Most sensitive acceleration direction of ST Multipurpose Sensor accelerometer?


(Mike) #1

Hi folks, would anyone know where there are deep specs on the ST Multipurpose Sensor. The brief specs and install guide that I can find don’t have accelerometer specifics.

I want to attach it to a surface that really only moves in one direction when bumped. But it’s sort of at the edge of whether ordinary bumping is enough for it to respond. (Some times it does, some times it doesn’t.) I know the accelerometer has the potential to be super sensitive under ideal conditions. So I want to make sure I’ve maximized the orientation sensitivity, in case that will help.

I am talking about the plain accelerometer. So, the additional magnet bar (used for doors and windows) is not present at all.

If anyone knows whether one of the axes / directions of this little guy is more sensitive, that’d be great. And if you know actual specification of the energy needed per direction, that’d be super great.


Using ST Multisensor accelerometer as mailbox mail detector
(ActionTiles.com co-founder Terry @ActionTiles; GitHub: @cosmicpuppy) #2

The resolution specs of the data is available in great detail if we can determine what chip is being used, and possibly what mode it is configured for.

Most 3-Axis chips are designed to be equally sensitive in all three directions, though.

My guess is it set to measure the range ±2g, in a resolution of 1024/g (ie, ±1024). To check this, watch the Device Sub-Tile numeric display as you place the sensor in all 6 basic orientations. Sometimes the chip is configured for ±1g,4g, 8g, or even 16g.

Is the chip model # legible (I’m guessing Invense brand…)?

I only have original model multi-sensor; have to take one apart to see the chip.


(Mike) #3

Thanks, this is all helpful to know. Especially that they usually try to make all 3 axes equally sensitive. I played with it some and the highest I got it was slightly over 2000 (positive or negative). I had to really whip it to get it that high. More commonly, values were in the mid hundreds to about one thousand, although often 2 of the axes would be very small numbers. This is with me simply whipping it in my hand, or tapping it. So I guess it is 2 x +/-1024, as you said.

Here are some pics. I hope they are enough. The big chip in center with a big S (I think) reads

EM3585 1510ACL0YE (e3) TW

If they try to make all 3 sensitive, would this still at least imply that the “corners” are less sensitive? (If it accelerates toward a corner of the device instead of perpendicular to one of its sides.)

Are the pics good enough? Again, this is the current model SmartThings Multipurpose Sensor,





(ActionTiles.com co-founder Terry @ActionTiles; GitHub: @cosmicpuppy) #4

So here’s the specs… https://www.silabs.com/Support%20Documents/TechnicalDocs/EM358x.pdf

Reading them now… Oops… That’s probably just the ZigBee chip (with MCU / micro controller). I don’t see Accelerometer on that spec at first glance.

Probably another chip somewhere?


(ActionTiles.com co-founder Terry @ActionTiles; GitHub: @cosmicpuppy) #5

I think it’s the IC chip 6521YL, but Google not helping.

As for corners… Yes… That will register as less sensitive in any one vector, because the g-force will be distributed to 2 or more axes.

Of course, mathematically it should still add up. I forget my physics… But roughly just a straight sum of all 3 values should be a good indication of overall force… Subtract the baseline (standing still / static) value.


(Mike) #6

Thanks man, I appreciate you taking a look. Sorry they don’t have detailed specs on their stuff.