Building an IR tachometer - logic problem?
r.daneel
Posts: 96
Hi,
I am building what amounts to a tachometer using an IR sensor, and I am tearing my hair out trying to find what's wrong with my logic. The hardware setup is simple - I have a TCRT5000 IR sensor module pointing at a rotating 50/50 black and white disc. The IR module works at 3.3V, is powered from the same 3.3V supply as the Prop, and is always on. The IR module has both analog and digital output pins - the digital out is connected directly to one of the Prop's IO pins. To test what's happening I have two LEDs connected to two other IO pins.
My simple test program looks like this:
It's pretty simple really - as the disc rotates and the black and white halves come into view of the IR sensor, the LEDs are supposed to blink alternately. This works just as expected - as I rotae the disc the LEDs do indded blink alternately.
What I really want is that a switch is turned on and off alternately on every revolution - so on rev 1 the switch goes on, rev 2 off, rev 3 on etc. SHouldn't be too difficult (I thought...). I changed the loop in the code to:
What I expected was that as the black and white halves of the disc were seen by the IR sensor the blue LED would blink as before, and as the black half of the disc came into view (and only the black half) the red LED would change state. What actually happens is that most of the time the LEDs behave as before - they blink alternately as the black and white halves of the disc come into view. Sometimes they will both come on, but they are never both off.
I've looked at this too long - I can't see where my problem is. Can anyone help?
Thanks.
I am building what amounts to a tachometer using an IR sensor, and I am tearing my hair out trying to find what's wrong with my logic. The hardware setup is simple - I have a TCRT5000 IR sensor module pointing at a rotating 50/50 black and white disc. The IR module works at 3.3V, is powered from the same 3.3V supply as the Prop, and is always on. The IR module has both analog and digital output pins - the digital out is connected directly to one of the Prop's IO pins. To test what's happening I have two LEDs connected to two other IO pins.
My simple test program looks like this:
CON
_clkmode = xtal1 + pll16x
_xinfreq = 5_000_000
VAR
long irPin, led1Pin, led2Pin
PUB Main
irPin := 9 ' IR sensor
led1Pin := 10 ' BLUE LED
led2Pin := 11 ' RED LED
cognew(@entry, @irPin) ' don't care about return value for this test
DAT
org
entry mov arg, par
rdlong scr, arg ' IR sensor pin
shl ir, scr ' mask
add arg, #4
rdlong scr, arg ' BLUE LED pin
shl blue, scr ' mask
add arg, #4
rdlong scr, arg ' RED LED pin
shl red, scr ' mask
mov scr, #0 wc
muxc outa, blue ' BLUE LED off
muxc outa, red ' RED LED off
mov dira, #0 ' all pins input
or dira, blue ' BLUE LED pin is output
or dira, red ' RED LED pin is output
waitpeq ir, ir ' start - wait for IR pin to go high (seeing black)
loop waitpne ir, ir ' wait for IR pin to go low (seeing white)
or outa, blue ' BLUE LED on
mov scr, #0 wc
muxc outa, red ' RED LED off
waitpeq ir, ir ' wait for IR pin high (seeing black)
mov scr, #0 wc
muxc outa, blue ' BLUE LED off
or outa, red ' RED LED on
jmp #loop ' loop
ir long 1
blue long 1
red long 1
arg res 1
scr res 1
It's pretty simple really - as the disc rotates and the black and white halves come into view of the IR sensor, the LEDs are supposed to blink alternately. This works just as expected - as I rotae the disc the LEDs do indded blink alternately.
What I really want is that a switch is turned on and off alternately on every revolution - so on rev 1 the switch goes on, rev 2 off, rev 3 on etc. SHouldn't be too difficult (I thought...). I changed the loop in the code to:
loop waitpne ir, ir ' wait for IR pin to go low (seeing white)
or outa, blue ' BLUE LED on
waitpeq ir, ir ' wait for IR pin high (seeing black)
mov scr, #0 wc
muxc outa, blue ' BLUE LED off
xor outa, red ' RED LED on
jmp #loop ' loop
What I expected was that as the black and white halves of the disc were seen by the IR sensor the blue LED would blink as before, and as the black half of the disc came into view (and only the black half) the red LED would change state. What actually happens is that most of the time the LEDs behave as before - they blink alternately as the black and white halves of the disc come into view. Sometimes they will both come on, but they are never both off.
I've looked at this too long - I can't see where my problem is. Can anyone help?
Thanks.
Comments
Welcome to the Parallax forum!
Your code works, as you expected. If the Blue LED represents Bit0 and the Red LED represents Bit1 in a binary counter, then the repeating sequence is ...
00
01
10
11
... and then back to 00 to repeat the sequence
What does your circuit look like, perhaps there is bounce on your signal that is causing a problem. You mentioned that "The IR module has both analog and digital output pins", but the datasheet for the TCRT5000 is a typical open collector transistor output without any signal conditioning ...basically analog.
You can also simplify this...
mov scr, #0 wc muxc outa, blue ' BLUE LED offto this...The IR module I'm using is built from the TCRT5000 sensor. It's one of these:
My circuit is pretty simple. Both the Prop and the IR module are powered from the same 3.3V source - so VCC and GND on the IR module are connected to the same VCC and GND as the Prop. No switching - when the Prop is powered so is the IR module.
The DO pin of the IR module is connected directly to Pin 9 of the Prop.
The AO pin of the IR module is not connected.
Pin 10 of the Prop is connected to a 100Ohm resistor which is connected to the positive lead of a blue LED. The other side of the LED goes to GND.
Pin 11 of the Prop is connected to a 100Ohm resistor which is connected to the positive lead of a red LED. The other side of the LED goes to GND.
That's it...
I thought about bounce - but the first test program works just fine. Wouldn't that be subject to bounce as well? If it is bounce, can I smooth it out somehow?
" If it is bounce, can I smooth it out somehow?" - A schmitt trigger on the output of the module might help but only after a small integrator circuit....
Small Integrator
1k From IR sensor >----/\/\----o------> to Schmitt Trigger | | o--||--> GND 0.1uFWithout hysteresis you will have 'edge' jitter on each transition of your IR detector.
Note; it is also possible to smooth out a signal bounce in software by oversampling. This is done by ensemble averaging, which basically increases your signal to noise ratio.
The Signal to noise ratio is essentially equal to the square root of the number of samples taken and closely relates to Nyquist. The caveat is that for every time you sample you take away from the maximum speed or rate in which you can detect a pulse from your IR detector circuit.
Here is something I put together a few years ago trying to explain Nyquist in a nut-shell...
http://wiki.ohmspace.org/What_is_Nyquist_doing_for_you%3F
Try the code below and see if that will help. Note, you might need to adjust the Delay value.
CON _clkmode = xtal1 + pll16x _xinfreq = 5_000_000 VAR long irPin, led1Pin, led2Pin PUB Main irPin := 9 ' IR sensor led1Pin := 10 ' BLUE LED led2Pin := 11 ' RED LED cognew(@entry, @irPin) ' don't care about return value for this test DAT org entry mov arg, par rdlong scr, arg ' IR sensor pin shl ir, scr ' mask add arg, #4 rdlong scr, arg ' BLUE LED pin shl blue, scr ' mask add arg, #4 rdlong scr, arg ' RED LED pin shl red, scr ' mask andn outa, blue ' BLUE LED off andn outa, red ' RED LED off mov dira, #0 ' all pins input or dira, blue ' BLUE LED pin is output or dira, red ' RED LED pin is output waitpeq ir, ir ' start - wait for IR pin to go high (seeing black) loopA waitpne ir, ir ' wait for IR pin to go low (seeing white) call #Debounce if_c jmp #loopA or outa, blue ' BLUE LED on loopB waitpeq ir, ir ' wait for IR pin high (seeing black) call #Debounce if_c jmp #loopB andn outa, blue ' BLUE LED off xor outa, red ' RED LED toggle jmp #loopA ' loop Debounce mov temp,cnt add temp,delay waitcnt temp,#0 test ina,ir wc Debounce_ret ret ir long 1 blue long 1 red long 1 temp long 0 Delay long 800 ' 1/100,000th of a second arg res 1 scr res 1I used three different techniques to capture the encoder pulse but I didn't use your waitpxx technique. I was getting erratic readings when using counters to monitor the encoders. Tracy Allen had some good advice about ground bounce.
You might be interested in the techniques I used to read the encoder as a comparison to your method. I wonder how many other ways one can read an encoder with a Prop?
I'll mark this thread as solved - thanks Beau and Duane.
http://www.techrepublic.com/blog/data-center/noise-and-jitter-nemesis-of-digital-communications/
I'm a software person that knows enough about hardware to make me dangerous, especially with a soldering iron in my hand. Anything beyond that explanation will have to come from someone like Beau.
I just hadn't thought of using it. I like the idea.
I don't really understand. Anything I said would just be repeating what others stated in the thread. As I mentioned in the thread, the issue only occurred when the sensor and servo motor were sharing the same power supply.
Steve
Sorry. Didn't see the reference to the servo on the same power supply.
Inductive kickback makes a lot of electrical "noise". Spiky, high voltage noise.
the kind of electrical noise that will drive logic circuits nuts.
Filtering and / or isolation are the only cures for that.
I had nothing on the same power supply when I was doing my test - well, nothing other than the Prop and the IR sensor. Waiting for the jitter to settle down (i.e. using the code Beau provided) solved the problem. Again, I'm no hardware engineer, but I get the impression that signal jitter is just part of the deal - introducing hysteresis, averaging the signal over a few samples or just waiting for the jitter to settle are just different ways of solving the problem.