Zenner diode selection for protection not regulation?

[jwhitmore]

There's a wealth of info out there on choosing the right Zener diode for voltage regulation. Calculations deal with the amount of current that your circuit needs, along with the input voltage and the required output voltage. So given two voltages and a required current you can find the value for a current limiting resistor. Like this question https://electronics.stackexchange.com/questions/28944/selecting-correct-zener-diode  I only have the second edition of The Art of electronics but that book didn't deal with protection, only regulation. Perhaps the 3rd edition has expanded to answer those questions, and I should put it on the shopping list.

In the mean time, unfortunately I'm trying to use the zener diode for protection of a micro-controller's GPIO (3v3) Pin. I've tried this in the past and I've always struggled with it. I'm using a OneWire bus. The OneWire bus is pulled up with a 5k1 resistor to the 3v3 supply rail. Then there's a 12V rail which must be connected to the bus for the program pulse of the device. So the uC is disconnected briefly from the OneWire bus and then the bus is connected to the 12V programming pulse. Once finished the 12V is disconnected and a short time later the uC is reconnected to the bus.

I'm struggling to understand the results I'm getting from the zener diodes I'm trying. As I say I've got a current limiting resistor, which is taken off the bus as I don't want to interfere with bus timings. I got a number of zener diodes thinking I'd surely find one that works for me. Even before I try the programming pulse the 3v3 zenner diode is pulling the bus down to 2.08V? I'd have thought that when a 3v3 Zener diode is connected to a 3v3 bus then the zener would be doing nothing, as in not conducting. I've tried 4 zeners of different values and even the 4.7V zener diode is pulling the 3v3 bus down to 3v before there's a pulse on the bus. That is not what I'd expected at all. I can't square that information with what I'm reading about Zener diodes.

If anybody could shed some light on what I've done wrong or what I'm misunderstanding I'd be really grateful. I'm trying to get past this OneWire zener problem so that I can then use a suitable zener to protect an uC ADC. In that case there'll be no current entering a GPIO so it might be easier.

Diode	Zener Voltage	Power	Idle Bus Voltage	pulse peak on reconnect
MMBZ5226B	3.3V	350mW	2.08V	3.24V
MMBZ5227BLT1G	3.6V	225mW	2.12V	3.44V
MMBZ6228BLT1G	3.9V	225mW	2.36V	3.68V
MMBZ5230B	4.7V	350mW	3.00V	4.4V

[Benta]

Zener diodes have a "soft" breakover voltage until you reach a certain current through them, usually in the mA range.

A better approach to input protection is usually using diodes to deflect input transients onto the supply rail. For lowest voltage drop, Schottkys are often preferred.

[JS]

TVS is your search term, you can get small packages with many of them designed for that purpose. They usually contain a single zener and diodes for each channel to clamp to the zener or ground. Also, is a nice practice to leave some resistance from the protection device to the µC so you can protect a bit over the rail but the current is limited to the pin.

JS

[Ian.M]

Its likely to be difficult to find a TVS diode that has a closely enough controlled breakdown voltage to protect a low impedance 3.3V I/O.

If Schottky clamping to the MCU 3.3V supply rail is unacceptable, its probably easier to use a TL431 shunt regulator set to about 3.1V or 3.2V, biassed at 1mA with a pulllup resistor to the +12V rail, then Schottky clamp to that, which will give you a very predictable clamping voltage, and a clamp that can handle nearly 100mA without issues.  Put 10uF + 100nF ceramic across the TL431 to stabilise the clamping voltage during fast transients.

[JS]

Its likely to be difficult to find a TVS diode that has a closely enough controlled breakdown voltage to protect a low impedance 3.3V I/O.

If Schottky clamping to the MCU 3.3V supply rail is unacceptable, its probably easier to use a TL431 shunt regulator set to about 3.1V or 3.2V, biassed at 1mA with a pulllup resistor to the +12V rail, then Schottky clamp to that, which will give you a very predictable clamping voltage, and a clamp that can handle nearly 100mA without issues.  Put 10uF + 100nF ceramic across the TL431 to stabilise the clamping voltage during fast transients.

  It's true that protecting a low impedance I/O reliably is quite hard, but for a very reliable protection optocouplers should be used, for a bidirectional bus with pullup 2 optocouplers and a few resistors are needed.

  There are some TVS which can be biased like this one, so the zener is ready for when the spike comes in.
https://www.semtech.com/uploads/documents/rclamp2504n_rclamp3304n.pdf

JS

[jwhitmore]

Thank you all for your responses, especially the info on the TL431. That looks really interesting and just the job. It does feature a 80 odd page datasheet so I've a bit of reading to do. I totally agree that Optocouplers would be a great solution, but I've an eye on using Infineon's Profet[1] and their Current sense pin (Is), which gives a measure of the output current. The circuit for that current sense features another Zenner diode to protect the ADC of the uC. I'll read the TL431 and order a few up to test.

Thanks so much again.

[1] https://www.infineon.com/dgdl/Infineon-BTS7004-1EPP-DS-v01_00-EN.pdf?fileId=5546d4626102d35a016147550a725555

[Ian.M]

There's also the TLV431 shunt regulator for when you need to go lower than the 2.5V reference voltage of a TL431 permits, or need a lower bias current, but it has the disadvantage of only being able to handle 20mA (abs.max).   Its reference voltage is 1.25V.