Vtyjkyuk

I have a problem that I cannot solve and need help.

I have an analogue input on a microcontroller and I want to have multiple switches which the micro can see and act on depending on which switch was pressed.
I have 5V that the circuit works on. So I am trying to get different voltages from 0-5 into the analog input of the micro.

The catch however is that I want to control it with a p type transistor or MOSFET so that the user can just short something to ground to activate the divided voltage. I've sat here for days trying to figure this out using MOSFETs and transistors but nothing I do works! The divided values are never correct!

I've been playing with BC327's and BS250 MOSFETs to try and make something work, but everything I try fails.

I realize I need help, so I'm throwing this out there on the chance someone can help me!

Your circuit won't work when the transistor base is sinking current to ground. Re-position your resistor and try again.

^{simulate this circuit – Schematic created using CircuitLab}

Below are two methods. The simulated voltage for the BJT circuit is 2.497V, for the MOSFET circuit 2.500V.

With 3 base/gate resistors increased from 10K to 100K (R7 to 10K), the values are 2.495 and 2.500V.

^{simulate this circuit – Schematic created using CircuitLab}

There is no reason at all to use a transistor or MOSFET in a switch detection circuit like this. You can simply connect as follows:

This takes way less components and results in easy switch detection via the ADC reading. No switch pressed reads a nearly 5V value. S1 produces a voltage less than 2.5V, S2 produces a voltage of about 2.5V and S3 produces a voltage above 2.5V. Of course you are free to change the resistor values to suit your application. I just recommend keeping them in the Kohm range so that some input impedance at the ADC does not change the switched voltage dividers very much. Also keep in mind that when you detect each switch you need to check a range of the A/D readings for each to accommodate resistor tolerance and noise.

Edit: (in response to OPs query about long wires)

This should work fine with wiring to tact switches mounted on a circuit board. It can also work if the switches are wired remotely with wires but consider several things. First there can be noise on longer wires so make sure to select the resistors such that the voltage difference from switch to switch is much larger than the noise voltage. Also keep the impedance lower in Kohm range like I suggested. If you used Mohm range resistors the lines will be more susceptible to noise pickup.

With remotely mounted switches where wire may be very long or if there is concern with ESD then it would be advisable to not use this analog sorting for switch detection. Instead each switch should be dealt with in a digital manner. Digital treatment makes it more practical to:

1. Provide ESD clamping on the signals at edge of circuit board.


2. Buffer switch signals if necessary to protect sensitive MCU pins.


3. Debounce switch signals with 1's and 0's instead of varying analogue values.


4. Organize larger switch counts into a matrix to save GPIO pins.

Originally you said you wanted to use a PNP transistor and a few different buttons that short something to ground in order to produce an output of 0-5V to send into a single input of your microcontroller.

Here's one way to do that.

^{simulate this circuit – Schematic created using CircuitLab}

[the values shown are just a rough example, you should play around with them and see what works best for your application]

If you push each button one at a time, the output will change according to the resistance connected. Adjusting your resistor values adjusts the output. (This won't always work if you're planning on hitting two buttons simultaneously though since that would put the resistors in parallel.)

Alternatively, you could put Ro on the collector and take your output there, but you won't be able to use as wide a range of resistor values at the base (not sure how many different values you need at the output), and in that case R1/R2/R3's value will vary inversely to the output. With the configuration shown, the output will increase with their resistance and the change will be slower (you can do a DC sweep to observe this).

Of course you could skip the transistor and Ro altogether and take your output directly from the resistive divider.