MMA3201 accelerometer experience?

Since Analog's ADXL150 accelerometer has been phased out (and the 8-LCC packaging on Analog's other accelerometers is not very user-friendly), I got a couple MMA3201D accelerometers (Freescale Semiconductor) to see if they'd be a reasonable substitute. After doing some preliminary prototyping, it looks like this sensor is really noisy. I was getting a standard deviation of 24mV (about .5g), and this is with a 12bit ADC sampling at about 40Hz. The ADXL150's I've used in the past were never this noisy.
Does anyone else have any experience with the MMA3201D? Is it just a noisy sensor that's better used in airbag systems (typ application), and not semi-accurate datalogging? Anyone have any suggestions for any other "non-8-LCC", economical (<40$) accelerometers with a range of +/-25g or greater?
Dave

wrote:
  Quoted Text Here  
I've no experience with the device in question... but would there be online support somewhere? Other manufacturers have manuals and application-notes on their Websites, and you may be able to find a solution to the noise problem in one of them. :-)

wrote:
  Quoted Text Here  
Just reviewing some of the posts, it seems that I posted my suggestion too early... :-)

I am using the MMA2201 for an altimeter I am designing. Should be similar.
You need to have a ground plane under the chip (on the layer that the chip is mounted on). Then all "nc" pins need to be connected to this ground plane.
Decouple the power supply pins well and make sure that your a/d and pic clocks not not running close to the same or a multiple of the MMA internal clock.
This should reduce noise if you are not using the ground plane already.
dave.harper wrote:
  Quoted Text Here  

jdMARS wrote:
  Quoted Text Here  
The data sheet for the MMA3201D clearly states that none of the N/C pins should be connected to anything.
  Quoted Text Here  

Oops, Like I said, I was basing my comments on the MMA2201. I have not seen the 3201 datasheet. Does it say anything about ground planes on the PCB?
jd

Oops,
David, you are absolutely right. Leave the NC pins free but still use the ground plane and connect chip ground and decoupling capacitor ground to this plane as close as possible.
jd

dave.harper wrote:
  Quoted Text Here  
I believe I've located the problem. I tried using the MCP3202 (the ADC) and the MMA3201 (the accelerometer) with a basic stamp, just to see if that would improve anything. The readings were MUCH less noisy (even with the circuit on a breadboard). The variables that changed were the PIC itself, the oscillator used, and the voltage regulator... therefore I looked at these 3 components to see if they might be the source of the noise.
The voltage regulator (purchased from radio shack) as the part number 7805A. A datasheet lookup indicates it was probably one of these: http://www.st.com/stonline/products/literature/ds/2144/l7805ab.pdf http://rocky.digikey.com/WebLib/On-Semi/Web%20Data/MC7800-D.pdf
The voltage regulator used on the basic stamp was PN 1121A5: http://www.linear.com/pc/downloadDocument.do?navId=H0,C3,P1370,D2188
In a nutshell, the voltage regulator I was using provides a line regulation (delta V_out) of 50mV, while the basic stamp's regulator was 10mV max (5 times more stable!). Using the basic stamp's regulated voltage supply with the 16F628 controlling the ADC yields a standard deviation of 2.8 lsb (3.4mV), which is down from 18.5 lsb (22.5mV)... a major improvement.
So it turns out the MMA3201 isn't a noisy sensor after all. Rather my voltage regulator was the source of the problem.
Thanks all for the help.
Dave

dave.harper wrote:
  Quoted Text Here  
Line regulation refers to the change in the output of the regulator when the input changes. Since the input is very stable (a battery) the input change is going to be very small to nonexistant.
The regulator used on the Roctronics hardware (with 16f628, MCP3204, and ADXL150) is the MIC5205. Its stated line regulation is 0.5% over temperature.
  Quoted Text Here  
While the line regulation is almost certainly not the problem, it still might be your regulator. The difference is likely not so much that different parts were used as the Basic Stamp has the regulator on the board and you put together your own from piece parts. Do you have the recommended capacitor on the input terminals? Is the 100uF output capacitor the only one? While it might seem that more is better, you also need a smaller low ESR (a 0.1uF ceramic would do the trick) type for the high frequencies.
Put a 1uF capacitor on the input terminals (as close to the regulator as you can manage) and an extra 0.1uF ceramic on the output and see what happens.
  Quoted Text Here  

David Schultz wrote:
  Quoted Text Here  
My bad, I misread that spec.
  Quoted Text Here  
Yes, I'm positive at this point it has something to do with the regulator I'm using (or rather the way I'm connecting everything). I only had a 1uF cap on the input, which is likely a factor. Also, I'm running off a 9V battery, which might be a little low... (9V can only source a couple hundred mA without a big voltage drop anyway). I'll fiddle with it some more tonight.
On a side note, that is an impressive project you did on the Kalman filter altimeter. I've been entertaining the idea of an apogee detection system using a differential pressure gauge (pitot tube out the nose cone) and a absolute pressure gauge...
Thanks again for the help, Dave