You are not dampening a response, and it is not incorrect leveling. The platform will always oscillate. That's the nature of a feedback circuit. The idea is just to subtract a known (error) oscillation from the incoming signal.
What you are doing is deducting a very low amplitude/low frequency signal from the input. This signal is (or rather should) correspond to the oscillation of the platform. The signal is generated internally in the INS/IRS computer, based on a math-model of the system.
Measuring raw input from the platform when an aircraft is parked would tell you that the aircraft is moving forth and back (both N/S and E/W) with an 84 minutes oscillation period. Basically it would tell you that you where driving around in a circle ;-). The INS/IRS computer has a model of this signal internally and deducts it from the input signal. If the modeled signal is a perfect match to the actual oscillation of the platform, you'll have a perfect INS/IRS, that tells you that you are actually parked when you are parked ;-)
This compensating circuit is working both in platform and strap-down units. It's just easier to imagine using a platform. but the accelerometer errors exists in both scenarios.
To add confusion, some models of the INS actually fed the compensating signal to the platform leveling system in stead of deducting it from the input. This will cause the platform to hold still even in parked scenarious. I guess that computers where to slow these days to handle all the calculations, so the more that where done mechanically the better.
Both approaches will work.
EDIT:
The best explanation out there I have found is a NASA document describing low orbiting satellites. They have an initial paragraph on schuler. Very good stuff..
I am apparently to stupid to figure out the link to the PDF, but search for "SCHULER PERIOD IN LEO SATELLITES ", and you will get it.
Last edited by lasseb; 6th Aug 2012 at 12:33.
Reason: Added link