Maker Pro
Maker Pro

Do accelerometers drift?

J

Jim Thompson

Jan 1, 1970
0
I'd like to build a 3 axis tilt sensor.

I would like to know if low-cost, hobby-caliber, solid-state accelerometers
drift. If you set one on a table do they register +1g all day long
consistently?

Does temperature affect them?

Even government grade accelerometers drift... it's the nature of the
beast. Most control systems augment the accelerometer with GPS or
other locating devices. The accelerometer is only of short term,
fill-in-the-data-points, value.

...Jim Thompson
 
H

H.J.

Jan 1, 1970
0
I'd like to build a 3 axis tilt sensor.

I would like to know if low-cost, hobby-caliber, solid-state accelerometers
drift. If you set one on a table do they register +1g all day long
consistently?

Does temperature affect them?
 
R

Rob

Jan 1, 1970
0
H.J. said:
I'd like to build a 3 axis tilt sensor.

I would like to know if low-cost, hobby-caliber, solid-state accelerometers
drift. If you set one on a table do they register +1g all day long
consistently?

Does temperature affect them?


Hello HJ, I can't offer a reply to your question, but I am interested in
finding a source of low cost accelerometers - if you don't mind, I would be
interested in finding out what you are using and who manufactures them.
Others are most welcome to reply / comment as well. Thanks

Regards
Rob
 
G

Glen Walpert

Jan 1, 1970
0
Even government grade accelerometers drift... it's the nature of the
beast. Most control systems augment the accelerometer with GPS or
other locating devices. The accelerometer is only of short term,
fill-in-the-data-points, value.

...Jim Thompson

True enough, but he was asking about tilt sensing which is a bit less
demanding than inertial navigation. While low cost MEMs
accelerometers of the sort used in every airbag recentlly manufactured
do have high thermal drift, high unit to unit variation, and not the
best shock resistance, they are suitable for use in low to moderate
accuracy tilt sensors, and once calibrated they will "register" +1g
all day long. Perhaps not +1.000g all day long, unless you are using
one of the better MEMS accelerometers like the ADXL105 with good
filtering, individual calibration and temperature compensation, and a
fairly stable environment.

Unless the OP wants high accuracy (in which case he should look at
electrolytic tilt sensors) he should give it a try. The mfgrs app
notes on tilt sensing are honest about the accuracies which can be
achieved, suggest you read them.

BTW I once dissasembled an inertial navigation accelerometer made in
the days before GPS, which used force balance inside a constant
temperature oven, very clever mechanism but obviously very expensive
too.

Glen
 
L

Leon Heller

Jan 1, 1970
0
Rob said:
Hello HJ, I can't offer a reply to your question, but I am interested in
finding a source of low cost accelerometers - if you don't mind, I would be
interested in finding out what you are using and who manufactures them.
Others are most welcome to reply / comment as well. Thanks

I've used the ADXL202 (Analog Devices). They work OK, and you can get free
samples.

Leon
 
V

Vladimir Vassilevsky

Jan 1, 1970
0
Glen said:
True enough, but he was asking about tilt sensing which is a bit less
demanding than inertial navigation. While low cost MEMs
accelerometers of the sort used in every airbag recentlly manufactured
do have high thermal drift, high unit to unit variation, and not the
best shock resistance, they are suitable for use in low to moderate
accuracy tilt sensors, and once calibrated they will "register" +1g
all day long. Perhaps not +1.000g all day long, unless you are using
one of the better MEMS accelerometers like the ADXL105 with good
filtering, individual calibration and temperature compensation, and a
fairly stable environment.


ADXL105 is a piece of junk in terms of accuracy and stability, as well
as other accelerometers of AD. It is really sensitive to mechanical or
thermal shock also.
The best you can get from it is about 0.05g accuracy.

VTI Hamlin does better MEMS accelerometers, though you will not get any
better "true" accuracy then 0.01g.
Unless the OP wants high accuracy (in which case he should look at
electrolytic tilt sensors) he should give it a try.

Electrolytic sensors are very gentle. They can't take heat, cold, or
vibration.


Vladimir Vassilevsky, Ph.D.

DSP and Mixed Signal Design Consultant

http://www.abvolt.com
 
H

H.J.

Jan 1, 1970
0
So if Accelerometers do drift, what do they drift to?

If one at rest initially registers +1g, does it drift to +4g after an hour?
and then go to -6g four hours later? i.e. it drifts in some random range

or

does it start at +1.000 g and then slowly tick up to it's max g limit after
some long period of time?

or ... what else?
 
O

onestone

Jan 1, 1970
0
Vladimir said:
ADXL105 is a piece of junk in terms of accuracy and stability, as well
as other accelerometers of AD. It is really sensitive to mechanical or
thermal shock also.
The best you can get from it is about 0.05g accuracy.

Crap. It isn't exactly rocket science to insulate them from both sudden
thermal changes or sudden shock, and long term monitoring shows that
handled correctly their stability is fine.
 
V

Vladimir Vassilevsky

Jan 1, 1970
0
H.J. said:
So if Accelerometers do drift, what do they drift to?

If one at rest initially registers +1g, does it drift to +4g after an hour?
and then go to -6g four hours later? i.e. it drifts in some random range

or

does it start at +1.000 g and then slowly tick up to it's max g limit after
some long period of time?

or ... what else?

MEMS accelerometers exibit large zero drift. The gain remains pretty
stable though.
Zero drift is quite random and very different from unit to unit.
The other problem with MEMS is the mechanical hysteresis.

Vladimir Vassilevsky, Ph.D.

DSP and Mixed Signal Design Consultant

http://www.abvolt.com
 
V

Vladimir Vassilevsky

Jan 1, 1970
0
Crap. It isn't exactly rocket science to insulate them from both sudden
thermal changes or sudden shock, and long term monitoring shows that
handled correctly their stability is fine.

Why didn't you tell me any numbers? Nothing to say?
Have you ever tried MEMS other then AD?
FYI, ADXLs are the sloppiest sensors ever.


Vladimir Vassilevsky, Ph.D.

DSP and Mixed Signal Design Consultant

http://www.abvolt.com
 
O

onestone

Jan 1, 1970
0
Vladimir said:
Why didn't you tell me any numbers? Nothing to say?

What, numbers to match those you supplied? Jeez, what a prune. And yes,
I've been using ADXL's since 1990/1 before they were commercially
available. I've also used several other types of accelerometer, and a
variety of other MEMs devices. I've been involved with micromachined
products since the mid 80's, both in development and as a user.

Have you ever tried MEMS other then AD?
FYI, ADXLs are the sloppiest sensors ever.

FYI your opinion only, don't blame the tools.

Al
 
N

netsurfr

Jan 1, 1970
0
If you're looking for a good 3-axis accelerometer packaged as a
standalone IC, try the Measurement Specialists ACH-04-08-05. It
measures in all 3 axis of movement. The 3 circuits operate
independently and look like JFETs. A typical circuit is shown on the
data sheet.

I've been working on circuits that interface to MEMS accelerometers
(and other sensors) for a while, and this one has good specs. I don't
know about price though. This one might be expensive. I recommend you
call them at 610-650-1500 and get a quote. I attached the drawing in
case you want to look it over first. Their web site is at
http://www.msiusa.com

netsurfr
 
N

netsurfr

Jan 1, 1970
0
If you're looking for a good 3-axis accelerometer packaged as a
standalone IC, try the Measurement Specialists ACH-04-08-05. It
measures in all 3 axis of movement. The 3 circuits operate
independently and look like JFETs. A typical circuit is shown on the
data sheet.

I've been working on circuits that interface to MEMS accelerometers
(and other sensors) for a while, and this one has good specs. I don't
know about price though. This one might be expensive. I recommend you
call them at 610-650-1500 and get a quote. I would attached the
drawing but this newsgroup does not allow binaries. The web site is
at http://www.msiusa.com

netsurfr
 
O

onestone

Jan 1, 1970
0
They were around the same price as ADXL202's, but are much higher ranged.

Al
 
Top