I'm using a differential ACD from Microchip called MCP3424 and an ATMega16 to make various measurements.

Here is the schematic:


Note: I have changed the values of R22 and R23 to 1.5k. MOSFET Q7 is off.

Today the ADC suddenly has stopped working. The ADC still responds to my commands but does a very weird thing: Each time I write the config I actually read it back to compare it with an expected value to check for errors. The weird thing now is that the channel selection bits (C1 and C0) seem to be stuck low. No matter which combination I choose these bits just stay low but the other configuration bits are still perfectly working! I can't take measurements from the ADC either because the RDY bit always stay high (which means no new result).

I have the suspicion that the negative terminals (CH1- and CH2-) need to be connected to GND to ensure that the voltages on CH1-/CH2- never exceed the maximum value of Vss-0.4 however there are clamping diodes inside the ADC so I'm not entirely sure if that really caused the fault.


  • I have taken ESD precautions, so damage through ESD is very unlikely.
  • The I2C part on the ATMega16 works just fine. I have checked the signals with an oscilloscope.

1 Answer 1


Ferrite beads are decent inductors at low frequencies. You have a FB in series with the power feed to your ADC. On the power pins of the ADC you have a 100 nF capacitor. Now, if the input power were raised very quickly (or shorted out i.e. fell very quickly) you'd create a substantial overshoot on the power pins that might easily damage your ADC. Maybe something like this: -

enter image description here

For a step input of 1V the output may "ring" considerably and can quite often reach peaks that are 2x the supply voltage.

I'm not saying it definitely is this but I'd be concerned.

  • \$\begingroup\$ Where did you find the induction value for the BLM? \$\endgroup\$
    – Marco
    Apr 27, 2016 at 18:11
  • \$\begingroup\$ It's inductance is unknown but relevant and fairly high Q in the MHz region. That's experience talking but remember it could be something else that caused the problem. Can you tolerate a 10 ohm resistor across it to lower the Q? \$\endgroup\$
    – Andy aka
    Apr 27, 2016 at 19:07

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.