Q
if we have the lower noise parts on the roadmap to compete, their noise density
is 50ug/Hz, while our best part ADXL103/203 is 110ug/Hz. (compete with
LIS244ALH)
A
comment on noise if Tilt measurement is the concern.
In principle customer has to look at four things
- Temperature coefficient of the sensor
- Frequency range it is being operated (i.e for 2 Hz at 0.05 degree, noise
density requirement is 500µg/root Hz)
- Scale factor (this decide what is the ADC needed)
- Power consumption, if this is used in portable devices.
With respect to making a tilt sensor the LIS344ALH is, to be quite frank, There
is no way it is up to the job.
Low noise is not the problem (I'll go through the math later) - the tempco is
just too big!. If you look at ST's own data you can see that more than half the
parts have a tempco greater than 0.6mg/°C in the X and Z axes. So if there is a
temperature variation of as little as ±15°C from where calibration happened the
error from this parameter alone would be equivalent to more than 0.5° of
tilt!!! You can forget about 0.05°...
Noise is not an issue. For a level, bandwidth of 2Hz is more than enough. If
you need a resolution floor of 0.05° the noise density required would be:
0.05° tilt = SIN (0.05°) = 0.9mg
At 2Hz, noise density required = 0.9mg / SQRT (1.6* 2Hz) = 500µg/root Hz
You can see that any of our low g accelerometers is low enough noise.
What about the converter needed to measure 0.05° of tilt? ST has a scale factor
of 660mV/g, so 1mg would be 660nV. You need a 13 bit converter to resolve this.
Is there one available?
Quite frankly, they are just seduced by what appears to be a low price. They
will believe anything someone with a low enough price will tell them. My
suggestion is that you have them look at the XL345. They will get similar
performance using some oversampling. They can eliminate the A/D converter. It
will use much less power. I'm sure Neil can help them with whatever design-in
work is necessary.
Once we have even lower tempco parts (XL350 and Quincy) the performance can be
improved further still.
With respect to the concern about power - The LIS344ALH consumes more than
700µA!!!!!!! If he cared about power he would be looking at a different device.