microcontrollerreedswitches
I'm a newbie in electronics.
I need to build an ultra low power (nA) reed switch signal reading for an mcu.
I'm using the following schematic diagram.
The Vo output goes on a mcu input pin that is a Schmitt Trigger.
Is there any chance this to work ?
Sounds like capacitively coupling it might work well. Reed switch goes from power a pull-down resistor and one side of the cap. The other side of the cap goes to a pull-up resistor and the GPIO pin. When the reed switch is closed, both sides of the cap will sit on the power rail. When you open the switch, the pull down resistor will pull that side of the cap down. If the cap is large enough and the resistors are properly sized, the GPIO pin will be pulled down as well until the cap can charge up. When the switch is closed, the capacitor will discharge. It might be a good idea to add a clamp diode to power on the GPIO pin side of the cap so the cap can discharge through the diode instead of the chip's ESD protection diodes. You'll need to play around with the values of the resistors and capacitor to get a pulse of the right length, though the pull-down will have to be significantly smaller than the pull-up - I would suggest trying a 10k pull-down with a 100k pull-up. To figure out about how long the pulse will be, calculate the time constant, tau = RC. This will be approximately how long the pulse will be. A 10k and 100k resistor with a 0.1 uF capacitor will give you a tau of about 110e3 * 0.1e-6 = 0.011 seconds.
With reed activations for every meter traveled and 200Hz sampling rate, you're effectively limited to 10 m/s in terms of speed you can reliably measure (if you have achieved a perfect 5ms pulse time from the reed). In non-ideal conditions, you should expect reliable operation at about a half of that speed, 5m/s or 18 km/h. Assuming a bicycle application, that's not much.
My point is, you should either increase the sampling rate or reduce the number of magnets you use. The latter is the easiest option, with 2 magnets you'll be able to measure speeds up to 90 km/h, which looks reasonable.
And here's the RC circuit Steve told you about in his comment. It has a time constant of 5ms, so you'll measure between 3 and 1V in the sample following a reed activation, and less that 1V in subsequent samples.
simulate this circuit – Schematic created using CircuitLab
Best Answer
What sort of response time do you need when the switch changes state? If the switch will often sit closed for extended periods of time and instant response when it opens isn't required, having a pull-up which is enabled when the CPU is interested in the pin state but disabled otherwise may be more efficient than having a weak pullup which is enabled all the time. If there is a source of periodic pulses, it may be helpful to use something like the following:
simulate this circuit – Schematic created using CircuitLab
Assume the "clock" outputs a 10us pulse, ten times per second. When the switch is closed, the circuit will consume 1mA, 1/10,000 of the time, which is to say that the switch will consume 1uA on average. Once the switch is open, PinState will go high the next time the clock outputs a pulse, or whenever the CPU outputs PullupReq; it will then remain high until the switch is closed. Note that if the CPU uses "wakeup on pin change" it will remain undisturbed while the switch sits closed, or while it sits open.
The only disadvantages of this approach are that unless CPU explicitly turns on PullupReq, it may take up to 100ms to find out when the switch is open, and it requires a source of pulses for the OR gate. Otherwise, it offers a pretty good trade-off of responsiveness vs power consumption, and--unlike high value resistors, will work even if the switch has some leakage resistance.