10
\$\begingroup\$

I'm working on a soft power switch for a microcontroller where a momentary switch can turn the circuit on (including microcontroller), and then when the switch is pushed a second time, the microcontroller can shut itself off after performing some clean up.

circuit drawing

I have the above circuit so far, but I'm not sure if it'll be reliable. I'm using a lithium-ion battery (3.7-4.2V) and the TC1015 regulator (3.0V output). The idea is that when the switch is pressed, the regulator turns on, then the microcontroller sets uC Power high, keeping itself on. When the switch is pressed a second time, an interrupt on uC Switch will allow the microcontroller to set uC Power low, turning itself off.

What I'm not sure about, is if I need to protect the microcontroller from battery voltage. The microcontroller I'm using has an absolute maximum voltage on the I/O pins of Vdd+0.4V, so I'm not sure how to handle that best.

Second, will this circuit actually keep the regulator from turning on when it's in the "off" state? I had thought about using a pulldown resistor on the enable line, but am worried about the current draw while the chip is powered on.

Edit: The microcontroller is the primary load that will be switched, so putting it into a low power mode unfortunately won't work here.

Edit #2 (After answers were posted):

I ended up using the circuit below:

fixed circuit diagram

The previously posted circuit didn't work very well, and had issues with a floating enable line when the microcontroller wasn't powering it.

The new circuit uses a flip flop, with the data line normally pulled low. Pressing the switch hits the clock, turning the system on. Subsequent presses of the switch drive the CLOCK line high (allowing the microcontroller to sense the press), but don't affect the output of the regulator. Once the microcontroller is ready to power off, it sets the DATA line high and then sets the CLOCK line high, which will cause the regulator to shut down.

One of the really nice things about this setup, is that the first button press turns the regulator on, and keeps it on until the microcontroller is ready to shut down. Bounce isn't an issue, because no matter how many times the clock line goes high, the data line is still held low by the pull down. In addition, the current draw should be very minimal (just the flip flop and the TC1015 while off), and there is minimal current draw through resistors while on.

The microcontroller does need to be protected from the battery voltage on the clock line, but as @Andy aka suggested, that can be done with a resistor on CLOCK.

\$\endgroup\$
0

4 Answers 4

4
\$\begingroup\$

R1 and R2 will limit the current into pins on your uC and this is usually sufficient to protect your device - you just need to check in the spec what that "limit" current is and choose a resistor value that is appropriate given that the uC supply may be at 0V (un powered). The zeners can be left out on this basis.

Reliability is another issue. Switch bounce may cause your uC to switch on then switch off a few times so write your code to be aware of this.

I think it may be advisable to have a resistor on enable but probably in the region of +10K and maybe this could be higher possibly 100k.

The voltage on the shutdown pin has to be at least 45% of Vin so this shouldn't be an issue.

\$\endgroup\$
3
\$\begingroup\$

This seems like a self latching system, which in theory should work, like a self latching relay circuit would (A button is used to turn on the relay, and then because the load pin and the coil pin are tied together, the relay stays on as long as power goes to the load pins).

To test it, without risking a microcontoller, you can do this. Add a dummy load to keep the regulator happy (a few leds, also so you can see that it is working), and then tie the output to the point where it is marked uC Power. After you press the switch, the regulator should start, turning on the leds and the uC Power, which in turn should keep the Enable pin at logic high (Shutdown Logic High is minimum 45% of VIN, so 1.89V at 4.2V In.).

So if you press the button and the leds stay on after you release it, it works. If it doesn't, it won't work as is.

Warning: I say that, not being sure how the Zener diodes will cause the circuit to react though.

\$\endgroup\$
0
\$\begingroup\$

The circuit linking the battery, microcontroller signals and the SHDN* input of the regulator (renamed EN) seems dodgy.

How about using a latch (running on battery power) to capture the closure of the key. Then the output of this latch can be OR'ed with a signal signal from the MCU to drive the SHDN* pin of the regulator (renamed to EN in the schematic). Upon booting, the MCU should drive its enable line first, then clear the latch, thereby ensuring that EN stays asserted.

The subsequent button action can be monitored through the latch: if the switch is pressed again, the latch goes high again. The MCU notices this and clears both the latch and its enable signal, triggering shutdown. Since the switch is latched, the MCU can monitor this very conveniently by simply polling it at a leisurely frequency.

One more optional refinement would be a circuit to ensure that when the battery itself comes online, the system powers up without the use of the button. This could be some kind of pulse which sets the latch.

\$\endgroup\$
0
\$\begingroup\$

Pull EN low with a suitable resistor and tie it to an io pin on the mcu. The switch goes on an input pin of the mcu. Pushing the button triggers an interrupt on the mcu input pin which toggles the output pin controlling the EN pin of your LDO.

Putting the mcu into deep sleep will allow the pin to be pulled low and disable the LDO. Triggering an interrupt with the switch will wake it up, pulling the pin high again and turning on the enabling the LDO again.

\$\endgroup\$
1
  • \$\begingroup\$ Sorry, I should have specified a bit more clearly, the microcontroller is the primary load that the LDO will be switching, so leaving it in a low power state won't work. \$\endgroup\$ Apr 11, 2013 at 13:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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