Data sheet for accelerometer wanted

geezer88

Senior Member
I've run across accelerometers in junk cameras and broken radio controlled helicopters, but I've never been able to find any data sheet for them. The device is a surface mount package with a metal top. Laser scribed on the top is a logo? that looks like several = stacked tightly together. Another logo? looks like a big C with a small m being eaten pak man style. Finally the number on the has been 820 or 821. Anyone figure these out?
tom
 

geezer88

Senior Member
Thanks for the tip. I rummaged all through the Freescale site and could not find anything describing the sensor I have been messing with. I did take the top off the package, and it is has a beam suspended by it's two nodes, and a silicon chip, flipped, to condition the signal, and four terminals to the outside world. There exist both beam style gyros and beam style accelerometers. Since this one came out of a stabilized digital camera, I guess it could be either.

The package is 4mm wide, 8mm long, and 2mm high. My eyesight, and magnifier are not up to the task of seeing much detail beyond the identification of the beam and flip chip. I can't see any markings inside, and my attempt to "unflip" the chip broke it.

I'm pretty sure the thing won't work anymore.

tom
 

inglewoodpete

Senior Member
I'm pretty sure the thing won't work anymore.
Yep, I'd go along with that.

I use the ADXL335BCPZ chip. Kind of tiny but very sensitive and has three axis outputs. The outputs are 0-Vcc (3.3v) and I read them with 3 x ADC input on the PICAXE. They should cost about $5 US (my guess).
 

geezer88

Senior Member
Thanks for the tip on the Analog Devices device.

For a rotating machine balancer I use to balance propellers, I've been using an ADXL103. It is only 5x5x2mm leadless chip carrier and has a single axis, +/- 1g, analog output that I examine with a PC based scope program, Scope. I use a retro-reflective photo sensor as the reference signal on one channel, and the analog accelerometer signal on the other. The Scope program is great because I can filter the bandpass to get rid of all but the range of interest, and get a great signal to noise ratio.

Reading the data sheet on the one you have been using, I see it is even smaller. So far, I haven't thought of a multi-axis problem to work on. I'll have to see what pops up.

My interest in the original post surplus units is mostly academic, but if they were of sufficient sensitivity, there are some other uses that might be interesting.


tom
 

Pongo

Senior Member
Is that a DIY balancer? if so do you have any pics or info available? I've been thinking of building a balancer...
 

geezer88

Senior Member
Yes, it is. It really is pretty simple.

The ADXL103 I use is powered by a 5 volt regulator fed with a small 12 volt battery for portability. It's output is 2.5 volts at zero G, and +/- 1 volt from there for one G +/-. So with the capacitor input of my laptop sound card, the dc offset is eliminated, and the scope can easily read a few milli-Gs. This signal represents the vibration level. I've got small, flexible wires going from the ADXL103 to the input of the scope. Since the accelerometer only weights about one gram, double stick tape will hold it in place on the test subject. More tape is used to hold down the wires, so they don't pull on the accelerometer.

A once per revolution reference pulse is needed. I was luck to score a Banner D12SP retro reflective photo sensor. It is intended to be used with fiber optic cables, but I found that if I simply point it at a spot of reflective tape or a bicycle reflector, it would sense reliably from inches to 10 feet, depending on how big a reflector is used. In the case of a motor or propeller, a spot of reflective tape one cm square works great at about 6 inches. The Banner requires 10-30 vdc to operate, so powering it off the 12 volt battery works fine. The transistor output switches pull to supply voltage. To keep from hammering the input of the sound card, I run the output through a current limiting resistor and then to a diode to ground. This drops the output to the scope to about 0.7 volts.

My Toshiba laptop has a stereo microphone input, so I can trigger on one channel for the reference, and then observer the vibration input on the other channel. Many laptops are only mono, so this could be a reason to buy a USB soundcard.

The balancing I've done has been with objects that are close enough to planar that a simple two dimensional analysis will suffice. I've balanced the condenser fan on my home AC system, and during building the thing, I balanced a small fan that was easy to run on the bench. In the case of the small fan, it was about 3" in diameter, and at 2000 rpm I could detect the difference of a piece of electrical tape about 4mm square. My next project will be to check the balance on an experimental aircraft propeller.

The basic test method consists of attaching the tape to a convenient place on the rotating assembly. Double stick foam tape holds down the photo sensor where it can "look" at the tape as it rotates past. The accelerometer is attached as near to the rotating assembly as possible. It doesn't have to be at any particular rotational position, but it is easier to figure out where things are heavy if it is at 0 or 180 degrees with respect to the reference sensor.

With that all set up, crank up the object and get a stable reading on the scope of the reference pulse. Adjust the sweep speed to get two pulses on the screen as far apart as possible. Now get the amplitude adjusted on the vibration input to get a decent sized signal. On the little fan on my workbench, the signal was clean and easy to observe without any filtering. In the case of the AC condenser fan, the compressor lives in the same box, and generates a ton of extraneous vibrations. The floppy sheet metal housing vibrates at a bunch of random frequencies based on the complicated geometry. In the beginning, there was 50 mV of junk in addition to the signal. I had a beer while thinking about it. Then I remembered the built in filters in the scope software. Since the rotational frequency of interest was about 20 hz, I set the high pass filter at 12 hz, and the low pass at 30 hz. Wow, what a difference that made. A nice, clear sinusoid emerged. The fan had three blades. Two of them had balancing weights already. I moved them around a bit, based on the relative phase of the reference pulse and the peak of the accelerometer. In the end, I had to add another weight, and got the measured vibration down to a few milli G.

My next adventure will be to check the balance level of the propeller on an airplane. It was homebuilt by me, that someone else now owns. It will be interesting to see what the vibrations will look like before filtering. The engine is a four cylinder, horizontally opposed, four stroke cycle, with a 2.43:1 reduction gear drive. From earlier experiments with vibrations on an electric drill, the gearbox produces a ton of vibration at high frequency due to the gear teeth. Anyway with all the monkey motion in that complex system, it should be entertaining to see what happens. I think I'll get to it in about two months.

Hope that helps a bit. The circuit is so simple that I don't have a diagram, and since it's not picaxe powered, there's no software to show, either. If this is verboten on the forum, you can PM me and continue the discussion. I will say that I started out thinking that I'd use a picaxe to do the calculations, but soon figured out the Scope software would do everything I needed and more.

tom

ps. Here's some reference material:

PC Scope software: http://www.zeitnitz.de/Christian/scope_en

ADXL103 datasheet: http://www.analog.com/static/imported-files/data_sheets/ADXL103_203.pdf
 
Last edited:
Top