HA Rules - How do you handle manual overrides?

[TonyNo]

It's not a strain gauge then, or, if it is, the support circuitry is capacitively-coupled. You should be able to put a leg on a load cell (or add a rosette with some strategic cuts in the leg) and turn it into a scale, though!

 

[ano]

It's not a strain gauge then, or, if it is, the support circuitry is capacitively-coupled. You should be able to put a leg on a load cell (or add a rosette with some strategic cuts in the leg) and turn it into a scale, though!

The resistance of these sensors change in relation to the stress it undergos AND also the temp. of the sensor. This is why they can NEVER be calibrated to trigger with a set load UNLESS the temp. of the sensor is kept PERFECTLY constant, which just isn't realistic. The processor you must use with these sensors IGNORES slow changes to the sensor like temp. changes, but triggers for FAST changes, like someone stepping on the sensor. A scale shouldn't change readings when the temps in the room change, unlike these sensors.

Imagine you had this scale sensor that read 100 ohms with no weight on it, but 200 ohms with 10 pounds on it at 70 degress F.
But at 75 degrees F it read 200 ohms with no weight but 400 ohms with 10 pounds.
O.K if your scale reads 200 ohms, what weight is on it? What you have no idea without knowing the temp?
But lets say you see the resistance rise from 200 ohms to 300 ohms in 1 second? You know likely the weight on the scale increased as it would be impossible for the temp. to change so fast. THAT is how these sensors work. So again, these sensors can not measure weight, just a fast CHANGE of weight.

 

[drozwood90]

I used to design automotive test equipment. Some of that included Load Cells.

That's why you design it with a bridge and known value shunt (relay injected).

Eww...ASCII art turned out bad...so I screen captured it and made it in to a JPG...

The load cell ends up being one of the R1-4, and the shunt calibrates the entire system. Then you need only read between S1 and S2 to get a reading on how much load is there. If you pick resistors that have the same thermal profile as the load cell, then the entire network stays equal all the time. This eliminates your "drift" that you were seeing...

BUT, you can use the shunt to re-calibrate the system at any given time (as the shunt R is in parallel to R2, you know what the difference should be and how much it was (you read it)). This can be done at a time when the system thinks there is "no" load on the system. Heck, if you wanted to, add a resistor on the same S2 junction, with a DAC on it, and when the system "thinks" there is a 0 load, then you can "zero" it out (I think that was another thing I worked on). You basically (in software) manipulate the values using the "shunted" value.

Thinking about it...I can't remember (8 years ago)...but maybe R2 needs to be the load cell...hmm...

I'll have to get back to you on that one...but the concept is there...I just can't remember...but I think it makes sense that you need to have the load cell in R2, so you can calibrate the thing...

As I said, I'll have to get back to you on that one...

--Dan

 

[Gangsta]

Has anyone tried mounting a Microwave sensor on/under their bed frame to detect occupancy. I am thinking of picking one up from the place that fits car alarms nearby. I believe its the same sensor that detects when you get too close to the protected vehicle. Apparently you can adjust the sensitivity to determine how far it will detect somebody? Just an idea, and will save me some cash if someone has already tried it.

 

[TonyNo]

Temperature effects on strain gauges.

 

[tmbrown97]

Ok, we need to come up with a workable solution here - this could be way better than my Google Calendar idea! Of course, I'd still like to capture the family schedule to show on the screen, but what sparked all this was wanting the house to go in Quiet mode when my wife is sleeping - because with her working 2-3 nights/week, she sleeps during some days and hates the phone ringing, the doorbell going off, etc.

imagine the possibilities - sensors under the bed - sensors under the crib mattress... so much fun automation! Plus I like the idea of the recirculation pump kicking on as you get out of bed.

 

[ano]

Has anyone tried mounting a Microwave sensor on/under their bed frame to detect occupancy. I am thinking of picking one up from the place that fits car alarms nearby. I believe its the same sensor that detects when you get too close to the protected vehicle. Apparently you can adjust the sensitivity to determine how far it will detect somebody? Just an idea, and will save me some cash if someone has already tried it.

Microwave sensors, like passive IR sensors don't detect occupancy they only detect movement. If your a really restless sleeper it might work, but once you stop moving, probably not.

I believe its the same sensor that detects when you get too close to the protected vehicle.

Yes, it will detect you when you GET too close, it won't detect you when you ARE close. There IS a difference.

 

[signal15]

Too bad this thing doesn't have an output to tell the HA system when you are asleep:
http://www.myzeo.com/

 

[drozwood90]

Temperature effects on strain gauges.

Thanks! That's what I had in mind.

So, the shunt would go across R4 (in their picture).

--Dan

 

[Gangsta]

Mabey this is too much, but

h**p://ursi-test.intec.ugent.be/files/URSIGA08/papers/KP1p16.pdf

this describes the basis for non contact heartbeat detection @ 1meter using a pair of modified wifi adapters.

 

[tmbrown97]

what about this?
http://www.cardaccess-inc.com/inhomeproduc...ex.php?a=sbs10a
more options:
http://www.teluhealth.com/catalog/

looks to be standard contact closure; nothing special required. Who's gonna try one?