I should have boards in about a week and a half. I'll order up parts tomorrow.Here's a pic of the top side and the parts list.UPDATE: I added a link to the updated schematic. The main changes to the original are adding an altimeter and airspeed sensor and switching from the Freescale MMA7260 Tri axis accelerometer to the Bosch BMA145. The Bosch is lower cost while having better noise and accuracy specs.Brian
Hey bcr,
I think I was comparing the BMA145 price from Digi-Key to the MMA7260Q from Sparkfun. I hadn't previously noticed the MMA7260Q was sold by Digi-Key although they added a T to the end of the part number which I guess is just the RoHS version of the part. Sparkfuns price is considerable higher.
I'm sure the MMA will work fine in your design, it's not like the BMA is night and day better. The MMA also goes to 6G instead of just 4G for the BMA. That might be important to you depending on your application.
The MMA7260QT is lower cost than the BMA145 at digikey, except in qty 1000+, where the BMA has a <$10 advantage on the $3000+ order. I agree on the other advantages you mentioned though. The BMA145 is smaller (4mm!) and generally has better specs. Figures that something better is out since I just fabbed boards with the MMA
Flar,
I'll post the updated schemo tonight when I get home. The main difference is I switched to a Bosch 3 axis accelerometer that was cheaper and had better noise, accuracy and linearity specs than the Freescale device. Also after starting the board layout and seeing that the size wasn't going to get too far out of hand I decided to put the altimeter and airspeed sensors on board instead of having a connector for off board sensors. The board size is 2" x 3.4"
Brij,
There are indeed anti aliasing filters. I believe they are necessary to prevent problems in high vibration situations. Either that or you have to significantly over-sample to get above Nyquist, in essence relying on the inherent roll off of the sensor and AD converter. I'd rather spend $0.05 for a simple RC filter and not burden the processor. I believe they are set for about 50Hz, but they are easy to change and may need to be raised for certain applications.
As far as the number of poles, I opted for just a single pole and plan on sampling at four times the cutoff frequency to make sure the filter has rolled off a bit before hitting the Nyquist frequency. It's sort of a compromise between no external filter and a multi-pole filter set right at the Nyquist frequency. The reason I try to avoid multi-pole filters unless absolutely necessary is the added cost and board space as well as the increased phase shift that occurs as you approach the cutoff frequency. The phase shift can lead to complications that need to be accounted for in the control loop.
Comments
I think I was comparing the BMA145 price from Digi-Key to the MMA7260Q from Sparkfun. I hadn't previously noticed the MMA7260Q was sold by Digi-Key although they added a T to the end of the part number which I guess is just the RoHS version of the part. Sparkfuns price is considerable higher.
I'm sure the MMA will work fine in your design, it's not like the BMA is night and day better. The MMA also goes to 6G instead of just 4G for the BMA. That might be important to you depending on your application.
B
I'll post the updated schemo tonight when I get home. The main difference is I switched to a Bosch 3 axis accelerometer that was cheaper and had better noise, accuracy and linearity specs than the Freescale device. Also after starting the board layout and seeing that the size wasn't going to get too far out of hand I decided to put the altimeter and airspeed sensors on board instead of having a connector for off board sensors. The board size is 2" x 3.4"
Brij,
There are indeed anti aliasing filters. I believe they are necessary to prevent problems in high vibration situations. Either that or you have to significantly over-sample to get above Nyquist, in essence relying on the inherent roll off of the sensor and AD converter. I'd rather spend $0.05 for a simple RC filter and not burden the processor. I believe they are set for about 50Hz, but they are easy to change and may need to be raised for certain applications.
As far as the number of poles, I opted for just a single pole and plan on sampling at four times the cutoff frequency to make sure the filter has rolled off a bit before hitting the Nyquist frequency. It's sort of a compromise between no external filter and a multi-pole filter set right at the Nyquist frequency. The reason I try to avoid multi-pole filters unless absolutely necessary is the added cost and board space as well as the increased phase shift that occurs as you approach the cutoff frequency. The phase shift can lead to complications that need to be accounted for in the control loop.
Brian
Flare