Powering off via UI while using a Mausberry Circuit

cyperghost

@caver01
So you soldering the SSR next to your switch. So you have two switches running.
I'm guessing the hardware switch is a momentary one - with normal state off?
So the SSR can also act as a momentary, unpowered via GPIO it is off, powerd via GPIO it is on....

So it think it should work ;)

caver01

@cyperghost said in Powering off via UI while using a Mausberry Circuit:

@caver01 But the mausberry works with permanent switches?
But I think the real button won't work anymore ;(

So use a GPIO to set event on/off ;)

Yep. use it to trigger an actual (parallel) button press.

What I can't figure out is why, even after the pi shuts down with a soft shutdown, the mausberry ignores my button presses. It is like it locks into a powered state. It must be checking the state of Pin 2:

If Pin2 = 0 and power button is closed, DO POWER ON
if Pin 2 = 1 and power button is closed, DO POWER OFF

so, after a soft shutdown, the mausberry thinks I am trying to turn the pi back on, but it is already on, so it does nothing. It's like an anti-bounce state, the time right after a POWER ON, but before the script is running and can set Pin 2 = 1.

caver01

@cyperghost If I get one of these to test, I will need help. Which pin on the SSR goes to what, and which pin gets the resistor?

cyperghost

@caver01 page 7 shows connection diagramm
Pin 1 goes to positive value (= GPIO Pin), Pin 2 is ground, set a resistor with right value (330ohm for 3.3V).

Pin 3 and 4 is the switch ... Connection does not matter here see page 4, it is independent if you switch the ground or positive value.

caver01

@cyperghost Looks like the diagram has the resistor on Pin 1. Does it matter?

cyperghost

@caver01 No but the resistor has to be placed somewhere in the line (+) LED (-)
So (+) R LED (-) or (+) LED R (-) doesn't matter

caver01

@cyperghost Oh I get it. . . because the device itself uses an LED internally for actuation.

Hey, thanks for your input here and the fun conversation about this. @madmodder123 I hope I didn't overstep by totally dominating your original post. I think we were on the same page with this, and it looks like a little testing may prove that while the Mausberry circuit is an excellent solution for an add-on power switch (especially when coupled with the updated script), it comes with an assumption that the button you add will be the only shutdown method used. It is not setup with awareness of additional soft shutdown procedures that aren't triggered by the button and the circuit itself.

cyperghost

@caver01 I'm pretty sure that a normally opened ssr works together with a momentary switch!

Annother suggestion. I think if we use a normally closed SSR and connect it to the $GPIOPin2 so there is no current as long as the Pie is active. But if is inactive it would close the reset pads of the mausberry... do you think that may work?

cyperghost

Some "improvments" to the original script! It might be a bit faster as there is no if-then-else....

#!/bin/bash
#this is the GPIO pin connected to the lead on switch labeled OUT
GPIOpin1=23

#this is the GPIO pin connected to the lead on switch labeled IN
GPIOpin2=24

echo "$GPIOpin1" > /sys/class/gpio/export
echo "in" > /sys/class/gpio/gpio$GPIOpin1/direction
echo "$GPIOpin2" > /sys/class/gpio/export
echo "out" > /sys/class/gpio/gpio$GPIOpin2/direction
echo "1" > /sys/class/gpio/gpio$GPIOpin2/value
power=$(cat /sys/class/gpio/gpio$GPIOpin1/value)

until [ $power = 1 ]; do
    power=$(cat /sys/class/gpio/gpio$GPIOpin1/value)
    sleep 1
done

[ $power = 1 ]  && echo "Shutdown now!" && poweroff

caver01

@cyperghost
Nice.

I used to use a python script that did edge detection. I was led to believe that keeping a python script active in the background would be even less of a CPU burden than a polling bash script. But I changed to the bash script you helped build when I saw the better commands. It would be nice to convert it to python.

cyperghost

@caver01 can you post python code please?
I think python is the better choice - if you enable IRQ mode, then it's the fastest alternative ;)
But the old bash script looked a bit weird so I made it a bit cleaner looking

caver01

@cyperghost I will post it later when I am back to my Pi.

cyperghost

@caver01 I will be online in a few hours again. At work I asked a colleague (circuit designer) how this issue can be solved. It's possible to switch the event just via a transistor - but I forget which type ... I think N-type, but you need some resistors (at lease one at base) and best a diode (for current switchbacks) the make sure switch behaviour.

On the other hand he said, that the optical-mosfet is also a nice method to trigger that and it's really fail safe as the circuits are isolated.

Then I asked the method with the reset trigger... Clear answer: NO! A reset is a reset as you don't know what there really happens. It can be named as "emergency exit" if the µc is trapped but never should be used as regular switch.

caver01

@cyperghost I am glad you confirmed the transistor idea. I was actually going to try that first based on other things I have read. I believe I need to use an NPN transistor, and I have a bunch of them collected from old hardware. Here is an example where someone is using a transistor to trigger a button press for a camera trigger. The SSR is suggested first, but someone concludes that a transistor method would be adequate.

Most discussions elsewhere online regarding this topic mention the use of the opto-coupler SSR as a safer method, as it isolates both sides and you don't have to worry about linking the grounds. However, I am not that worried about isolation, and I won't have to buy anything to try it.

I didn't get a chance to lookup my python script last night. I will post something soon, and I will definitely try the transistor trick.

cyperghost

@caver01 Thank you for the linked document with the transistor. As you use the same potential there should be no drawbacks with connecting the Pie and the mausberry.

As said, the SSR is the easiest method at all.

caver01

@cyperghost The only risk I see with the transistor here is if the momentary button wired to the Mausberry is not connected to ground on one of its poles. Maybe even that does not matter, but the diagrams I have seen including the one linked above has the emitter connected to ground that is common to both the switch and the Pi. I may look at the Mausberry circuit closely to see if I can trace one of the button terminals to ground. If not, I may just try it anyway.

cyperghost

@caver01 Well take a voltmeter and measure switch pins against ground. I hope that there are 3.3V ;)

caver01

@cyperghost Good idea!

caver01

@cyperghost well, that NPN transistor totally worked!

First, I modified my mausberry script by adding another variable for GPIOpin3 which I set to 18.
Then, I added two more echo lines to export the pin and to set its direction to OUT:

#!/bin/bash

# need to call this from /etc/rc.local

# End Emulationstation if condition of running binary is true (v1.59)
# v1.0 07/21/17 by cyperghost - Inital run 
# v1.1 07/22/17 - Added chown command to set right user permission for creating es-shutdown // cyperghost
# v1.2 07/23/17 - Some small improvments, easier to maintain, removed echo, removed else branch // cyperghost
# v1.5 07/27/17 - Great step to exit ES even if emulators is running by runcommand.sh are started // meleu
# v1.55 07/29/17 - all kudos go to @meleu for his alltime genious RegEx hack! // meleu
# v1.56 07/30/17 - All emulators will be detected. // meleu
# v1.58 08/02/17 - generel method: Use PPID to detect child PIDs now (ScummVM fix) // cyperghost
# v1.59 08/03/17 - nothing new, just polishing the code // meleu
# I just checked with SSH command - it saved my metadata! Maybe you need to extend sleeptimer!
# greetings @pjft for his famous favorits and @meleu for his RegEx sniplets and his constant help! 

#this is the GPIO pin connected to the lead on switch labeled OUT
GPIOpin1=23

#this is the GPIO pin connected to the lead on switch labeled IN
GPIOpin2=24

#this is the GPIO pin connected to the 1k-ohm resistor, leading to the base of NPN transistor (emitter and collector to power switch poles)
GPIOpin3=18

echo "$GPIOpin1" > /sys/class/gpio/export
echo "in" > /sys/class/gpio/gpio$GPIOpin1/direction
echo "$GPIOpin2" > /sys/class/gpio/export
echo "out" > /sys/class/gpio/gpio$GPIOpin2/direction
echo "1" > /sys/class/gpio/gpio$GPIOpin2/value
echo "$GPIOpin3" > /sys/class/gpio/export
echo "out" > /sys/class/gpio/gpio$GPIOpin3/direction


while true; do
    power="$(cat /sys/class/gpio/gpio$GPIOpin1/value)"
    if [[ "$power" == 0 ]]; then
        sleep 1
    else
        emucall="$(sed -n 4p /dev/shm/runcommand.info | tr -d '\\"' | tr '^$[]*.()|+?{}' '.')"
        if [[ -n "$emucall" ]]; then
            emupid="$(pgrep -f "$emucall")" 
            pkill -P "$emupid"
            kill "$emupid"
            sleep 4
        fi    

        espid="$(pgrep -f "/opt/retropie/supplementary/.*/emulationstation([^.]|$)")"
        if [[ -n "$espid" ]]; then
            touch /tmp/es-shutdown && chown pi:pi /tmp/es-shutdown
            kill "$espid"
            exit
        fi
        # End Emulationstation if condition of running binary is true

        sudo poweroff
    fi
done

Finally, I added the line:
echo "1" > /sys/class/gpio/gpio18/value to the emulationstation.sh in place of the sudo poweroff in
that file.

I soldered up two components, an NPN transistor and a 1k-ohm resistor. The resistor goes on the middle pin of the transistor, and the outside pins go in parallel to the power switch connected to the mausberry. Then, on the opposite end of the resistor, I connected this to GPIO18 as specified in my updated script.

Now, when I select SHUTDOWN in the ES menu, it completely powers off.

Thanks for all of your help!

caver01

I am going to create a new thread for working on the Python version of the BASH script. I think that makes more sense than burying it here.