How to use a brownout monitor to cycle the same rail it is monitoring?

[gaddison]

Hi all,

I'm using a TI TPS386000 integrated supervisor / watchdog part in a design I'm working on at the moment, which is great for sequencing the rails at power on, i.e. power is applied, once it is within an acceptable window, a switching regulator is enabled, once the output of that is above a certain threshold a downstream LDO gets enabled and so on, all with a programmable delay to avoid large inrush currents, which is all well and good.

My confusion is during regular operation if, for example, the output of the switching regulator goes out of bounds, I would like to cycle the enable pin of that regulator itself. But I can't just use a regular voltage supervisor to achieve that, because when you first apply power, the regulator output voltage will be below the acceptable threshold, thus the enable input will be held low by the supervisor and the regulator will never be enabled in the first place.

I've been trying to come up with some convoluted analogue delay circuits to achieve what I want, but it feels as though it's all getting a bit messy and needlessly complicated. Am I missing something simple here? Is there a standard way of achieving this behavior, or am I barking up the wrong tree completely?

Any thoughts would be appreciated,

George

[ajb]

You would basically need a circuit that produces a pulse triggered by the falling edge of the supervisor's reset output.  Here's an example of a very simple circuit that does that for a rising edge.  Switching the pulldown resistor R2 to a pullup resistor will convert it to trigger on falling edges: https://electronics.stackexchange.com/questions/461985/rc-positive-edge-detector.  The logic gate can be anything, but should have a schmitt trigger input to avoid output oscillation. 

If you need to do this across multiple channels, or have more complicated interlocking logic, a small MCU or Greenpak device would simplify the hardware (possibly even replace the supervisor entirely), albeit at some additional development expense.

[nctnico]

I'd be carefull with replacing the supervisor alltogether even though using a microcontroller to do complicated power sequencing could be a good option. Designing a reliable power-on reset circuit is an art in itself and not all microcontrollers have one onboard that is reliable. I'd at least start with keeping the option open in the design to place a power-on reset chip.

[gaddison]

Thank you both for the input, perhaps I am over-thinking this - I ideally wanted to have quite tight control over the reset timings which an RC circuit might not be the best for. But to be honest, I suppose the duration of the reset pulse itself isn't all that critical in the scheme of things. And a schmitt trigger on the output guarantees a nice clean edged signal. I also wanted some sort of glitch filter on the input, the ability to ignore voltage out of bounds excursions for under a few microseconds, which again is difficult to do accurately across a wide temperature range with an RC circuit. But I can select a supervisor IC with internal glitch filter, then it's just an inverter and a few passives which seems like probably the simplest solution.

I had considered a small dedicated MCU (or those GreenPAK devices, I haven't come across them before, they look really interesting), but it is just one more layer of complexity, another device to keep on top of software for, and realistically I don't have the time to put much dedicated design effort into it, so it's probably going to end up less reliable than an off the shelf solution.