At 06:31 PM 10/17/99 -0400, you wrote: >At 03:17 PM 10/17/99 -0700, you wrote: >>The ICsensors (now EG&G) can go for <$100, and I've got one going on the >>space station with a gain of 1.9 Million to one--giving 200 micro-G's full >>scale. Thats good and cheap. Had some real problems getting there, >>though. A lot of 60 cycle stuff kept showing up--tried shielding and >>all--no effect. Finally stuck the sensor in a foam lined box, and noise >>went away--was acoustically coupling off the air conditioner motors on the >>roof. >> > >Just found it! Which one did you use to get 200uG full scale? We are >looking to make an INS that has no greater error than 0.5mG,but less would >be great! They really only have one device that would be applicable--it is a 2-G sensor in a package roughly 5/8" square by 3/16" thick. Model is 3022 and 3028 for uncompensated, and 3052 for compensated. Uncompensated have lowest noise, BUT--and this is a BIG BUT--about 0.3% per degree gain and zero drift! Ideal solution is to put the accelerometers in a thermally controlled block--just need a 1-2 watt resistor as a heater element, and a power op-amp and thermistor to control the temperature to 30-50 degrees above ambient. 3058 has internal compensation and reduces the error down to a few percent over the full temperature range--but at the expense of noise. Specs are on their web page. Alternate method is to provide a integrating feedback on the instrumentation amp so it serves as a hi pass filter with a time constant for 100 seconds or so (See burr brown's web page for their app-notes on how to do this)--this will take care of zero drift rather nicely, but leaves the larger gain drift. Math in the pic can deal with that--as long as you have a temperature sensor near the accelerometers too. We do this with our earthquake alarms in a PIC711--plus a lot of other signal conditioning. ALSO do not trust the calibrations from ICS if you require high accuracy! They are Ok for a few percent--but have proven to be very unreliable. A 2-G sensor is easy to calibrate yourself at DC levels--just remember the earth has a 1 G field--just rotate the device through 180 degrees and you have +/- 1 G available. Cosine of the angle in between. Enjoy Kelly William K. Borsum, P.E. -- OEM Dataloggers and Instrumentation Systems & San Diego, California, USA