I am looking into making a 3.3V MCU 5V-compatibletolerant and maybe protect it further. The Wikipedia article for TVS diodes has a circuit with combined zener/TVS and rail-to-rail clamping diodes. Simulating that, it does not seem like the upper diode is used at all - which makes sense, since the TVS keeps the voltage level below 3.3V + clamping diode drop. So, what is it good for?
If I add a 15mA PTC, would the simulate circuit with or without the diode to top rail be an appropriate setup to make a 3.3V MCU tolerant for 5V input and protect it from further mild abuse?
I put the smaller resistor and the Schottky in parallel to the 10k, since I might want to have the full output drive, while the capacitance of the pin is rather small and I do not want to drain so much current from a 5V input.
Sorry, read the circuit wrong, zener does not go to the pin. In that case, the voltage will be higher by the forward voltage of the clamping diode, which does not seem to be good design for connecting the teensy to a 5V input
An IO pin, of course. You can assume basic sanity with may question But sure, given regular questions from people who want to power something through a voltage divider, I see that that may be a valid question.
Actually. multiple IO pins. I also need the output capability, so a unidirectional level shifter is not an option, even if it was one that still safely read 3.3V as high.
I don't want to make it a 5V IO, but only 5V tolerant (just made that more clear in the initial posting). It should still have an output high level of 3.3V, but should be able to cleanly deal with 5V as input. The input could be a 5V open-drain network, so the input impedance should be as high as possible at 5V, though the pullup on that network will be quite strong (25mA+).
The application is for a trigger box for a camera system. The ports will usually be connected to 3.3V Aptina cameras for trigger out and exposure in signals, but might also be used on 5V open drain networks for Basler ace cameras. In that case, they will only have to serve as inputs. (There are additional pins with iso1540 for output capability there.)
They will also be used in case we want an external sync signal between multiple boxes, or an external trigger.
It keeps the input between the power rails and it dumps excess energy from the power rail if required, like if the input got hit by a big static zap from the user.
Yes, I got that. Before post #2, i got the circuit wrong and thought the Zener was in parallel with the clamping diode to ground. Hence the question.
The issue I see there is that it does NOT keep the voltage between the rails, but between the rail + the forward voltage of the diodes. At 3.3V, even Schottkys will exceed the absolute maximum ratings of a lot of devices, not to talk about the normal ratings. So it is not sufficient for making the device 5V-tolerant.
My simulation is essentially what the Ruggeduino does, so I guess it is fine. It is just not very high impedance at 5V (but sufficiently for the planned application).
The only alternative I can think of would be if someone could point me to a buffer that is
bidirectional
can be powered with 3.3V on BOTH rails (not the case for Grumpy_Mikes, as far as I see in the datasheet - min Vref_B is Vref_A+0.8V)
ElCaron:
The input could be a 5V open-drain network, so the input impedance should be as high as possible at 5V, though the pullup on that network will be quite strong (25mA+).
That scenario can be fixed with internal pull up on the 3.3volt pin, and a (schottky) diode in series with the input.
Anode to the 3.3volt pin, cathode to the (0-40volt) sensor.
Leo..
Wawa:
That scenario can be fixed with internal pull up on the 3.3volt pin, and a (schottky) diode in series with the input.
Anode to the 3.3volt pin, cathode to the (0-40volt) sensor.
Leo..
I see where you are getting it, but as I said, this is supposed to remain an IO. Is far as I can see, a pin with such a circuit could not switch anything on the outside low (when used as an output with other 3.3V stuff), because the the diode blocks current from flowing to the pin.
ElCaron:
The issue I see there is that it does NOT keep the voltage between the rails, but between the rail + the forward voltage of the diodes. At 3.3V, even Schottkys will exceed the absolute maximum ratings of a lot of devices, not to talk about the normal ratings. So it is not sufficient for making the device 5V-tolerant.
IMHO this is not true. Most CMOS inputs have internal "protection diodes" to Vcc and GND to keep input in the "safe area" from GND - 0.3V to Vcc +0.3V. If you apply higher voltage to the pin it means the diodes are conducting so much current they are going to break. Lot of simple ICs have stated "maxim clamping current". Knowing this you simply add current limiting resistor large enough and you have "high voltage tolerant" pin. Problem is many MCUs don't say what is safe diode clamping current. You may try it or add external diodes to be safe. External diodes will carry most of the clamping current protecting the internal diodes (voltage drop on diode grows with increasing current so if internal diode needs to carry "large" current its voltage drop will be higher than voltage drop of external diode that is rated for much higher current).
Well, that is essentially what I am saying. The question is rather if Vcc+(0.3V) is really a safe operation range for e.g. a Teensy 3.6, at which it should run in an intended operation case.
Why do you think this solution would be better than a Zener to ground that could place me at Vcc? In both cases, my input would only have limited impedance at 5V.
It depends what you need. Zener diodes are far from ideal - they have rated voltage only at specified current and temperature. To reliably protect the input pin you will get considerable current through the Zener even at voltage under 3V3. Also the Zener does not protect the pin when Vcc of the protected device is less than expected. OTOH the Zener is simpler and if you don't mind the other drawbacks...
I cannot say if Vcc+0.3 is absolutely 100% safe. From all I know it is safe and AFAIK it is widely used. But I am no EE...
But sure, given regular questions from people who want to power something through a voltage divider, I see that that may be a valid question.