hawkward517
Posts: 2
Joined: Fri Jun 03, 2016 2:45 pm

MS5611 drift a lot

Fri Jun 03, 2016 3:08 pm

Hi everyone!

I'm working on a hover function for my drone and I thought that the MS5611 pressure sensor was the best option available. So I hooked my sensor and read it every 20ms or so (using the max OSR). I precise that the sensor is in the drone on my desk and the motors are turned off (so no air flow to disturb the sensor)
As you can see on the graph (~10 min readings), the temperature increase a lot and the pressure decreases.
raw_temppress_alt2.PNG
raw_temppress_alt2.PNG (59.3 KiB) Viewed 919 times
I thought that it would be compensated with the hypsometric formula but no (I use my first readings as a reference pressure to have the altitude compared to the ground). As you can see, I filtered my raw pressure readings and then plotted my altitude (in cm) and it gives a 5 meter drift with bad random spikes. I tried other MS5611 and I have to admit that I think this one gives me the best results...
Since the CRC is valid at every frame I guess that it really comes from the sensor.
Does anyone have an idea?

hawkward517
Posts: 2
Joined: Fri Jun 03, 2016 2:45 pm

Re: MS5611 drift a lot

Tue Jun 07, 2016 9:37 am

Me again, just to add the fact that I get plausible values if I have quick changes.
30sec3steps.PNG
30sec3steps.PNG (26.63 KiB) Viewed 898 times
On the graph you see (after filtering) altitude changes 30 seconds each position (the lines are the real height values)
Does anyone have an idea why so much drift and if the quite big noise I get is in "normal"? :?
Thank you!

MadAl
Posts: 1
Joined: Fri Nov 03, 2017 8:47 pm

Re: MS5611 drift a lot

Fri Nov 03, 2017 8:53 pm

Hi,
I get similar drift results with temperature. This is indeed weird, as the pressure compensations (first and second order) using the stored calibration and correction values should take care of this. It apparently doesn't. As to the noise, this is pretty normal. Also the measurement frequency plays a big role in pressure drift. For instance, when I measure at 5 Hz or 1 Hz creates huge offsets in pressure readings. I contacted the manufacturer but did not get a satisfactory answer to any of these issues. If you have found a cure I would be very much interested.

a.texer
Posts: 8
Joined: Thu Dec 03, 2015 6:18 pm

Re: MS5611 drift a lot

Wed Apr 18, 2018 2:09 pm

What about natural pressure variations?
See below, all 4 sensors were sitting at one spot collecting data at 1Hz, so ~22 hours of data total.
MKS e-Baratron is my reference sensor.
BMP280 is from Adafruit.
MS5611 is GY-63 one.
Zubax GNSS 2 module also employs MS5611 sensor.
Attachments
Graph10.png
Graph10.png (69.55 KiB) Viewed 350 times
Graph12.png
Graph12.png (82.14 KiB) Viewed 350 times

Return to “Automation, sensing and robotics”