I have a project with the following requirements:
* "Discrete Output" with high impedance in the OFF state and approximately 100 Ohms to ground in the ON state
* Maximum voltage on the "output" pin of 32.2V
* Withstand lightning waveforms of 600V and 300V in both polarities
* Very small enclosure which limits high wattage part options
* No high speed switching requirements
* High vibration environment (no relays)

I originally envisioned the attached circuit. There are two 50-Ohm 1W wire wound resistors to split the lightning event voltage for the TVS diode. The ON and OFF states would be controlled through a MOSFET with a 5V gate driver.

This circuit can handle the specific lightning events I am targeting, but the problem is the potentially continuous 32.2V applied to the output. When the MOSFET is ON, this would sink about 322mA. Each resistor would require 5W+ each. This is a small enclosure with space for only surface mount parts. I could handle about 1W wire-wound resistors max. The transistor must also be very small.

I'm at a loss for other solutions with these requirements. Am I missing something or is there a better method for this type of function?

 

Don't offhand know how to reduce the resistor power, but the two resistors should be in series before the TVS diode.
The resistor between the diode and MOSFET does nothing to limit the effects of the lightening event.

You may need to go with a TVS that can handle more power so you can reduce the value of the resistors.

 

Thanks, crutschow.

Here's the logic with the TVS diode between the two series resistors.

If it was configured how you're suggesting it with them placed before the TVS, when the FET is ON during a lightning event, the TVS would never get clamped as the cathode would be connected to ground through the FET. This would result in the FET sinking the full current of the lightning event only limited by the 100 Ohm resistors. None of the lightning energy would be absorbed with the TVS.

With the two resistors in series split by the TVS diode as shown, this creates a voltage divider for those lightning events and allows the TVS to clamp when the FET is on.

 

32 volts into 100 ohms is (32**2)/100 -> over 10 watts of power. I don't see how you can continuously present a
100 ohm load at 32 volts without having somewhere for the 10 watts of power to go.

 

This is my conclusion as well, Michael.

Just wondering if there's some clever ideas around here to help. Feeling like I'm going to need to push request a revision to the requirements.

Thanks for the input. It at least gives me some validation.

 

It all depends on how you interpret the design spec of "approximately 100 Ohms" without any tolerance.

I would go for a design of near 100 milli ohms max and see if I get that by the preliminary design review.

 

Are you expecting the lightning pulse to appear at the node marked 'output'? If not, then where? Lightning is rarely (if ever) predictable.

 

Alec, in this case, the "lightning" follows a standard and goes through a lab testing process. The waveforms are applied at the external connection (OUT).