SparkFun Forums

[eforman]

A few questions for those of you that have worked with these:

- I am trying to detect when a person is present in an 8' circle. This is for an outdoor sculpture. For various reasons I could get into if anyone cares, normal solutions like PIR and ultrasonic have disadvantages. Do you think using a MLX90614 (standard 90° FOV) mounted 4'-5' high and pointed down could detect someone on the edge of the circle? I.e. it would only see legs and feet. I do NOT need to know the temperature, and accuracy is not an issue - I only need the sensor to see a numerically significant difference from its own temperature. The issue is only a very small part of its FOV could have the person in it. This is also for outdoors year-round, so ambient temperatures and clothing types will change a lot.

- I'm curious, why would anyone *not* want the "gradient compensated" version? e.g. MLX90614-ACA. From the datasheet, it sounds like it's always useful:

"Melexis has introduced the xCx version of the MLX90614. In these MLX90614xCx, the thermal gradients are measured internally and the measured temperature is compensated for them. In this way, the MLX90614xCx is much less sensitive to thermal gradients induced from outside"

Thanks all

[Valen]

As a person is, or even his/her legs are, usually much larger than a 8 inch circle, I don't quite understand how this detection range/area is situated. Is it vertical? On the ground? Somewhere in the air between waist and ground? Also "a very small part of it's FOV could have the person in it" makes not much sense to me. Do you want to detect entrance in to the border of some circular area?

Also, is heat detection really a requirement? How about capacitive detection, which can also be non-contact. But you would need large antenna like conducting areas to sense a object's pressence.

[Mee_n_Mac]

Assuming a 90 deg FOV, to detect out to 8' (I assume you meant a circle w/a radius of 8') you need to have the sensor mounted 8' above the center of the circle and looking straight down. Think isoceles triangle w/equal sides being the height and radius. To keep the math simple let's make the height about 5.6' and that gives a circle w/an area of 100 sq ft. Let's say you step into the outskirts of this circle w/both feet and part of you legs ... such that their area (as seen by the MLX) is 1 sq ft. So the question now becomes how mush different in temperature must those feet+ be to cause a detectable difference in measured temperature.

This gets hard because the MLX doesn't measure temperature directly, it measure IR power over some waveband, falling on the sensor. The power emitted from those feet+ is a function of their temperature, the emissivity of their coverings and the amount of IR reflected off them. The PDF mentions a resolution of 0.02 deg C but doesn't mention any noise spec, no doubt because the MLX does some internal filtering of noise. Because your usage is outside, you can't use a fixed threshold as the background temperature will be changing over time due to the Sun, and temperature and perhaps other changes (? grass growing, snow ???). You need to detect the increase (or decrease) in temp shortly after the person steps into the circle because you'll be running an averaging filter to remove the hopefully very slow changes due to the background. Again you hope that the internal filter of the MLX doesn't slow down the small change due to the feet+ such that it falls victim to your averaging filter.

Given temp is related to the IR power by T^4 I'll GUESS that it should work but I'd have to know more about the background and the statute (? this is the same problem as your other post ?) and do some research on the IR emissivity of clothing to have any real idea.
http://en.wikipedia.org/wiki/Stefan%E2% ... tzmann_law
http://en.wikipedia.org/wiki/Emissivity