I want to hold a microchip(could be Arduino, might not be) in reset if the power (12V battery) feeding it's 5V regulator drops below 7V. If the power recovers, I'd like the reset to end at a higher level, let's say 8V. So there should be some hysteresis. This is for a battery powered telemetry system, so it needs to function autonomously.
On chip brownout detection isn't the answer, because the upstream regulator isn't dropping to the internal threshold, it's just becoming 'flakey', and peripheral devices aren't behaving, although the micro is performing it's function.
Any circuit ideas? Available functionality? Off the shelf solutions?
Thanks
You will want to look very carefully at that chip's specs.
The ATmega series (e.g. ATmega328) consumes several mA (6 mA, IIRC) while in reset, so it is MUCH better to put it into power down sleep mode, where it, if configured correctly, consumes less than 1 uA.
A reset controller might still be needed, if the battery dies completely.
This is the problem, indeed. The systems can be remote, and discharge is guaranteed in certain circumstances. However, by that point, most of the unwanted behaviour has died away. We'll look at sleep modes.
Thanks
Putting the processor into reset, or sleep, still doesn't turn off the peripheral sensors/devices in your circuit. Perhaps one of those power management ICs could instead turn off a mosfet, which would power down the entire circuit, except the PM IC itself.
Edit: It looks like the TL7709A, at 7.6V, would work, but it needs 2mA supply current, which is a lot. I don't know what other off-the-shelf options exist, but it seems you could rig up a comparator with a voltage reference at one input, and a resistor divider off the battery at the other input, with some positive feedback to provide hysteresis, and that might work well enough. The open-drain comparator output could connect directly to the gate of a P-channel mosfet that supplies battery power to the circuit. But perhaps there is a lower-power version of the TL7709A from another manufacturer.
Not too worried about a few mA, the system is a biggie. the TL7709A or something similar may suffice to drop power to the whole package. However, it seems to be setup to monitor a 9V package, whereas we're looking at 12V. Have to read up, see if the 9 is important, or if it's just "at least 9". The 7-ish threshold is what we want.
Thanks!
At Digikey these are shown as Marketplace items. Digikey doesn't have any in stock. But Rochester appears to have a whole bunch in stock. But you would need to check the hysteresis. It's only 20mV, which may not be enough. But it seems you might be able to add some positive feedback.
Yeah, the hysteresis is a bit tight, esp. if it's monitoring the input to the regulator, but the source is a battery so line noise isn't an issue, only noise due to load variation. I've passed the idea on to he who would make the decision(i.e. order the part). We'll see what he wants to do.
Thanks!
If the micro works well (as it should with BOD enabled) why not simply measure battery voltage and turn off (or simply ignore) the flaky peripherals at low voltage?
I suspect what's happening is that the voltage regulator is 'flickering', so the customer isn't impressed with the symptoms, but I'm two levels removed from the customer right now. Personally, yeah, I'd implement a shutdown at 7.5 volts and do nothing regardless of conditions until BOD, or recovery above 8V. I'm not really sure why they haven't done this already.
All TBD, I may get to meet with them this evening, and all might get explained. But I want to go in with options to explore, not a puzzled look on my face.
I think the big risk would be that when you put the processor to sleep, load current would be reduced, and battery voltage would recover to some extent. So it might oscillate.
Well, if it becomes important, it looks like it would be possible to increase the hysteresis of the TL7709A on the battery recovery side by adding two resistors and a diode. So once the battery voltage has gone below the 7.6V, and the reset is triggered, it would have to go back above 8V, or more, before the reset would be released. If you're interested, I could draw that up.
For the moment, let's hold off on that. Last night, the path forward became
monitor the regulator input voltage in a watchdog loop, put everything in a safe state and send a warning message at a 7.5VDC threshold
Do not resume normal operations until the voltage is above 8.5 VDC; reset what needs resetting on resumption.
Enable the BOR (brown out reset) at 4V(processor voltage).
Start the monitoring loop at startup.
This will be what the customer gets on this release. If it proves unsatisfactory, then the project will revisit the issue. I'm confident that this will achieve what is wanted, although the two voltage levels may change.
Thank you, everyone, for the guidance.