Voltage spike on RESET when Serial port is closed

[matthijskooijman]

When the serial port is closed, the DTR pin goes from low to high. But since the capacitor in the reset line is already charged by this point, the voltage on the RESET pin becomes 2xVcc (DTR + capacitor charge) for a short while (until the capacitor discharges through the internal pullup of the RESET pin). I haven't verified this behaviour with a scope, but if it indeed happens, we should probably fix it, since it is out-of-spec behaviour (unless the reset pin is somehow more resistent to high voltages, which might actually be the case given there is something like high-voltage programming which might work on the RESET pin. Still, might be better to fix anyway - the voltage spike might also be visible on the reset pin on the header.

It seems that the Arduino Uno has a diode from RESET to VCC, which looks like it is essentially intended to kill this voltage spike.

[amcjen]

I just tested this on a scope, and I'm getting about 1.5 x Vcc for about 100µs, with a clean slope from max of 1.5xVcc sloping down to about 1.25xVcc after 50µs, and down to Vcc at 100µs.

I re-reviewed the Uno schematics, and I don't see a diode in their reset circuit. Am I missing it? http://arduino.cc/en/uploads/Main/arduino-uno-schematic.pdf

[matthijskooijman]

Seems they added it in a later revision. I was looking at the Uno rev3 schematic: http://arduino.cc/en/uploads/Main/Arduino_Uno_Rev3-schematic.pdf

[amcjen]

Ah, gotcha. Yep, seems like a prudent addition.

[amcjen]

It looks like the later change to the Uno board also had an external pull-up on the reset line, of a 10kΩ pullup. I supposed we could add an explicit one too, and not rely on the internal RESET pullup?

Found the same spec diode as the Uno one, but in a smaller package (http://datasheet.octopart.com/CD0603-S01575-Bourns-datasheet-8529516.pdf)

I can add this in parallel with the 10kΩ to solidify the reset circuit.