Talk about sensors, OSD systems, radios and anything else related to electronic hardware.
User avatar
User

Korken

Posts

83

Joined

Tue Mar 18, 2014 8:11 pm

Invensense's MPU-9250 vs ST's LSM9DS0

by Korken » Thu Jul 10, 2014 8:56 pm

Hi all!

I'm always having an open mind about MEMS sensors and after being a true Invensense follower, I have been looking at the LSM9DS0 sensor for a while since it was released with great interest. So I thought I would do a smack down of datasheet "performance" and see what the result is (and some mailing with ST for clarification), and this is the result of some of the common "standard" performance parameters. Rate is assumed to be 2000 deg/s.
comparison.png
comparison.png (18.54 KiB) Viewed 7513 times

The LSM9DS0 is better/equal than the MPU-9250 in all aspects except initial zero rate.
Does this mean it is better? Maybe, maybe not. A real life test is needed to determine this.

But I will certainly be looking more at the LSM9DS0 sensor! Most likely make a test board based on this sensor instead of the MPU-9250.

Does anyone here have experience with this sensor and can give a professional opinion?
Or any other comments? Would there be interest to characterize these sensors?

Additional information, the LSM9DS0 is approximately 3 USD cheaper than the MPU-9250 as well.

Best Regards
Emil
Last edited by Korken on Fri Jul 11, 2014 7:58 am, edited 1 time in total.
no avatar
User

jcdoll

Posts

21

Joined

Sat Apr 26, 2014 9:32 pm

Re: Invensense's MPU-9250 vs ST's LSM9D0

by jcdoll » Fri Jul 11, 2014 6:00 am

If you are looking at minimizing integration errors over time then bias instabilities for the gyroscope and accelerometer will be important. Bias instability of gyroscopes and angular random walk are the major parameters that have improved across generations and differentiate navigation grade macro scale gyros from their MEMS equivalents. It isn't consistently reported, but you can see the Allan deviation plots in the MPU-3300 datasheet for example (15 degree/hour). This is coming from a non-gyro MEMS guy with limited background in inertial navigation, so not sure where the main shortcomings are for flight and no professional opinion. Also not sure how easy it is to procure Bosch gyros, but those are the big three companies.

- joey
User avatar
User

Korken

Posts

83

Joined

Tue Mar 18, 2014 8:11 pm

Re: Invensense's MPU-9250 vs ST's LSM9DS0

by Korken » Fri Jul 11, 2014 8:00 am

Indeed the bias stability is very important!
I have contacted ST about this on their LSM9DS0 and LSM9DS1 and I will see what they respond with.
Also, I have purchased this sensor to be able to do characterization of it, and to get a good feel for it. :)
User avatar
User

Korken

Posts

83

Joined

Tue Mar 18, 2014 8:11 pm

Re: Invensense's MPU-9250 vs ST's LSM9DS0

by Korken » Mon Jul 14, 2014 3:21 pm

Now I have gotten really interested in knowing how sensor perform vs each other, and I have the equipment at the university to test almost everything.
So I designed a "SensorStick of Destiny" to be able to compare sensors.

The sensor stick includes:
    - ST's LPS25H Barometer
    - Meas Inc's MS5611 Barometer
    - Honeywell's HMC5983 Magnetometer
    - Invensense's MPU-9250 9DOF IMU
    - ST's LSM9DS0 9DOF IMU
    - ST's LSM9DS1 9DOF IMU
    - LP2985A 3.3V low noise PMIC

This will be an interesting summer project. :D

Render of the board:
sen_stick.png
sen_stick.png (281.44 KiB) Viewed 7453 times


Schematic (C16 should be 10nF):
stick_schematic.png
stick_schematic.png (66.68 KiB) Viewed 7453 times
User avatar
User

VArcht

Posts

287

Joined

Thu Mar 06, 2014 5:24 am

Re: Invensense's MPU-9250 vs ST's LSM9DS0

by VArcht » Mon Jul 14, 2014 6:13 pm

Cool, looking forward to the results. One thought though, would it be better to array them around center so that their translation would be relatively the same?
User avatar
User

Korken

Posts

83

Joined

Tue Mar 18, 2014 8:11 pm

Re: Invensense's MPU-9250 vs ST's LSM9DS0

by Korken » Mon Jul 14, 2014 7:31 pm

I'm going to be looking mostly at temperature drifts, bias drifts and gain errors.
For this the position of the sensor is arbitrary. :)

Edit: Plus the distance between the inertial sensors is approximately 5 cm. Which is quite small.
User avatar
User

peabody124

Posts

1339

Joined

Tue Feb 18, 2014 12:33 pm

Re: Invensense's MPU-9250 vs ST's LSM9DS0

by peabody124 » Tue Jul 15, 2014 4:45 pm

Yeah we did Allan variance plots between mpu6000 and l3gd20 years back. It takes a long time to run though and requires making sure you don't gave any AC cycling or anything like that.

Also keep in mind you can get things like mpu3000 (??) which have ultra low drift but you spend 50$ just for the gyro... However I think those benefits get buried by the noise and better algorithms are probably more critical.
User avatar
User

Korken

Posts

83

Joined

Tue Mar 18, 2014 8:11 pm

Re: Invensense's MPU-9250 vs ST's LSM9DS0

by Korken » Tue Jul 15, 2014 7:19 pm

Yeah, low drift is only important if you loose all other forms of references - äny good algorithm would compensate if it has referenses.
I am more looking getting the best "bang for the buck" (as Americans might say?).
I just got answer for ST as well, the LSM9DS0/1 have a bias stability of 20 deg/hour. Not far from Invensense's MPU-3300.
On top of this ST will send me samples of the LSM9DS1 which will be nice to try as well. :)

With this I can compare gyros, accs, mags and barometers.
Everything I've ever wanted to know. ;)
no avatar
User

ernieift

Posts

193

Joined

Sun Mar 02, 2014 11:21 am

Re: Invensense's MPU-9250 vs ST's LSM9DS0

by ernieift » Tue Jul 15, 2014 10:29 pm

This is a labor pcb, right?
For a better noise distance, i'd separate the Vcc of each sensor (with ferrites or more 3V3 regulators) and put a little resistor in the signal lines.
For mixed signal boards, keep digital wires away from analog by putting a null current ground track between them.
Otherwise, it looks good. I'm keen on your results...
User avatar
User

Korken

Posts

83

Joined

Tue Mar 18, 2014 8:11 pm

Re: Invensense's MPU-9250 vs ST's LSM9DS0

by Korken » Wed Jul 16, 2014 11:15 am

Good of the ferrites! I added them.

On the digital part, I have put digital lines on one half of the board and analog on the other.
This should be enough I believe.
Next

Who is online

Users browsing this forum: No registered users and 1 guest

Powered by phpBB ® | phpBB3 Style by KomiDesign
cron