[TriEmbed] TriEmbed Digest, Vol 39, Issue 20

[Michael Monaghan]

Chip,

I have lots of thoughts, but I don't think any of them get you around near
end cross talk (NEXT) which is essentially what this is.  Any filtering
would have to allow for the round trip to the reflector which will suffer
air gap loss so shorter reflection paths will have higher return power and
likely overcome polarization and filtering.  Separating the transmitter and
receiver a few feet apart might help as may baffles to reduce NEXT.  You
might also measure delay from transmit to detect, but some part of me
screams we don't know what the light path will look like and it might be so
short as to be meaningless lacking a coherent light source like a laser.
Short of placing the receiver on the other side of the path, I think heavy
fog and rain create an out of service condition that you can likely live
with and use handwavium (tm) to explain away.  "These are not the counts
you're looking for."

Mike

On Thu, Aug 25, 2016 at 11:03 AM, Chip McClelland <chip at mcclellands.org>
wrote:

> Mike,  interesting. Had not thought about reflected light. Is there a way
> to use polarization to differentiate a signal reflected from the other side
> and not just from the fog?
>
> Thanks,
>
> Chip
>
> On Thursday, August 25, 2016, Michael Monaghan <mike at chipworks.net> wrote:
>
>> Shane,
>>
>> That's elegant and conserves power.  Only think I can think of that
>> _might_ improve on it is to profile the beam break time to optimize the
>> polling period.  Chip might get by with 3 tests a second instead of
>> hundreds and that will feed his battery need.  ;)
>>
>> I can confirm your suspicion that fog will reflect an IR beam.  We had a
>> gates in Fernbank Forest that used a reflected IR carrier.  In heavy fog or
>> rain, the sensor would false positive.  Another had remote sense instead of
>> reflected.  In the same conditions it wouldn't detect carrier at all.  Both
>> of these false positives would leave gates open for vehicle traffic
>> crossing unsigned pedestrian pathways.  In the end we put a timer on them
>> to close the gates and hoped no one ever stalled under them.
>>
>> Mike
>>
>> On Thu, Aug 25, 2016 at 10:46 AM, Shane Trent via TriEmbed <
>> triembed at triembed.org> wrote:
>>
>>> Chip,
>>>
>>> I think you could set timer for the carrier generation to produced 15
>>> cycles and have the interrupt from the IR detector stop the pulse
>>> generation and set your result flag. So if you reach 15 cycles, the path to
>>> the reflector was blocked indicating an object in the path. This would let
>>> you keep your IR LED drive time as short as possible by turning off the
>>> drive as soon as the carrier is detected. So if your detector is awesome
>>> and locks in 7 cycles, it only get 7 cycles. If the detector needs 12
>>> cycles, it gets only 12 cycles.
>>>
>>> I wonder if heavy fog or rain would reflect enough of the IR to trick
>>> the receiver into believing it could ALWAYS see the reflective target?
>>>
>>> Shane
>>>
>>> On Wed, Aug 24, 2016 at 11:57 PM Chip McClelland via TriEmbed <
>>> triembed at triembed.org> wrote:
>>>
>>>> Shane,
>>>>
>>>> That makes a lot of sense.  By looking at a specific time and for a
>>>> specific carrier frequency (38 kHz), I can better see the signal against
>>>> the background noise.
>>>>
>>>> So, instead of a phototransistor circuit, I would use an IR receiver
>>>> sensor such as this one - http://www.vishay.com/docs/82474/tssp4p38.pdf
>>>>
>>>> To reduce the duty cycle, I would only run the 38kHz signal in bursts
>>>> or 10-12 pulses.
>>>>
>>>> Will check this out.
>>>>
>>>> Chi
>>>>
>>>>
>>> _______________________________________________
>>> Triangle, NC Embedded Computing mailing list
>>> TriEmbed at triembed.org
>>> http://mail.triembed.org/mailman/listinfo/triembed_triembed.org
>>> TriEmbed web site: http://TriEmbed.org
>>>
>>>
>>
>
> --
> Sent from my iPhone - please excuse the typos
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.triembed.org/pipermail/triembed_triembed.org/attachments/20160825/b8c6436b/attachment.htm>