High-performance gyroscopes have traditionally been exclusively made with non-MEMS technologies such as fiber optic gyroscopes (FOGs) and ring laser gyros (RLGs). However, MEMS based inertial sensor technologies are continually improving and now increasingly being used in aerospace and defense applications. We recently spoke with Dr. David Arch, Marketing and Product Manager for inertial sensors at Honeywell Aerospace. In this detailed interview, David reviews the most important performance metrics for gyroscopes, outlines the current status of MEMS gyro development and discusses ongoing and future trends. Dr. Arch also provides size and cost comparisons of the latest MEMS gyroscopes versus FOGs and RLGs.
MEMS Investor Journal: What are the most significant performance metrics for high-end gyroscopes?
Dr. David Arch: For gyroscopes, the main performance metrics are angular random walk, bias stability, bias repeatability and scale factor stability.
MEMS Investor Journal: Can you briefly explain what each of these mean in layman's terms?
Dr. David Arch: Angular random walk (ARW) is a measure of the noise of the device (lower is better); this is usually measured in units of degrees per root-hour.
Bias stability is a measure of how stable the output of the gyro is (it’s important that it always gives the same output at the same input rate); this is measured in degrees per hour or degrees per second.
Bias repeatability is a measure of how good the output is every time you turn the gyro on (you want it to be the same each time); this is measured in degrees per hour or degrees per second. Bias repeatability is a harder spec to meet. Turning the device on and off and on many times and achieving a good repeatability requires very stable devices and a good control over the thermal, mechanical, and electrical characteristics of the device. In contrast, bias stability is measured after the device is turned on and for a particular length of time.
Scale factor stability is a measure of how good the gyro output is versus rotation input (you want this to be the same every time you measure rate); this is usually measured in parts-per-million or percent from linear.
MEMS Investor Journal: How do MEMS gyros currently compare with non-MEMS based technologies such as FOGs and RLGs?
Dr. David Arch: Making such comparisons is difficult because devices are designed for particular applications. If one is to look at ultimate performance, however, RLG and FOG technology is a factor of 1000 times better in terms of bias stability and bias repeatability compared to MEMS. ARW for MEMS is a factor of ten to hundred times greater than RLG or FOG.
MEMS Investor Journal: How do costs compare? What is the typical cost range for high-end MEMS based IMUs? What about for RLG or FOG based IMUs?
Dr. David Arch: Again, this is difficult to compare because of performance differences. A good navigation grade IMU (FOG or RLG) will be in the $50K - $75K range. Tactical grade FOG or RLG IMUs are in the $20K range. MEMS IMUs are still poorer in performance than these tactical IMUs and in the range of $10K.
MEMS Investor Journal: What about sizes? How does the overall component size compare for MEMS vs. RLG and FOG IMUs? What are the typical size ranges?
Dr. David Arch: MEMS IMU sizes vary considerably -- from less than a cubic inch to 20 cubic inches. The best performance MEMS IMUs are not in the smallest packages, at least not yet. RLG IMUs are as small as 33 cubic inches but the higher performing ones (like FOGs) are in larger packages, primarily because optical gyro performance gets better with larger sizes.
MEMS Investor Journal: Could you please briefly describe the main applications and the approximate specifications range for each?
Dr. David Arch: Applications for gyros range from motion sensing, platform stabilization, attitude and heading to precision navigation. Each application has different gyro requirements. Motion sensing may have requirements of several degrees per second bias stability, while precision navigation will have bias stability requirements of 0.001 degrees per hour and 0.001 degrees per root-hour.
The technology used will be that which meets these requirements including cost. MEMS gyros right now are approaching 1 degree per hour for bias stability for the best devices so they do not meet precision navigation requirements. RLGs and FOGs do meet precision navigation requirements.
MEMS Investor Journal: For which applications are MEMS inertial sensors still not good enough?
Dr. David Arch: Navigation for extended periods of time (for example, greater than 5 minutes) where long term stability is critical. Also, accurate pointing applications where ARW is important.
MEMS Investor Journal: So, to clarify, is precision navigation the only application that cannot currently be accommodated by MEMS based IMUs?
Dr. David Arch: It depends what you mean by precision navigation. Even if you need moderate stability but you need it for long periods of time, MEMS is not there yet. MEMS bias stability looks pretty good for short periods of time but starts to drift after 5-10 minutes. If you were trying to track a first responder or a soldier for any length of time, you could not do it because of this drift.
MEMS Investor Journal: For navigation, what does the bias stability have to be to go beyond the extended periods of time (for example, greater than 5 minutes)? Also for navigation, are there other specs that are important?
Dr. David Arch: It depends on how long and what application. If GPS is utilized, that alters the requirements. One really has to look at the application to provide a clear answer. Even with short periods of time, there are applications where one needs more stability that what MEMS can provide.
MEMS Investor Journal: Can you provide 2-3 application examples along with the typical IMU performance requirements?
Dr. David Arch: Aircraft navigation needs bias stability better than 0.01 degrees per hour for extended periods of time, even with GPS (you must be able to navigate accurately if and when GPS is unavailable which happens more than people imagine). For an attitude heading applications, the IMU must do 0.05 degrees per hour, or thereabouts, if one does not rely on other aides. For tracking a first responder or soldier for any period of time (say 30 minutes) you will require less than 0.1 degrees per hour.
MEMS Investor Journal: For accurate pointing applications, what is the lowest ARW MEMS gyro that you are aware of and who makes it?
Dr. David Arch: ARW for MEMS devices is in the 0.1 degrees per rt-hr range -- Honeywell’s gyros are in this category and several other vendors are approaching this value.
MEMS Investor Journal: Are there accurate pointing applications where the much higher ARW of MEMS based IMUs is not good enough? If so, which specific applications?
Dr. David Arch: There are many applications -- if you are trying to find true north, for example, for an aircraft attitude and heading reference system, or for a marine vessel requiring good heading.
MEMS Investor Journal: Is there a fundamental limit for MEMS inertial sensor performance due to the small proof mass in MEMS devices?
Dr. David Arch: Intuitively I would say yes, although thermal limits say that MEMS could be much better. Packaging, surface effects, fabrication tolerances, and electronics are all first order effects with MEMS and contribute to errors in the MEMS sensors. This is much less the case for RLGs.
MEMS Investor Journal: Which one of these factors (packaging, surface effects, etc) is the most limiting one for the current generation of MEMS devices and why?
Dr. David Arch: All of these factors are being worked by most MEMS developers. These are all very difficult challenges and become more challenging as you try to get better performance. The electronics may be the most challenging because we are attempting to measure very, very small signals (and reducing the size of the devices makes this even more difficult) and there are many noise factors that confound the measurement. As we push performance, the signals get even smaller. Packaging and surface affects introduce more noise. Separating the causes of noise and eliminating them (or at least reducing them) is very difficult. Progress has to be made on all fronts (and is being made) but it is very slow.
MEMS Investor Journal: Why are thermal limits important for MEMS sensor performance? Could you please expand on this?
Dr. David Arch: These limits simply state that if everything is perfect (electronics, packaging, etc.), thermal limits would dictate MEMS gyro performance -- you cannot get better than this. However, we are a long ways from this in actuality.
MEMS Investor Journal: What is the highest performance MEMS inertial sensor (both accelerometer and gyro) that Honeywell currently offers? For which applications are these sensors ideally suited?
Dr. David Arch: Honeywell does not offer individual MEMS sensors at present. We do offer MEMS based 2-axis and 3-axis rate sensors and MEMS based IMUs in a 17 cubic inch configuration and a 5 cubic inch configuration. We also will introduce a MEMS AHRS product in early 2011. In run stability for our best configurations are in the 1 degree per hour range with a bias repeatability of 20 degrees per hour; ARW is ~0.1 degrees per root-hour. For the accels, the bias stability is on the order of 3 mg and bias repeatability 5 mg with a scale factor repeatability of 300 ppm. We believe these numbers are state-of-the-art.
*********************************************
Dr. David Arch has worked in sensor development at Honeywell’s Advanced Technology Center from 1980 to 2006 in a variety of capacities. David has been Manager and Director of Technology from 1986 to 2006 with focus on MEMS based sensors in general and, specifically, on MEMS inertial sensors for avionics. Dr. Arch is presently Marketing and Product Manager for inertial sensors at Honeywell Aerospace. Dr. Arch holds a Ph.D. degree in Solid State Physics from Iowa State University.
Copyright 2010 MEMS Investor Journal, Inc.
Angular random walk (ARW) that Dr. David Arch is referring to will always exist in a MEMS Gyro no matter how precise the fabrication process is. Since most MEMS Gyro's work on the principle of Coriolis effect, their electromechanical behaviour at the micro-scale is not completely deterministic, i.e., the laws of structural dynamics need to be amended to include random effects, perhaps using Monte Carlo simulation methods. Further the dynamic system identification must closely take into account the characteristic length of the cavity and the gas properties (if any encapsulation is used, like Nitrogen, Helium, etc.), in particular the mean free path. The MEMS cavity sizing needs to carefully designed around these parameters.
Dr. MP Divakar
Posted by: Dr. MP Divakar | December 03, 2010 at 11:51 AM
can you please provide me Dr. David Arch's email address?
Posted by: farhat | April 18, 2012 at 04:54 AM