Just got an email from Martin Harris, he have made himself the RC-Thoughts Jeti Altmeter. Not only have he tested it in true flight-conditions but he also had MVario2 sensor and the older Vario-sensor installed in the same plane. What a perfect opportunity for a true life comparison!
Whole flight in one picture
MVario 2 is on red and RC-Thoughts Jeti Altmeter on blue graph-line. And they really can be compared, very minor differences and time is also really good following each other. Certainly excellent result for a 5€ sensor!
Details and what can we read?
If we zoom in to a 15 second slot we are seeing some patterns on the very minor differences. Comparing the time vs value axis and more precisely the pattern of it we can read that Jeti Altmeter has some fluctuation on values, meaning the linearity is good but could be better. The accuracy on altitude-measuring with these devices always leave something to desire on but the behavior of both sensor really are extremely similar.
What can we do to make Jeti Altmeter better? Is it necessary?
Well, necessary, honestly no, Martin proved that. But if we can make something better, why not do it. We can see about the linearity that normalization in Arduino of the pressure-sensors value could use some work. This is after all version 1.0 with very minor testing. I’ll look into normalization and put a new version up, simple as that. The value Arduino gets from pressure sensors fluctuates a little and therefore needs to be stabilized in code.
One interesting thing can be seen on the graphs. I use somewhat slow way to normalize the value from pressure sensor in Arduino and I was a bit surprised that it’s quite much on-par with MVario2. Appears that Jetimodel might be using somewhat similar approach in getting the pressure-reading stabilized.
Thank you Martin, really great info!
One thought on “Performance of RC-Thoughts Jeti Altmeter sensor”
Great work again! Maybe filtering with sliding window, kind of "low pass filter"? I'done such to filter acceleration measurements, n[samples] depends on sampling/reporting frequency but could be eg. 5 measurements?