0
\$\begingroup\$

Say you were wanting to measure a signal with a higher voltage than the ADC could handle. Is there some best practice or commonly used circuitry that goes beyond a simple voltage divider to protect the ADC?

Or is it common for professional designs to simply rely on a voltage divider?

\$\endgroup\$
  • 2
    \$\begingroup\$ Diode clamp to the rail. Still need the divider so the ADC can step down the voltage though so it can be read. Diode clamp just protects it if it goes over even what the divider steps it down to for some reason. You can also buffer the divider with a voltage follower so the divider impedance doesn't mess with the ADC. \$\endgroup\$ – DKNguyen Jan 29 at 4:10
  • \$\begingroup\$ electronics.stackexchange.com/questions/418499/… \$\endgroup\$ – DKNguyen Jan 29 at 4:13
  • \$\begingroup\$ Does a diode clamp affect the signal significantly? Or is its impact low enough that you would see diode clamping of some variety on most ADC circuits in real world products (where over-voltage could occur)? \$\endgroup\$ – zubin3333 Jan 29 at 4:21
  • \$\begingroup\$ When it's not clamping it should appear as a tiny capacitor to the rails. Every pin your MCU or ADC probably as diode clamps inside it for ESD protection but they aren't very hefty. You're already sticking a massive resistive divider across the signal and that would affect it much more. \$\endgroup\$ – DKNguyen Jan 29 at 4:23
3
\$\begingroup\$

schematic

simulate this circuit – Schematic created using CircuitLab

If you have need a resistive divider, it can limit current instead of an explicit current limiting resistor.

The diode rail clamps can only clamp to the rails if there is voltage on the rails. They won't offer protection if the ADC is unpowered. But zeners or TVSs which breakdown in reverse will provide protection when not powered (as well as powered, but they might not clamp to as constant voltage as the diode rail clamps.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Schottky will protect the rail better than a silicon diode. \$\endgroup\$ – mkeith Jan 29 at 4:30
  • 1
    \$\begingroup\$ @mkeith Indeed. I use schottkys when I do it but regular diodes work too. \$\endgroup\$ – DKNguyen Jan 29 at 4:38
1
\$\begingroup\$

There is actually a lot to consider here. The divider plus Schottky diode is pretty good protection. But there are two more things to consider. First is power sequencing and second is current drain.

Sometimes the high voltage may be present when the processor containing the ADC is inactive. To protect against this, you really want some type of high-side switch so the voltage divider does not get energized unless the ADC is energized. Otherwise you may have accidental power up of the microprocessor via the voltage divider (this is a bad thing).

Also, if the high voltage is coming from a battery, and you are trying to conserve power, you may want to be able to turn off the voltage divider to avoid consuming battery power.

schematic

simulate this circuit – Schematic created using CircuitLab

In this circuit, whenever VCC_ADC is high, the voltage divider will be energized. If, for any reason, VCC_ADC should go low, then M2 will turn off, and M1 will turn off, and the divider current will be zero.

If you want to control the divider with an IO pin, simply connect the gate of M2 to a microprocessor IO instead of directly to VCC_ADC. Then code will have to drive the IO pin high prior to sampling the divider voltage. BAT54 is a Schottky, but I didn't see a Schottky symbol in the library.

| improve this answer | |
\$\endgroup\$
  • 1
    \$\begingroup\$ I think the polarity of D1 should be reversed (Anode to ADC_IN) for the circuit to work as intended. \$\endgroup\$ – Peter Karlsen Jan 29 at 10:14
  • 1
    \$\begingroup\$ Yes, D1 is forward biased. V_ADC_IN is therefore erroneous. \$\endgroup\$ – stowoda Jan 29 at 13:45
  • \$\begingroup\$ Yeah D1 was backwards. Sorry. Fixed now. \$\endgroup\$ – mkeith Jan 30 at 7:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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