ADXRS MEMS Gyroscope
Posted 1 Oct 2002 at 14:01 UTC by steve
According to a story
on smalltimes.com, Analog
Devices is about to ship their long awaited ADXRS
Gyroscope. The MEMS-based device will sell for as little as $10 -
$30. While targeted at hand-held devices, the low cost makes it
ideal for robot projects (nBot
and autopilot come to
mind). The ADXRS Gryo comes in a range
of resolutions but all are in standard BGA packaging measuring 7mm x
7mm x 3mm and draw 5ma at 5 volts. A library
of technical documents containing example applications of the ADXRS
is available online. The full press
release is also available from Analog Devices.
...with an adapter board. BGA's are a little beyond my prototyping
I found this interesting, from the site:
"the same process that has been used to manufacture over 100 million
accelerometers since 1993."
Would these 100 million accelerometers be found in automobile airbags?
What other existing product uses these?
ZIF sockets?, posted 1 Oct 2002 at 15:53 UTC by steve »
I've seen ZIF sockets for BGA packages but I don't know how expensive
they are. I bet they'll cost as much as the gyroscope! :-(
I haven't had a chance to read through the data
sheet yet, but I've already asked for samples.
The Tokin CG-16D angular rate sensor ("gyro") that we're using on the autopilot 2.2
boards has been End-Of-Lifed by Tokin, so we're looking for a new
solution. It is a great part, but their customers were mostly buying
the L34 3.3 volt unit instead. We could adapt our boards to it, but the
solid state and better noise resistance of the ADXRS looks better.
My one complaint about both of them has to do with their sensitive axis.
The Tokin can do three axis sensing with only two boards since it is
parallel to the mounting plane, but the Analog part will require three
boards since it senses perpendicular to the plane. That will add quite
a bit of cost to our IMU design. I wish they would offer three axis
versions of the gyros and accelerometers to save us the cost and
hardship of orthogonal mounting.
I have put in a request for a sample or two also.
But I do agree that this is going to be tricky to solder.
My idea is to make a small PCB for the parts.
Where the pads are underneath, I plan to have .020 plated through vias
to the other side with pads as well. Then "somehow" holding the chip
accurately on the PCB, heat up the pads on the other side of the pcb,
and with a bit of solder allow it to flow through the via and solder
the pins on the chip. I'll be using a dab of liquid flux for this, but
only on the pads not under the chip itself.
One problem is going to be hard to eyeball if the pads and measurements
are correct as the little pins are under the chip itelf and not that
close to the edge.
I wish they'd keep the TSOP or TQFP packages.
I hope they do have a eval board for this like the accellerometers.
Well, the normal way to do it is with reflow or wave soldering.
Basically, add solder, place component, heat - the surface tension of
the solder will move the chip into the exact position.
I suggest trying the following:
*) design a board with the correct pads
*) cover the board with lots of flux spray or liquid flux or something
*) using either low temperature solder (silver?), solder paste or - with
a lot of effort - normal solder shavings, cover the pads with a little
*) Place the component as accuratly as possible
*) put the whole lot in the oven set to an appropriate temperature for
about 5 mins!
For normal solder I think you would need 300 degrees but I don't think
ovens go up that high?! Mine does 270 I think which might be good
enough. With silver solder or solder paste I think the temperature is
lower so you may have better luck :)
I received my sample, and .020 pads are too big, so I'll be trying the
heat it up and hope it sticks method, as mentioned above.
I happen to have a hot air SMD rework tool, so I think that'll heat up
Those little .020 in size BGA pins are tiny. They aren't making this
easy for us anymore.
If this keeps up, we'll have to figure out how to glue down dies and
do the wirebonding ourselves next.
Hmmm, hope somebody takes this as a hint and develops a reasonably
low cost wirebonding machine for us.
I was successful at mounting my Hyro chip on a PCB. Plus the circuit
works too. Really, really neat.
I had a PCB layout done for me by ExpressPCB.Com, I then applied some
flux onto the PCB and carefully placed the chip. Then I used a hot air
rework station to carefully heat up the stuff and solder it down. Of
course you have to carefully hold the chip in place to prevent it
moving as the hot air flows under and around it. It took about three
seconds to heat it up real nice. You could tell as the solder started
flowing as the chip would drop down ever so slightly.
I am planning a project submission for the DPRG website detailing my
PCB layout for a combo low cost Gyro/Accelerometer system on one PCB.
Of course low cost is relative. Right now the Gyro chip costs $39.90 in
single units, and the Accelerometer costs about $29.90 in single units
as well. So we have about $20.00 for the PCB (piggy back with other
layouts), $70.00 for the two chips, plus maybe about $10.00 for other
assorted parts like capcitors, inductor, regulators, connectors, etc.
So it looks to be about $90.00 to build one up. But then a while back
another company wanted something like $4500 for a gyro chip setup. So
in that case it is low cost.
But it is really cool, works nice, the gyro is real fast and very
sensitive to small tilt changes. The Accelermeter allows you to
compensate for the creeping gyro errors that occur in things like a two
wheeled balancing robot.
I love it when things actually work. It is so hard to debug and fix
things in the unknown area.