Debounce a Switch and Increment a value ?

scottc · Joined: 16 Aug 2010 Posts: 95

Hi,

I have a push button switch connected to RB0. I am trying to figure out
how to increment the count by 1 "only" regardless of how long the switch is pressed. Basically the switch press increments "count" The count variable is tested in main, Depending on the actual count value 1,2,3,4 etc a certain item is displayed on my lcd.

Here is what I have so far, its not great but works. Any ideas to make it
better would be appreciated.

This function is called from main, in a while loop.

Douglas Kennedy · Joined: 07 Sep 2003 Posts: 755 Location: Florida

The way many do this is with a timer. The timer is triggered to start when the
the button is pressed. The timer continues to count while the button bounces if the button signal is still there when the timer times out it is accepted as a true press. Never ever put much code inside an isr but in this case the code to test the state of the button press is just an input statement so it is placed inside the timer isr. Detect the button press enable the timer have the timer signal the state of the button as it times out after spanning the bounce time of the button. I'd do this all yourself as it it a good way to learn about timers and the enormous speed differences between mechanical devices (button) and the instruction cycle of your PIC

scottc · Joined: 16 Aug 2010 Posts: 95

Douglas, I like the timer idea. I'm thinking thats a better way to handle
what I want to do.

Will give it a go.

Thanks Scott

arunb · Joined: 08 Sep 2003 Posts: 492 Location: India

You could use the algorithm shown below to achieve the same result

Gabriel · Joined: 03 Aug 2009 Posts: 1067 Location: Panama

usually what i do is when a change in pin state is detected,

i go into a loop that will exit once its gotten the same pin state 5 times in a row... this takes very few usec. ... longer for bad contact buttons.

that will debounce a button quite nicely and take very short time.

all you would need is a pull up.... and it will debounce a button no matter how bouncy it is...

gabriel
_________________
CCS PCM 5.078 & CCS PCH 5.093

Humberto · Posted: Mon Oct 18, 2010 1:14 pm

For sure there are so many ways to do a debouncing algorithm, the point is what the original poster stated:

BertKoerts · Joined: 05 Jun 2006 Posts: 2

Perhaps the method from Scott Dattalo is interesting, it is called vertical counters and I use it for almost all my projects. You need to call the function frequently (set up a timer interrupt with 1ms rate for example). It needs only two variables, little program memory and is fast. It handles a complete byte for debounce input so you can use it for all your inputs on port B for example, or just for one bit if that is all you need.

http://www.dattalo.com/technical/software/pic/debounce.html

I used the pseudo C code from this page.

Wayne_ · Joined: 10 Oct 2007 Posts: 681

The simple solution

scottc · Joined: 16 Aug 2010 Posts: 95

Wayne, the example header was random madness Smile

I was also
testing returning values.

I tried your code snip out and it works very well. I modified it as I was
using count as a global value.

Humberto · Posted: Wed Oct 20, 2010 5:58 am

John P · Joined: 17 Sep 2003 Posts: 331

I don't understand why people keep telling each other that you need to fool around with counters in order to get a reliable reading of a pushbutton.

All you need to do is set up a repetitive timer routine that gives you an interrupt at intervals longer than the bounce time of the switch. Say 50msec, i.e. 1/20 second. That's enough to catch even a fast press, but it should cover any contact bounce. Check the data book describing the switches and do some experiments, obviously. If you already have a timer running but it's faster than you need, divide it down and pretend it was at the slower rate.

Then in the interrupt, all you have to do is read the button once. You'll most likely need an additional single bit per pushbutton, storing the button's last state. Then for every instance of the last state being "off" and the present one "on", you can increment your times-pressed count.

Does this method give anything up? Well, conceivably you might want to get the button state at the instant that it changes and becomes stable. Then you'd have to watch it more closely, but in general what a pushbutton does is check for user input, and if it's a twentieth of a second earlier or later, nothing's affected, unless you're trying to measure user reaction time. A limit switch in a machine might be a different story. But don't waste you own time and processor resources on unnecessary stuff.

Wayne_ · Joined: 10 Oct 2007 Posts: 681

Humberto
"pushed" is defined as static, this means it will retain it's value from the last call to the function. It will not get initialised to false each time. This is what static means. It is like declaring a var as global but associating it with a function.
The var pushed is used to flag the state the button is in each time the function is called.

Because the main routine is missing some of you have a different idea of what the scottc wants to do, I assumed the following:-

scottc · Joined: 16 Aug 2010 Posts: 95

Wayne, the code works fine and does what I had expected. Main thing
was to press the switch and increment the count by 1 regardless of
how long the switch was pressed in. Indeed it does this very well.

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

The idea is that virtually all switches or pushbuttons finish bouncing in
much less then 10 ms. So if you sample the switch at a 10 ms rate, the
worst case response will be 20 ms, to detect a pressed or released state.

In the diagram below, a pushbutton switch is initially pressed (giving a low
level) and then released. It bounces for a few milliseconds after your
finger is removed from the pushbutton. It's release about 4 ms after a
sample point (in the interrupt routine). When it's next sampled in the isr
it's bouncing. It happens to bounce low, so it's still read as being pressed.
Finally on the next bounce, it's read as being released.

John P · Joined: 17 Sep 2003 Posts: 331

PCM Programmer has explained my point better than I could have done myself. Thanks.

But note that in general you do need to dedicate one bit of storage to each pushbutton, to store its previous state. Otherwise, when the processor examines the present state of the button, there's no way to know whether it changed in this cycle, or an hour ago, or never. It's the difference between "now" and "previous" that lets you know that a change occurred.