Raising the temperature

I flew Phoebe a few times yesterday (with blu-tack), and she consistently drifted several meters during a 10s flight – worse than ever.

This morning I brought her indoors just to confirm this massive deterioration in flight stability was temperature related.  Sure enough, indoors she just hovered perfectly:

Checking the temperature from Andy Baker on Vimeo.

Another temperature test

I ran a quick passive exploratory test tonight, again temperature related with the blu-tack installed.  The aim was to understand how much the IMU temperature drift is related to being worked hard, and how much is related to the wind from the props.

I ran a test indoors with no prop power – here’s what I got.  Look at the resolution of the temperature: over an 11 second flight, the temperature range spans 35.79 – 35.93°C, or 0.14°C peak-to-peak which is tiny.  I think I can safely say it’s solely environmental changes (i.e. the breeze front the props) which cause IMU temperature- and therefore sensor drift.

Passive flight temperature change

Passive flight temperature change

I think I need to add some more blu-tack to the underside of the IMU PCB to give it the same breeze resistance the top now has.

A bigger blob of blu-tack

A bigger blob of blu-tack

A bigger blob of blu-tack

Another test this morning with a bigger blob of blu-tack providing thermal insulation and protection from the breeze.

The temperature profile is much better: it dropped only 0.25°C in 11s seconds, compared to yesterday’s 1.2°C with a smaller, thinner piece of blue tack, and 1.4°C when completely exposed to the ambient temperature and breeze from the props.

Given the sensors drift about 0.1% / °C according to the specs, and a 0.1% error in gravity is 0.01ms-2 acceleration error or a 1m vertical drift during a 10s flight, it’s got to be good thing the blu-tack has made the thermal drift linear, and so the butterworth can deal with it well.


And yet more blu-tack

While testing stuff from the previous post, P’s battery broke loose again – even with her yellow sponge landing dome, the shock of hitting the ground due to an aborted flight is enough to rip the duct-tape strapping the battery under the top plate.

So the battery has now moved to the bottom plate held in place by the same super-sticky black blu-tack as holds the IMU on the breadboard.  Seems to be a lot less likely for the battery to separate completely, and easier to fix should the battery slip.

The reason the battery was dangling from the top plate in the first place is because the space on the bottom plate it’s large enough to strap it down tight, but by using a the blu-tack, that’s no longer a problem.

A £1 coin and some blu-tack

For the last while, Phoebe’s drift in non-windy conditions has always been backwards.  Until yesterday, I’d written this off to poorly calibrated X-axis accelerometer leading to slight horizontal acceleration while Phoebe thought she was stationary.  This horizontal acceleration is only limited by friction through the air so can soon get out for control.

But yesterday, a conversation outside the blog with Phil, combined with a dislodged battery got me thinking about another possible cause..

Now the position of the battery in her frame shouldn’t cause a problem; the PIDs should handle any physical imbalance and resultant drift.  But what if the props produce a small lateral force in certain situations in addition to the significant vertical one, or perhaps the vertical force is not equal in some way…

Bear with me now while I start some unbounded speculation based upon the theory this lateral force does exist…

Individually, when spinning in free space, this lateral force is balanced across a propeller blade pair having zero net effect.

But put an object near the single spinning propeller, and there is a reduced amount of lift as the blade passes the object.  This is why helicopters always crash into buildings in the movies. However, across four props spinning at the same speed, at the same distance from the object, the net force is still zero.  In this example, that object is Phoebe’s perpex dome protecting her and her circuitry.

There may also be a similar effect due to one pair of prop neighbours spinning in opposite directions, producing an imbalanced lateral force in the zone at the closest point the props pass each other.

Now imagine the quad is slightly unbalanced, perhaps due to a battery being set back slightly from centre making it arse-heavy.  To acquire and maintain horizontal hover, the rear blade pair spins faster compared to the front pair to lift the extra weight of the battery.

But in doing so, they increase the lateral force of the back pair of blades.  Depending on the blade rotation direction this lateral force could either over correct the unbalance, or counteract the balance.  Either way, the quad accelerates laterally, and only friction limits the resultant drift speed.

In a normal quad with a human feedback loop, this can be corrected by setting a fixed horizontal velocity compensation in the transmitter, but autonomously it cannot meaning the only solution is to make sure the quad is physically well balanced.

So on reseating the battery yesterday, I took a little extra care to ensure the battery was seated as central as possible.  And as a result, the drift ceased to be as significant, and now was more in a left-right direction (the reseated battery was still skewed that way).

And here’s where the blu-tack and pound coin came in.  There’s enough space on the flange of Phoebe’s dome to add small weights – in this case a £1 coin.  And things improved further.  Not perfect but for me to tune the weight balance more, I need to sort out the scaling problem between the accelerometer axes first, as Phoebe is climbing to 2m+.  Thank heavens for her yellow sponge ball!


Duct tape, velcro and now blu-tack

I’ve just added blu-tack to my range of adhesive components sticking Phoebe together.

I spotted yesterday Phoebe was showing a 3° nose down tilt when on her horizontal test platform.  That’s a big problem (3°tilt = 0.05g or 0.5ms-2 horizontal acceleration) and it took me a while to find the obvious source – the MPU6050 breakout had shifted in its attachment to the breakboard – probably loosened finally by too many crashes; it has been surprisingly rugged for a very long time.

Hence the blue-tack to keep the MPU6050 breakout firmly attached to the breadboard – although I’m actually using a different black variety which is softer and stickier when massaged.  This means it works itself into nooks and crannies, and then is strong enough to take paint off a wall (it was for sticking my DIY hi-fi speakers to custom stands; it squidged out very thin when putting the two together, and the two were virtually inseparable thereafter).  It also matches Phoebe’s colo(u)r scheme better 🙂

I wonder how much this has been affecting my drift control testing recently?  Weather is good this weekend, so we shall see.

P.S. That’s not my complete set of sticky stuff I use; the last is double-sided padded sticky tape which sticks the Raspberry Pi and breadboard to the frame’s top plate which provided a very strong bond while the padding does provide some noise suppression.  I left it out from the blog’s title as it was too long!