Old-school Vs. MEMS Accelerometers / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Old-school Vs. MEMS Accelerometers

1 2 next

Move Thread

LAN_403

Alucard1137

Thu Apr 27 2006, 04:21AM

Registered Member #128 Joined: Fri Feb 10 2006, 05:02PM
Location:
Posts: 19

I am currently building a UAV for use in medium range (30 to 60 miles from target) aerial photography. The basic avionics package will employ GPS for long term and an IMU for short term positional stability. Long term attitude control will be maintained by an optical horizon sensor and short term attitude control will be derived from the IMU. The IMU will be comprised of four, 3-axis accelerometers arranged in a tetrahedral configuration.

My main concern is drift in the IMU. The accelerations likely to be experienced by my aircraft are going to be relatively low in frequency and amplitude (under 5 Gs), so I am not worried about high-end frequency response or range. Which type of accelerometer would be most accurate in this case? Would it be oldschool mass-spring (double balanced flexure mount with optical readout in my case) or the new, MEMS-based chips? The aircraft is small, but large enough that a few extra grams of weight will not matter.

What factors affect the accuracy and SNR of an accelerometer? My physics background makes me think that, all other things being equal, the reference mass will be the dominant factor in signal to noise ratio (and thus sensitivity and accuracy.) I would guess larger test masses would improve low-end frequency response and have a much higher SNR.

AndrewM

Thu Apr 27 2006, 04:41AM

Registered Member #49 Joined: Thu Feb 09 2006, 04:05AM
Location: Bigass Pile of Penguins
Posts: 362

why is it that we always end up doing the same projects?? I don't know specifically which method provides the best intrinsic SNR, but I'm using MEMS devices with some success.

Heres my setup. Thats a 16F628A for a brain, and you can see a single 2 axis MEMS accelerometer on this board. Also notice the two 150m FM modules at the rear, and the leads coming off which attach to a cellular phone for long range control.

Heres my control software. the 3d model displays the orientation in real time, and the sliders are individual servo controls. the joystick controls channels 1, 2, and 3.

Alucard1137

Thu Apr 27 2006, 05:01AM

Registered Member #128 Joined: Fri Feb 10 2006, 05:02PM
Location:
Posts: 19

Andrew wrote ...

why is it that we always end up doing the same projects?? I don't know specifically which method provides the best intrinsic SNR, but I'm using MEMS devices with some success.

Hahaha! I always wondered if I had an evil twin brother/arch-nemesis. Either that or we're both physics majors with an interest in aviation. Nice setup! What's the drift like that you have encountered so far? Any flights? Got a project website we can marvel at?

Below is a picture of the 1/2 scale model used to test whether or not my design flies (it does, and is quite tolerant of turbulence as intended.) I will have to actually resort to more sophisticated construction of the foamy jet (IE: smooth curves as opposed to the highly angular construction) in the larger model.

Carbon_Rod

Fri Apr 28 2006, 03:21AM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

Donâ€™t they sell MEMS gyroscopes to the public yet?

Perhaps a centrally (cg) located cheap electrolytic tilt sensor with uC shock-corrected anti-sway routines tied into the flight control system may work.

fun =]

Ben

Fri Apr 28 2006, 04:33AM

Vigilatny
Registered Member #17 Joined: Thu Feb 02 2006, 02:47PM
Location: NL
Posts: 158

Wouldn't a MEMS gyroscope just be two parallel accelerometers, space some distance apart?

A decent forum for such discussions(at elast for helicopters) is here:

Also the guys from this company have been know to post there:

AndrewM

Fri Apr 28 2006, 05:22AM

Registered Member #49 Joined: Thu Feb 09 2006, 04:05AM
Location: Bigass Pile of Penguins
Posts: 362

no website for my efforts, yet anyway. In general you seem to be more ambitious than I am; I'm not after an inertial measurement unit, but rather and just achieving orientation control with accelerometers. This will make one-button maneurvers possible so long as they occur in the same plane as the force of gravity; loops, banks, rolls, immelmans, but not simple turns.

again, i don't have a great answer regarding drift; my sensors are only transmitted with 8 bit accuracy. however, off the cuff, i would expect MEMS gyros and maybe even a pitot tube to be more accurate than an IMU as described.

Alucard1137

Fri Apr 28 2006, 07:49AM

Registered Member #128 Joined: Fri Feb 10 2006, 05:02PM
Location:
Posts: 19

Yeah you may be right about that one. My classes are about to end, so I will be able to conduct preliminary tests to know for sure.

Carbon_Rod

Fri Apr 28 2006, 08:33AM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

â€œWouldn't a MEMS gyroscope just be two parallel accelerometersâ€ â€“you could have 6 synchronous sensors and your relative position would still decay in accuracy after a single reading. Accelerometers are not perfect â€“ even the super sensitive adiabatic gas temperature differential bridge design has the curse of compounding errors.

Perhaps even a surplus artificial horizon would give better results. ;o]

AndrewM

Fri Apr 28 2006, 03:17PM

Registered Member #49 Joined: Thu Feb 09 2006, 04:05AM
Location: Bigass Pile of Penguins
Posts: 362

Don't let us discourage you, I want to see if your idea works!

Alucard1137

Fri Apr 28 2006, 04:10PM

Registered Member #128 Joined: Fri Feb 10 2006, 05:02PM
Location:
Posts: 19

No worries, I am far too stubborn to be discouraged. :P The goal is simply to have an IMU that keeps the aircraft stable in between updates from the GPS and horizon sensor (and the chirped ultrasonic altimeter below 100 ft.) This aircraft will be able to take off and land on its own, so attitude and position control becomes critical for the first and last moments of flight (where the horizon sensor will be essentially useless, the GPS updates are too slow, etc.)

After reading more and thinking about it, I think MEMs would suffice in this arena. I suppose I will have to be patient if I want an all-weather UAV. :)

It seems you have a head start- when did you start the UAV? I started thinking about mine around the time of the Model RC/Wireless Video thread.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob