Projects‎ > ‎Scooter Computer II‎ > ‎Sensors‎ > ‎

Reed Switch Debounce

The reed switch for the wheel sensor bounces and caused erroneous readings, so this needed to be cleaned. As an example left below good (80-100km/hr) and right bad (20- 1700

scii_test_good_part.png scii_test_bad_part.png

The first issue was that it was probably due to using connections that are floating - below is the new config showing a 10K resistor holding down FIO4 down to GND, with the reed switch connecting +5V when switched.

reed_switch_hookup.png

This should have worked better than just wiring up the reed switch with no pulldown but it was worse! So I played around with a capacitor to clean up the readings using my lathe as a test driver. I hooked a 0.1uF capacitor across the resistor above (GND - FIO4). Below is the difference - left no cap, right 0.1 uF - there is still a bit of bounce but the smooting removed all errors in the readings, so for now is it good enough to run with!

bounce_no_capacitor.JPGbounce_with_0u1_capacitor.JPG

Below is a test log using the 0.1 uF capacitor debounce method (plotted in Excel) - a 20 minute ride with not one error (recorded at 2 samples per second)

scii_debounced_reading.png

Next

The above is a hack and it was just what I had lying around the lab - this really needs to be improved with a proper debounce circuit...

Debounce circuit http://www.labbookpages.co.uk/electronics/debounce.html

Update: 19/07/2011

I ordered a MAX6816EUS+T debounce IC which "should" have been perfect as it is a debounce unit which just requires a capacitor. Trouble is that when it arrived I discovered it is tiny surface mounted device, great for a production circuit board but not for wiring up manually (it's about 3mmx3mmx1mm in size)... time for a visit to JayCar.

From: http://robotgestation.com/SensorForce.html

Update: 16/08/2011

I got the diodes etc above and tried a trial but couldn't get it to work... then realised the "ICIC" wasn't referring to the LabJack input but an additional IC, the 74HC14 Schmitt trigger, which I hadn't purchased. Since I am streaming now I might go with a combination of software debounce and what LabJack recommend (below). Since I am measuring every pulse it should be pretty obvious if I get a double trigger (an I can add the value to the trailing time).

2.8.1.3 - Input: Mechanical Switch Closure

Update 21/08/2011

And the result of the above... a pretty good falling edge to trigger off and it looks like one that I will keep.

final_debounce.JPG

Above: Debounced signal from reed switch driven by a lathe.

Update (23/08/11)

OK all though I get a clean signal I an still getting wavering results. I think this is due to the reed switch itself. I am seeing a few microseconds but this is enough to throw out my calculations. The animated diagram below is from http://reed-switch-info.com

todo - add reed switch debounce animation

Update (10/09/11)

After still experiencing issues with jitter and incorrect reads (even buying and trying a signal generator), I posted a question on the LabJack forum. After working with the support guys there, we worked out that there was a few issues:

  1. The U3 LabJack clock, as it is rated to only 1.5% accuracy introduces jitter and the more expensive U6 would provide much more accurate readings (saving up now)
  2. The way I was reading the LabJack was too quick and the LabJack wasn't returning data correctly (reduced read rate)

Below is the thread for reference:

http://forums.labjack.com/index.php?showtopic=5479&st=0

ą
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:53 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:53 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:53 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:53 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:53 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:53 AM
Ċ
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ą
U3-HV.JPG
(26k)
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ċ
cicuit.png
(161k)
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:54 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:55 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:55 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:55 AM
ą
Paul McIntosh,
Mar 25, 2015, 12:55 AM
Comments