1
\$\begingroup\$

Many PIC microcontrollers such as PIC24EP512GU810 have 5V tolerant pins.

My question is how does this work. Do these devices have clamping diodes for ESD protection?

Would clamping the input voltage to [3.3V + diode forward voltage] to interface a 12V signal be a good idea.

\$\endgroup\$
3
\$\begingroup\$

Not clamping diodes per se, but there will be some kind of ESD protection structure, one that does not kick in until you've exceeded the absolute maximum input voltage specification.

Yes, your clamp to Vdd should work fine, provided Vdd does not go out of spec because of the clamp current. It is better than a Zener (given the caveat) because it goes down with Vdd, also Zeners < 5V or so are really crummy.

Note that in general 5V tolerance may not be guaranteed if Vdd is 0V, which may be of concern since the 12V is from an external source. You'll be exceeding the 300mV maximum most likely, but I doubt that is a real problem, especially with limited current, but I wanted to flag it for evaluation and possible testing.

Edit: I looked at this particular chip, and Microchip has used a more reasonable abs max rating than on some other parts- the maximum input voltage is limited to 3.6V on 5V-tolerant inputs when Vdd < 3V, so it should be ok with the clamp diode, assuming the Vdd=2.999V condition isn't a problem! If you're paranoid, use a Schottky clamp such as BAT54C and keep the current low enough to guarantee the 0.3V drop but high enough that leakage isn't an issue.

enter image description here

Another approach is to make a special clamp supply bus using something like a TL431 shunt regulator and connect all your diode clamps to there. You're free to choose whatever voltage makes sense (via the two resistors that set the voltage) and the TL431 is guaranteed to be able to sink (at least momentarily) 100mA total. By separating the shunt regulator from the clamp diodes you're free to use very low capacitance fast switching diodes and still have hefty sink capability. With a TL431 and a BAT54S per pin you could easily handle transients of hundreds of volts using a 10K resistor.

\$\endgroup\$
  • \$\begingroup\$ No, BAT54C if you're clamping two inputs to a positive voltage and BAT54AS if you're clamping one input to high and low (eg. +V and ground). \$\endgroup\$ – Spehro Pefhany Jul 25 '14 at 18:38
  • \$\begingroup\$ Right. I was thinking of clamping both positive and negative voltages of a single input. \$\endgroup\$ – Hassan Nadeem Jul 25 '14 at 20:05
1
\$\begingroup\$

ESD protection and a 5V tolerant pin are different things.

ESD protection is achieved, as you say, with clamping diodes, while making an input 5V tolerant might be as easy as making the input mosfets gate thicker in order to withstand an higher \$V_{GS}\$. Since when the input is high, i.e. 5V, the upper protection diode would be conducting a series resistor is needed to not exceed its maximum current. Series resistors are anyway usually present, so that's not much of a problem. I believe a circuit including a diode and a zener diode with the cathode connected together might work too: from Vcc to the pin you have the normal diode reverse biased, while from the pin to Vcc conduction starts when \$V_{in}=V_\gamma+V_z\$. Tuning \$V_z\$ you can get your 5V or whatever you need.

Clamping is a good idea indeed, please be sure that the source of your signal can provide the necessary current: I'd use the classic resistor - zener diode topology, keep in mind the pin input current (very low) and the current needed to correctly bias your zener and choose R accordingly.

\$\endgroup\$

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.