F458) using an ADC module. It was successfully interfaced and showed the correct readings.

Later I noticed in the datasheet of the controller that "the maximum recommended impedance for analog sources is 2.5 kΩ" and the datasheet of the ADXL335 shows that its output impedance is 32 kΩ. I have not used any signal conditioning circuit between the sensor and the controller.

I wonder why I am getting the correct output, even though the source impedance requirement for the ADC module of the controller is not satisfied.

  • \$\begingroup\$ How often do you take a measurement, or, in other words, how much time are you giving the ADC for each measurement? \$\endgroup\$
    – ocrdu
    Oct 12, 2020 at 10:50
  • \$\begingroup\$ Data sheet links and schematics please. \$\endgroup\$
    – Andy aka
    Oct 12, 2020 at 10:56
  • 3
    \$\begingroup\$ If you have the datasheet-suggested 100nF capacitors on the accelerometer outputs, then the impedance is not 32kohms but much lower. \$\endgroup\$
    – Justme
    Oct 12, 2020 at 11:11

2 Answers 2


Here's probably what made it work fine: -

enter image description here

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?

  • \$\begingroup\$ I haven't used capacitors also still getting output (I agree it is not 100% correct) and that's why I am surprised \$\endgroup\$
    – Ank09
    Oct 12, 2020 at 12:16
  • \$\begingroup\$ @AKumar Well, if you are not sampling at a high rate you might get away with it but, it's likely that there are signal amplitude issues that you are just not aware of but, your "application" is OK in what it does. \$\endgroup\$
    – Andy aka
    Oct 12, 2020 at 12:18
  • \$\begingroup\$ i am using it for tilt angle measurement \$\endgroup\$
    – Ank09
    Oct 12, 2020 at 12:22
  • \$\begingroup\$ Do you pre-calibrate your circuit at various angles before launching it into doing a real job? \$\endgroup\$
    – Andy aka
    Oct 12, 2020 at 12:30
  • 1
    \$\begingroup\$ Then you will have "calibrated out" static errors that could be due to ADC input pin bias currents and, providing the temperature doesn't increase or decrease much above or below the temperature when you calibrated it, it will continue to work. If you need extended working performance over a large temperature range then I would use an op-amp buffer (properly designed and not just any old op-amp circuit). \$\endgroup\$
    – Andy aka
    Oct 12, 2020 at 12:37

The source's output impedance (and capacitance) will affect the sampling 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 maximum impedance recommendation is as it may be, but ignoring it doesn't mean measurements will fail if done properly, taking the source's impedance into account.

It is probably working properly because you give the ADC plenty of time for a measurement, and/or because you have added a capacitor on its input.

  • \$\begingroup\$ I haven't added external capacitor but yes there was few seconds time between two readings. \$\endgroup\$
    – Ank09
    Oct 12, 2020 at 12:19
  • \$\begingroup\$ It's about how much time (how many clock cycles per sample) you allow the ADC before getting the measurement, not so much about the time between measurements. I don't know how this is set for your specific ADC module, or if it can be set at all, though. \$\endgroup\$
    – ocrdu
    Oct 15, 2020 at 11:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.