Foolproof input protection on µC pin (up to 150V AC)

[PeterFW]

Hello!
What would be the simplest solution to ruggedize the input pin of a microcontroller against damage with repeated voltages up to 150V AC?
The 150V AC carry allmost no current, they are the output of a piezo element.



This is the best i could come up with, it works. But id like to ask, if there is a simpler way to do this.

Current limit resistors on the input/output side to protect in case of a short circuit.
Diodes to protect against reverse voltage/feeding and rectification.

Then there is C1 and R1, they form a envelope detector for some smoothing, without them i had a bit of a problem with multiple triggering since i do not want to use software debounce on that single pin, and R1 serves as a pull down for the gate.
D2 is a zener to limit the voltage on the gate.
Then there are D3 and R4 wich are used for a internal, low voltage trigger input wich needs to be inverted.

I would like to have the option to connect either an external voltage source as a trigger, the piezo element or a simple switch.

[wolfield]

What about some TVS diodes, huh?
For instance: http://www.semtech.com/images/datasheet/srv05-4.pdf

[dannyf]

The solution depends on what source it is and what you are measuring. But generally, a large enough resistor will do for slow-moving signal from a low-impedance source.

[Ian.M]

The 100K input resistor + 1K pulldown looks like a good way of making the input sufficiently insensitive that it may not turn on the BSS123 as it limits the maximum possible gate voltage to approx 1.5V with continuous 150V DC in and the BSS123 is specced as having a gate threshold between 1V and 2.8V

[ConKbot]

Maybe not simpler,  but Ive used a Jfet current source (actually, it was a depletion mode mosfet, but it behaves about the same, so its a direct substitution, http://www2.eng.cam.ac.uk/~dmh/ptialcd/csink/csink.htm ) to limit the current into a optocoupler and a series diode to protect against reverse voltage input, so it came on at 2.2v and was protected all the way up to ~250V (it was a 350V mosfet, but I ran out of bench supplies  and 9v batteries to put in series) It happily took 120VAC mains for a torture test, and 1kV isolation from input to output (checked with a megger)

This was for a trigger input on a DAQ, so we could take trigger sources from just about anything, and even prevent stuff from getting blown out if it accidentally saw mains.  The trigger source wasnt our equipment, so protection from someone else's oops was definitely warranted. Best part was it all fit in a plastic backshell for our standard connectors, going from a BNC input, to the 7 pin input on our DAQ. 

Depending on frequencies, a 1mA 350V depletion mode mosefet current source, and a zener/tranzorb/voltage clamp of choice could stand up to a lot of naughty things on the input, only 4 components.

[Ian.M]

To save a lot of part hunting, can you disclose which MOSFET your design actually used?

[mikeselectricstuff]

All you need is the 100K resistor. For belt & braces, shottky diodes to supply & ground to stop body diodes conducting.

Remember it's not the voltage that kills, it's the current.

[lewis]

I've used the following successfully in commercial products:

470K series resistor with 4.7V zener down to ground. 10-100nF cap in parallel with zener takes the edge of high dv/dt transients and helps with EMC immunity. The higher the better, but reduces responsiveness.

Or 2x 220K in series if transient peaks are likely to exceed 300-400V. (Or 3x 180K for 1kV pk. immunity, etc. Use the peak voltage rating of the resistors and relevant creepage/clearance requirements [www.creepage.com] to determine how many in series are required.)

2.7V zener for 3.3V supply.

MCU's internal pullup can be used to tie the pin to a known state in the absence of the input to save cost.

Use an analogue/schmitt-trigger pin.

[ConKbot]

To save a lot of part hunting, can you disclose which MOSFET your design actually used?

dont know off the top of my head, but digikey and mouser both have depletion mode mosfets stocked, just not categorized well, but searching does turn them up.  The part i used was an AO semi part, but IXYS looks to have a decent selection stocked/available.

As others are saying, resistors are a good go to, as they are cheap, even a $1 mosfet is expensive compared to 3-4 resistors, plus the footprint can be similar.