2-axis accelerometer in SOIC package?

[Futterama]

Hello forum,

I need a 2-axis accelerometer in a manual solderable package, like SOIC (I don't have access to reflow soldering, I only have my soldering iron). The QFN and LGA packages can't be hand-soldered as far as I know (at least not recommended).

The device will be mounted on a gasoline-driven RC car (for measuring car acceleration) with lots of vibration and that's why I don't trust a hand-soldering job of the QFN and LGA packages because the manufacturers recommend not to (at least freescale recommend not to use hand-soldering).

I know freescale has a 1-axis accelerometer in 16-lead SOIC (MMA1260EG) and I could use this, but the car tends to make wheelies, and then the static acceleration of the earths gravity will corrupt the reading, so I need that extra axis to compensate for that. I could use two MMA1260EG's but it would be nice to avoid all that extra work.

The possibility of free samples of the 2-axis SOIC accelerometer would also be nice.

Any ideas?

 

[dknguyen]

If you extend the pads so they are longer than the pad, you can hand solder QFN packages. But I have never seen any MEMs sensors in SOIC packages.

What are you trying to do? Measure your car's acceleration? The wheelie would corrupt your reading with two sensors unless you had some crazy filters (or at least veery well experimentall tested filters) to separate gravity from acceleration. You could also use a gyro as a dynamic tilt sensor to work under dynamic conditions, resetting it to level using the accelerometer tilt measurement whenever the gyro gave 0 reading (I assume you can't balance your car in a non-horizontal position so...)

Can you just assume no slip and measure wheel RPM?

 

[Futterama]

I also thought about extending the pads, but freescale has a special PCB land pattern guideline:

"Signal traces with 0.1mm width and
min. 0.5mm length for all PCB land pad near package
are recommended as shown in Figure 8 and
Figure 9. Wider trace can be continued after the
0.5mm zone."

But you think I can safely ignore this and just extend the pads?

Regarding the measurement, yes, I would like to measure my car's accelerarion. The powerful 26ccm gasoline engines on my type of car will easily spin the wheels even on a dry road, so I really need the accelerometer to get the real acceleration.

Are you sure it's not possible with a 2-axis accelerometer? Did I not think this through? I'd better make some drawings because this is really hard thinking for me right now, trying to determine the effect on the accelerometers when the car is not level...

 

[dknguyen]

If you were planning to use the two-axis to detect incliniation so you could remove the gravity component from the acceleration, then yeah you can't do it without some very trial-and-error filters. THe problem is that accelerometers can't measure inclination in dynamic conditions becase the acceleration of the vehicle gets compounded into those readngs- you have a catch 22 problem see. You need tilt readings (via gravity) to remove gravity to get horizontal acceleration, but you need to remove horizontal acceleration to get tilt readings (via gravity). Add on a gyro- they drift over time, but your car probably won't remain non-horizontal long enough for it to matter- just reset the tilt whenever the car gets level again (ie. when the gyro reading is zero) using the accelerometer's tilt readings (though this may require the car to stop to remove the horizontal accelerating component).

How about a premounted surface mount accelerometer?
https://www.sparkfun.com/commerce/categories.php?cPath=23_80

 

[Futterama]

What kind of gyro are you talking about? Is there some kind of solid-state gyros out there?

The premounted accelerometers are pretty expensive :-/

 

[dknguyen]

Yeah, but the discrete ICs and PCB cost just as much if I'm not mistaken.

There is only one gyro I know of that is flat mounted and measures rotation in the IC package's x or y-axis when mounted horizontally, all others measure the rotation about the package's z-axis You can either:
-use a horizontal PCB with a small vertical mounted PCB with a z-gyro on it to provide dynamic tilt sensing (not good for vibration obviously)
-use a horizontal PCB with the one x-y axis gyro
-use a z-axis gyro on a single vertical PCB (but then you may need an accelerometer that can measure acceleration it's z-axis as well so it could also mount flat...probably a 3-axis accelerometer since I don't know any single axis accelerometers that measure their z-axis acceleration).

The IDG200 is the one that can measure rotation about the X-Y axis when mounted flat.
https://www.sparkfun.com/commerce/categories.php?cPath=23_85

But what if you just not measure the acceleration during wheelies? Get one 3-axis accelerometer (you only need the axis in the forward and downard directions but you won't find just those two in an accelerometer unless you flip a 2-axis accelerometer or the PCB it's mounted on vertically), and reject all acceleration readings of the forward axis (ie. the forward acceleration, duh!) if gravity (the reading of the vertical axis) does not equal ~9.81m/s^2?

 

[Futterama]

Sorry, I can't seem to find the IDG200 anywhere...

 

[dknguyen]

IDG300, my mistake.

 

[Futterama]

Well, it seems to me, that using a gyro will be too expensive (no free samples...) so I'll just have to manage with a accelerometer. And the user of my project will have to avoid wheelies. Also, the cars are only rear-wheel-driven, so I could measure RPM on the front wheels in addition to the accelerometer.

 

[Helder Ferreira]

Look in the ELEKTOR magazine, they have a project in wich the supply a sample 2 axis accel. factory soldered to a small pcb

 

[Pommie]

Why not use one of the breakout boards from sparkfun. Link.

I made my own.

Mike.