1
\$\begingroup\$

I want to stop supplying power for a short moment if a watchdog has not received a reset order for a while.

Shutting down power is important as this is the only way for some buggy device to function again.

More details: Buggy device run on 5V. GPIO voltage can be 5V or 3.3V. In normal operation the whole circuit is consuming about 0.5A sometime peaks overs 2.5A for a really short moment.

We can consider that the GPIO can freeze and therefore stay in +5V position.

\$\endgroup\$
  • \$\begingroup\$ What voltage does said buggy device run on, and what voltage is the GPIO running at? Also, how much current does said buggy device draw? \$\endgroup\$ – ThreePhaseEel Sep 15 '16 at 23:44
  • \$\begingroup\$ The term you are looking for is supervisor ic or reset ic \$\endgroup\$ – Axis Sep 15 '16 at 23:50
  • \$\begingroup\$ Yeah -- depending on his voltage rails, it may be possible to do this with nothing more than a SVS chip, a PMOS, and a few ceramic caps. \$\endgroup\$ – ThreePhaseEel Sep 15 '16 at 23:55
  • 1
    \$\begingroup\$ Is the GPIO on the device that may freeze up, or on some other (reliable) device that is monitoring it? \$\endgroup\$ – Bruce Abbott Sep 16 '16 at 0:08
1
\$\begingroup\$

You can purchase supervisory chips that will provide a reset pulse if the WDT input has not been toggled in a certain length of time. For example, the Analog Devices ADM83xx series, but there are many others. The WDT will provide the pulse if the input gets stuck high or low, or simply fails to respond in time.

You could also design a circuit that has that function, but the integrated chips are very small and not unreasonably expensive.

Use a parametric search and look at the datasheets to find a suitable product if you want a fully integrated version. The general category is supervisory chips, but not all incorporate a watchdog. You can probably find windowed watchdogs too which will reset the system if the pulses come too frequently as well as too infrequently.

enter image description here

If the reset pulse that the device provides (the above type has a number of options for pulse width) is not sufficient, you could stretch it with a monostable multivibrator chip.

Interrupting the power to the subsystem is then merely a matter of connecting a suitable high-side power switch to the /RESET output. You may need to control the ramp-up current to keep from causing a glitch on the supply due to decoupling capacitance on the switched side, and/or you may need to actively discharge said decoupling capacitance to positively reset the circuit within the available time.

Needless to say, this is a nasty bandaid fix and it would be better to solve the root cause first.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.