With the latest release of the OpenPilot software version 14.01 come two new features (among others) that when combined, make flipping and rolling a multirotor (MR) even easier. A new flight mode called "Rattitude" is a hybrid of the self levelling "Attitude" and the more aerobatic 'Rate' mode. In "Rattitude" mode as the control stick on the transmitter moves out from the centre, the flight mode of the craft changes from 100% Attitude through to 100% Rate with varying degree of mixture in between. The effect means that letting go of the sticks means the MR will return to self levelling (ie Attitude mode) but you can still throw it around as the control sticks move out from centre, you get the aerobatic response of Rate mode.
Normally you would have to flick a switch to go from Attitude to Rate. In Attitude mode, because it's always trying to self level, to maintain forward flight you have to constantly be holding the stick forward. It feels like you're fighting it to a degree for the whole flight. Also Attitude mode limits the maximum angle of pitch and roll so if you try to roll in Attitude mode, you'll roll to the angle defined as the maximum and not go any further around.
In Rate mode, it goes where you point it. If you roll to 45 degrees, the MR will attempt to hold that angle until you tell it otherwise. Rate mode can be intimidating for the novice pilot as the safety net of self levelling is removed and depending on the settings, Rate mode can fill quite 'slippery' if you're not used to it.
The other new feature in 14.01 is Cruise Control which from my limited reading of the topic attempts to maintain a steady altitude when the MR is turning or banking by slightly increasing the rpm to counter the slight decrease in lift). With Cruise Control comes a 'Max Angle' parameter that when set to 90 degrees, combined with Rattitude mode, means that as the MR rolls past 90 degrees, it automatically spins down the motors. That means the part of the roll when the MR is passing through the inverted (ie the props would normally be pulling it to the ground), the motors are automatically shut down which seems to make for a much tighter roll.
From personal experience I went from rolling and flipping with at least 10m height to doing the same at maybe head height. The first time I tried this combination it was quite weird to hear the motors shut down as it passed through the 90 degree vertical point but they came back up a few milliseconds later. Posts in the OpenPilot forum suggest that setting the Cruise Control Max Angle (CCMA) to 80 degrees and the degrees / second to 360 make for some nice smooth rolls.
The screenshots below show captures from the OPenPilot GCS software version 14.01 and the settings required to change to get this to work. I first tried it on my "beater" quad with a CC3D flight controller and liked it so much I put it on my "nice" quad with a Revo FC and Ov3rquad frame. The video below shows the results from these settings on both my CC3D and Revo controlled quads with CCMA set to 90 degrees and the Rate mode "Degrees / second" setting up at around 450-500.
The next thing I want to try is lowering the CCMA to 80 and deg/sec to 360 for the nice, "natural" roll and then go the other way to CCMA = 100 and deg/sec =720. Maybe with these settings I might get in a few double or triple rolls before hitting the ground.
Today I flew the PVC quad on another shakedown flight (after fiddling with the configuration via the GCS). I was flying over a fairly large oval, not very high but on the far side from where I was sitting when all motors lost power. I believe failsafe on my Futaba kicked in. You can hear in the video that the motors stop and then just before hitting the ground, they briefly start again. I walk over to the quad from the other side of the oval, perform a quick inspection to make sure nothings broken or fallen off and without any other changes or resets, power up the OP CC3D by holding yaw right and away I go.
I've used this Futaba Tx and Rx pair in nitro fixed wing models at much further range and never had an issue. The main difference is now I have a 150mW 1.3GHz video transmitter (http://www.readymaderc.com/store/index.php?main_page=product_info&cPath=11_30_38&products_id=1018) about 20cm away from my 7 channel 2.4GHz receiver. I've read about guys using low pass filters on the vTx so I thought it might be worth checking the interaction between the 1.3Ghz vTx and the 2.4GHz Rx. Fortunately a guy from work that came out to watch carries a spectrum analyser around in his car so we set about looking at how "dirty" is the 1.3GHz vTx.
The screenshots of the specan show a 2GHz wide band centred at 2GHz (so it's showing from 1 - 3 GHz). The first screenshot is using the "Max Hold" function on the specan which is effectively a cumulative trace that shows the max detected levels across the band and holds it. This screenshot shows my 2.4GHz Tx as well as the in building wifi and a few other small peaks. At this point, my video transmitter is powered down.
The second screenshot on the same scale (1 to 3GHz) after the vTx has been powered up for a few seconds. You can see a peak at 1280MHz (vTx frequency) but disturbingly, across the whole band are is a whole lot of noise coming from the vTx. The second harmonic at 2560MHz appears to be right where we see the Futaba Tx.
The effect of this noisy little transmitter means it's desensitising my 2.4GHz receiver. It's like someone talking really loud (1.3GHzvTx) standing right next to you (2.4GHz Rx) when you're trying to listen to instructions from someone talking in a normal voice on the other side of the room (the 2.4GHz Tx).
To address the problem a 1.3GHz low pass filter can be fitted in-line on the vTx between the output and the antenna. The low pass filter allows the video signal to be transmitted down in the 1.3GHz band but anything above roughly that frequency is heavily attenuated. The noise coming from the vTx is no longer shouting in the ear of the 2.4GHz receiver so I can hear the instructions from the transmitter much easier and at a further distance. I've ordered this low pass filter (http://www.foxtechfpv.com/1400mhz-low-pass-filter-p-439.html) from foxtechfpv to hopefully solve the problem.
What I found most useful today was being able to see the problem right there on the spectrum analyser. Being able to see the video transmitter peak at 1280MHz and then all the smaller peaks across the rest of the band including the 2.4GHz area right where my Futaba Tx is.
What will be interesting to look at next is after I fit the low pass filter, how does it look on the spectrum analyser. I expect it should look much cleaner. Something else I also want to look at on the specan is the 1 watt 1.3GHz vTx fitted to my DJI F450. I've flown that at much further range without issue. Is the vTx a much better one where it transmits only on the frequency its supposed to? I'll hook it up and find out.
I haven't had the chance to actually go fly the YAPVCBQ since the somewhat unsuccessful maiden so the next best thing is to tune it indoors and I'm learning heaps. For example, the first time I fired it up on the line, only motors 1 & 3 were turning consistently. First thought was I had dodgy connections to 2 & 4. What I figured out (I think) was that I had to disable the yaw and pitch stabilisation because strung up on the wire like that it was trying to compensate in an axis it had no control over.
Well I think that logic stands up because after disabling yaw and pitch, all four motors worked fine.
I set out to tune the inner loop in rate mode for a start with the default setting of Kp 20 and Ki 0. Rotating the quad until it was level was easy enough but it would pretty much start drifting off level straight away. I upped Ki to 15 and it held it longer before starting to drift.
That's all I've had time for so far but I've found tuning it strung up like this much better than going out to fly. First it lets me concentrate on one axis at a time and second it's much easier to plug the laptop in straight away to adjust the values.