0
\$\begingroup\$

I'm using this IC (https://www2.mouser.com/datasheet/2/609/ADM1270-878589.pdf) to control the current and the voltage on some of the outputs of a PCB that controls the power to other PCBs, which have several sensitive components like microcontrollers and other semiconductors.

This IC controls the gate of a pmos. I'm thinking about using this MOSFET (https://www2.mouser.com/datasheet/2/389/sts10p4llf6-956791.pdf).

The IC gate current is very low (GATE Pull-Down (On) Current = 25 uA), and the MOSFET gate has a fair amount of charge (Total gate charge = 34nC), so I did the math: Q=it => t= 1.36 ms, which is the amount of time to switch the MOSFET without any resistor (I think...). If I use a resistance to limit the current the time to switch will increase even more.

My doubts are:

  • Are my assumptions/calculations correct?
  • How long do I have before my sensitive components get fried? Is 2 ms of delay on my protection IC good enough?

My objective is not only to protect singular components, but also to know which subsystems are functional and to be able to isolate the damage, if it can't be prevented completely.

(Context) I'm a Mechanical Engineering student doing an electronics project in an automotive context.

Thank you in advance for the help!

\$\endgroup\$
  • 1
    \$\begingroup\$ It's not clear that the things you are hoping this will protect against are the things it was designed to protect against - specifically it seems aimed more at loads that draw excessive current, than at externally introduced stresses that would typically damage things like microcontrollers. You probably should explain a bit about the adverse conditions you are trying to prevent. \$\endgroup\$ – Chris Stratton Oct 3 '18 at 17:49
  • \$\begingroup\$ Done! Thank you for the advice. I do think the IC fits my needs. Even if it only opens the circuit 2 ms after some fault, it's still better than any fast actin fuse i've seen since it can open as soon as it passes the treshold, while fuses take longer the closer to the treshold the fault is \$\endgroup\$ – David Ribeiro Oct 3 '18 at 18:27
  • \$\begingroup\$ You've still not stated a requirement for which this seems suitable. This is basically an overcurrent protection device. MCU's are not typically damaged by overcurrent, but rather by overvoltage. It's not to say it could have no benefit, but it would not really protect against the primary threats that typically exist to the type of parts you are trying to protect. \$\endgroup\$ – Chris Stratton Oct 3 '18 at 21:05
  • \$\begingroup\$ It also has overvoltage, undervoltage, and it's very flexible, I can pick my tresholds. Besides i can connect the output 'PowerGood' to the MCU and see the real time state of each system in my interface. I undertstand I'm focusing in CMOS components because i don't understand as much the standards of protection for this components but current protection is needed for purposes besides protecting the sensitive components. Meanwhile you helped me a little as i now understand that this components are more susceptible to overvoltage events. Thank you! \$\endgroup\$ – David Ribeiro Oct 4 '18 at 13:53
2
\$\begingroup\$

Check the datasheet of each component in your system and find the table of maximum allowable voltages and currents. If you exceed the maximum allowable ratings of any of your components then you must assume that they are instantly damaged. This damage may cause an obvious, immediate failure of your circuit or it may cause a failure a short time later.

\$\endgroup\$
0
\$\begingroup\$

Define all the automotive compliance specs you must meet and the mechanical and electrical energy being switched.

This gets reflected back when stopping and must get dumped somewhere.

Usually an active switch or power diode returns the energy stored back into Vbat with an ultra-low ESR cap shunted with the battery low ESR.

Long cables have an effective series inductance ESL typically 1uH/m which with low ESR and low RdsOn will cause ringing and thermal losses in FET. So define these and it can be modelled easily. TBD

Each component has surge energy limit defined in active parts by the Safe Operating area SOA chart and passive parts by the max ripple current.(rms)

Component temp rise depends on thermal Rja. [W/‘C] with some thermal time constant depending mass and to some extent thermal velocity.

If you don’t know already, E(L)=1/2LI^2 or E(C)=1/2 CV^2 and you should know the inertial energy stored.

\$\endgroup\$
0
\$\begingroup\$

Ok so the conclusion i reached with the help of the other answers/comments and more searching is that microcontrollers and other semiconductors are more susceptible to overvoltage.

Overvoltage failures typically happen due to a breakdown of an insulation layer and may be caused by external factors. This events are fast and additional protection like a TVS or a zener in parallel is more appropriate.

Overcurrent failure is usually caused by some internal failure (such as a fried capacitor turning into a short) and the best approach form the point of view of sensible components is:

  • To prevent it from happening by protecting/good design of those other components.
  • Good design around the pins, limiting inrush currents and over-stress through the use of a series resistor.

As for my calculations they are invalid since the current wont be constant, there will be a peak of inrush current to feed the GATE capacitance, meaning that it will be way faster, but on the other hand the current will reach a peak of 180mA assuming a trace resistance of 13mΩ (simulated on LTSpice). This value is way above the 25uA typical current but it will only last 1ns, so it should be fine. This current could be reduced with a series resistor, but the typical application circuit given in the datasheet doesn't use any series resistor, corroborating the hypotheses that this peak is fine.

\$\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.