3D Robotics


I noticed that Digikey is now selling Honeywell's newest aerospace-grade IMUs, which cost $1,328 each (note that's just for the IMU; it's not a full autopilot). How do the specs of these aerospace IMUs compare to those we use here? Are they worth the extra money? 

In terms of overall approach, the Honeywell IMU seem very similar to modern autopilots such as Pixhawk 2.x and 3.3: they both have MEMS sensors with internal environmental isolation and temperature compensation.

As for the sensors themselves, I'm no expert on specs, so I'll just post the basics here, comparing the Honeywell sensor to the Pixhawk 3


On the face of it, the Invensense and ST sensors in the Pixhawk 3 appear at least as good, if not better. But I imagine that there are some other factors that may be more important, such as gyro drift and vibration filtering. The Honeywell specs in drift are shown here: 


Meanwhile the Invensense ICM-20602 sensor in the Pixhawk 3 gives its drift in different units: ±4mdps/√Hz. I really don't know how to compare those.

Finally, I'm sure that a lot of the performance depends on the software running on the Pixhawk boards, be it PX4 or APM, both of which use GPS to augment the raw IMU data to compensate for drift, along with a lot of other smart filtering. 

So for those IMU experts out there: how do you think these two approaches compare? Are aerospace-grade IMUs worth the extra money?

E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • Hi

    I want to add this point that rich community of diy autopilots and their active contribution is a great advantage that i think really challenge the superiority of closed source products even by big manufacturers.

    For instance, i have a MTI3 AHRS from Xsense company. It's really a very good ahrs. it works very accurate even in very hard and dynamic conditions. But in support side they are not as fast and responsive as every one expect from a well known inertial navigation company. for example four months ago i realized that there is a walking drift in angles estimation. after many try and error i found that it's because of improper gyro bias calibration. So i re-calibrated the gyros with their gui. After that everything seemed good. but after a power off and power on, the problem came back. After contact with their staff they said that you must calibrate your gyro drift ( keep the device still for about 6 second) every time you power the module on. Such a simple and easy to solve problem!! It was very annoying condition for me and it's hard to re-calibrate the device every time i turn it on. They stated that this problem added to firmware requests.  from that time there is no evidence about the time of solving this problem.

    I think such a simple problem will be discovered and solved very faster if it exist in an open source community like ardupilot or px4. So it's a very big advantage, big user community and big developer community. In many situations, solving possible problems is very easier in open source community than closed source ones.

    of course, reliability of the product is the another side of the issue. in this side, closed source solutions of big companies have very better condition. 

  • John, good point.  I know we've seen datalogs where the sensors are not performing within spec.

  • Well we all know where Honeywell went with their ubiquitous compass chip! HMC5883, et al!

    This IMU is nothing new. There are literally dozens of tactical grade and other grade "IMU"s on the market.

    This is like comparing apples to oranges! The pix is NOT an IMU, but uses an amalgamation of sensors which aggregate into an IMU.  The Honeywell device does the same. The real question is how do the sensor's specs compare. The Honeywell device is NOT an AHRS system either.  It's a block with the sensors; has an SPI interface.  It is quite possible the Honeywell "IMU" has an Invensense sensor inside it (ICM-20602 or whatever in it).  I challenge someone to spend $1,200 and crack one open! What makes it worth $1,200.

    In addition, THE major aspect of how well the IMU will work is dependent upon how the device is driven and how the data from the device is processed.  One could have an extremely accurate gyro and software that doesn't process it adequately.  It's mostly a matter of post processing the data from the sensor. Saying the Pix does "x" because of this or that sensor and will work just as good because the sensor specs are "better" is just mush... One does not need to compare anything other than the standard deviation of the processed sensor data.

     A Pix could just as easily be connected to this Honeywell IMU instead of an Invensense sensor; has an SPI interface.  THAT would be the REAL comparison.  In other words, use the same post-processing on the data from the different sensors and compare the results.  This can be easily done by computing the standard deviation of a sample set. The real test is what is the standard deviation of the different sensors running the same post processing of sensor data.

    How well the Pix performs is wholly dependent on the software it is running and what sensor it is using, but garbage in equals garbage out!

    Note that the ICM-20602 is used in the iPhone 5!!

  • John makes an important point: Honeywell endeavors to publish honest real world specs on their products (I've spoken with Honeywell engineers who work at the site where their IMU's are developed and manufactured and they made that point specifically ... even if it leads to simplistic conclusions by potential customers, they do their best to publish honest specs.)  These aren't marketing or best-case-in-the lab specs and they aren't trying to out-spec their competition on paper.  To really say anything for sure, someone should run one of these new honeywell sensors side by side with an MPU9250 and look at relative sensor noise, relative bias drift, startup bias, how do they hold up under temperature changes? how do they hold up in a dynamic and high g-load situations?

    I also do agree with the other sentiments that the DIY community does really well on the back of inexpensive sensors like the mpu6000/9250.  But consider how quickly the EKF solution blows up (especially position) when we lose GPS, the IMU's we have now in the DIY community are still pretty noisy and have lots of small errors that are difficult to identify and calibrate out ... we need constant gps correction to maintain an accurate attitude solution.

    Anyone want to buy or lend me one of these?  I could run some side-by-side tests and report back. :-)

  • Developer

    The difference is that the Gyro bias repeatability, in-run stability etc. numbers from the Honeywell are actual confirmed numbers, that they guarantee as long a the IMU is used withing the operational parameters.

    Our numbers are more a kind of 'when I did some tests on my system I saw about x degrees of.." type of deal that may or may not change depending on a lot of factors.

  • Great topic! For hobby project I wouldn't think twice about using Pixhawk 3. However many people in this topic made good point about certification and reliability. I would never consider using pixhawk and hobby grade hardware for safety critical projects.

    I think accountability is a big thing here. I some chunk of the Honeywell price goes towards insurance.

  • 100KM

    APM 2.8 with the MPU6000 (0.005 deg/sec/root(Hz), 0.3 deg/root(hr)) is at least 2 times better than any of the HG1120s

    Pixhawk 1 with its MPU6000 (0.005 deg/sec/root(Hz), 0.3 deg/root(hr)) is at least 2 times better than any of the HG1120s

    Pixhawk 3 with its ICM-20602 (0.004 deg/sec/root(Hz), 0.25 deg/root(hr)) is at least 2.5 times better than any of the HG1120s

    The Pixhawk 3 also has a MPU9250 as well(0.010 deg/sec/root(Hz), 0.6 deg/root(hr)), which as good as the best HG1120.

    The best HG1120 (0.010 deg/sec/root(Hz), 0.6 deg/root(hr))

    I think one of the reasons the APM autopilots have been so successful is the low gyro noise of the MPU6000.

  • Aerospace grade, implying AS9100 certification or equivalent, doesn't relate to maximum performance.  It relates to verified reliability.  

    There's nothing stopping Drotek doing the work necessary for Pixhawk 3 to receive aerospace grade certification, but it would significantly increase development costs.

    The HG1120 can be used in manned aircraft applications with reasonable confidence that it will perform as expected.  The Pixhawk 3 will likely be tested via trial and error, with early adopters expecting bugs and deviations in performance.

  • 100KM

    Comparing them in terms of angle random walk, the Invensense ICM-20602  0.004 deg/sec/root(Hz) is equivalent to 0.24 deg/root(hr).   This is 2.5 times less rate noise than the best HG1120.  The ICM-20602 is an excellent value.

    The double sided rate noise power spectral intensity(in deg/sec/root(Hz)) can be converted to angle random walk(in deg/root(hr)) by just multiplying by 60.

    Two or more ICM-20602s are going to be more reliable too.

  • A quick search about bias in high-end gyros points toward Allan variance curves, from which come in-run bias stability and Angle Random Walk.

    I have quite no idea what I am doing. So read further at your own risk.

    Angle random walk at Honeywell and drift at Invensense looks really similar concepts.

    Converting ARW °/sqrt(hr) for the medium device above of 0.7 in milli degrees per second gives 0.7*1000/sqrt(3600) =12 mdps

    Then, ICM drift measurement of 4 mdps/sqrt(Hz), at, say 8kHz as is recommended by Invensense, gives 4/sqrt(8000) = 0.044 mdps

    According to this paper http://ieeecss.org/CSM/library/2010/feb10/06-AsktheExperts.pdf, table 2 line bias stability, it would put Invensense in the intermediate range and Honeywell in the moderate one, backing the thought Invensense has 10^3 less drift.

    I support the argument above about Honeywell price difference coming from build quality, superior MTBF, probably higher-grade material. And there are lots of other factors for gyro performance such as bias repeatability, temperature range, other constraints supported...


    That is a try until non-fake knowledge of high-end gyros and math is displayed here!

This reply was deleted.