Sorry to reply late to a couple of topics, but I've been either off the net or very busy the past few days.
--
Ring laser gyros measure changes in angular orientation, and, as previously mentioned by JD-EE, are commonly installed in a "strap-down" configuration (always aligned with the axes of the aircraft or other platform), rather than, as in the older precision inertial nav systems, with the mechanical gyro mounted on a plate having 3- or 4-axis motor drive to keep the gyro erect in an inertial frame, while the aircraft (missile, whatever) rotated around it. That difference is mainly an interesting point, a laser gyro could be installed either way.
Ring laser gyros that have been incorporated into an inertial navigation system (with suitable mathematics), are used, along with accelerometers, to provide the short-term attitude (actually a very long, short-term
, since the laser gyros have such low drift), while the accelerometers provide long-term attitude (gravity reference) and accelerations along the axes to integrate into position. (The latter - double integration of accelerations - is where a lot of drift in INS systems comes from, and thus the requirement for some sort of aiding; in a modern aircraft, that aiding would come primarily from GPS and the altimeter).
That is all background for a discussion of tumbling. Tumbling of a mechanical 3-axis gyro (which trys to stay aligned in an inertial frame), occurs when two of the gimbals supporting the gyroscope happen to align, a state known as gimbal-lock, which can physically yank the gyro's spin axis off its original alignment to some other direction. This is bad because it takes a long time, using other aiding, to re-erect the gyro. More sophisticated systems may use 4-axis gimbals, which are steered so that no two ever align.
A strap-down system can suffer a related effect if the mathematical attitude solution involves only the three axes: roll, pitch and yaw. The problem occurs when pitch becomes near vertical and yaw becomes indeterminate. The vehicles that I have worked with are incapable of approaching pitch of +/-90 degrees, and so I cheat and use only roll, pitch and yaw. I'm not sure if, in an INS, this might cause loss of short-term attitude accuracy, or if it will recover a correct indication as soon as the platform leaves the vertical. However, this should never happen, because there is a mathematical construct known as "quaternions" that avoids this problem (don't ask me, I don't know
); I suspect that it is a mathematical equivalent to the 4-axis mechanical solution. Hopefully, any aircraft INS is thus designed to avoid mathematical "gimbal-lock".
One more point about laser gyros in INS systems: they allow the system to perform the function of a gyro-compass. After a short alignment period, using the accelerometers and the rotation of the earth, an INS not only knows which way is up (down) but also the direction of true north. The laser gyro INS systems that I am (a little) familiar with acquire true north to an accuracy of a few tenths of a degree within 10 minutes of a cold start, as compared with 6 hours for a mechanical gyro-compass typically used on a ship. This alignment does not drift, even when the platform is in motion.
--
CSMU:
I will begin by confirming that I know nothing about the design of any CSMU in flight recorders, except what I have read on the web. However, I do know a lot about designing electronics for the deep ocean.
Some of the prior discussion, about whether or not to open the recorder, and how water might be detected inside, seems to be "misplaced". It was made clear by BEA at the press conference held the day the recorders arrived in Paris, that the following steps would be performed: open the case, wash the memory board in deionized water, dry the board in a oven, microscopically inspect the board (and, if I recall correctly, repair any obvious damage), and only then try to read the memory.
I do not think that CSMUs are designed the way many posters think, nor the way I would design a typical deep-sea system. Without quoting the dozen or so pages I found (sorry, hurrying), I'm not convinced that a CSMU is actually intended to be water-tight in the sense that a typical pressure housing is sealed and has pressure-proof penetrations for wires. CSMU designs appear to vary considerably between vendors, and are kept as trade secrets or may alternatively be patented.
In the descriptions I have seen, the outer shell, stainless or titanium, is not thick enough (at 1/4 inch) to act, alone, as a pressure housing at 6000 meters depth (20,000 psi, sorry for the mixed metric/English units, I use the MKSA system, but still have not internalized Pascals
). It would collapse but for one or both of two mechanisms: either it is designed to leak (not sealed at all) and/or it is internally supported by incompressible materials.
The next layer inward is typically thermal insulation. One reference identifies this as "silica" which is a mineral and is possibly packed densely enough to support the shell, but I wonder how good its insulation properties would be if it were packed that hard.
The next layer is the thermal mass (works in conjunction with the insulation to limit the temperature inside. In some designs, this is identified as paraffin, with the phase change providing the primary heat absorption. This opens the possibility that the paraffin is also used as a water block.
The description of the Honeywell data recorder, which has been previously linked in this forum, mentions that their CSMU "uses modular "dry-block" materials for both the insulating liner and thermal mass, there is no need to deal with the sticky thermal gels or special insulating fluids." This implies to me that it is common to protect the memory boards from water with gel or fluid, rather than using an enormous increase in weight to make the CSMU a traditional pressure vessel. It is also possible that I am wrong about that, and the gels/fluids are secondary protection; in which case, the outer shell must be water tight and adequately supported internally.
It is possible that the boards are conformal coated with something that will resist water for a long time. I am speculating about that; I did not find that mentioned in any reference.
Note well that it is
not necessary to protect the memory chips from pressure, only (possibly) from water. Epoxy encapsulated integrated circuits (the most common and inexpensive packaging) do not have voids, and so there is no space to implode, which would subsequently damage the die (circuit) inside. [There are some special devices, such as solid-state accelerometers and pressure sensors, that require a void space to operate, and thus they are not pressure tolerant, beyond the strength of the case.]
While we normally deploy complex electronics in pressure cases, we have also designed and tested numerous "pressure-tolerant" circuits that operate in an oil filled space that is equalized to ambient pressure. Appropriately packaged semiconductors; ceramic, mica, and some solid tantalum capacitors; most resistors; carefully characterized inductors (core materials can change properties with pressure); ordinary wire and connectors; and many other components; all work fine at pressures of 20,000 psi and even greater.