mixing different motors in an X8 configuration?

[Andrew Chapman]

Hi all, I'm just getting back to all this after being away on a 6 month project. Amazed to see all that has been achieved while I was gone, congratulations everyone!

I've been doing some static thrust testing of different prop and motor combinations for a new large X8 build, as well differential thrust mixing between the upper and lower motors (seems to be most efficient with about 4-8% more thrust on the lower motor in my case). I also tested using different size and pitch of props between upper and lower, and I know in theory what should work best, but no combination I had resulted in a significant win over simply a matched pair of the best performing props (T-Motor 18" CF, closely followed by Foxtech Supreme 1865 CF, btw).

One thing I found is that it seems a little more efficient to run a higher kv motor on the bottom in order for it to spin faster, rather than a matched pair with a higher throttle on the lower one.

So, I was about to order the motors to be 4x 380kv on top and 4x 400kv on the underside, but thought I'd check if anyone could foresee any strange flight characteristics in doing this? On APM:Copter each of the four upper+lower motor sets is yaw balanced (effectively it's own quad), unlike some other platforms which have all top motors spinning one way and all lower motors the other, so I was confident there wouldn't be any problem with yaw in a normal hover, but then I started thinking about combinations of yaw and pitch/roll and wondering if it could result in some unexpected skew given the upper and lower motors are running at different kv but the flight controller is expecting them to all perform the same.

And actually, now I think of it I guess the same question applies even if they're a matched pair of motors and we're doing software mixing for higher throttle on the lower motors.

I'll loop back and tackle issue #1467 soon if nobody else does, but I can at least for now confirm that it is worth doing.

AC.

[Robert Lefebvre]

Hi Andrew,

My feeling is that any time you don't have a completely "symmetric" layout (and I'm using the term broadly to mean any situation where thruster performance is not completely equal and linear), you will end up with our simple control loops having errors, which must be detected and then responded to.  So, having different motor KV is a step in that direction, however, as you point out, and X8 is already not symmetric since the lower props are operating in the downwash of the upper props.  So, will having slightly different KV make any real difference?  Hard to say.  

My gut feel is that it will be fine.  I've flown an H8 on simple round Octo code, and it was OK.  You could actually hear that things weren't quite even, particularly when yawing, as there was an initial response of the motors, and then a response to the response.  ie: From a level hover, give a yaw input, you would hear the motor sound change twice.  But at the end of the day, it was still completely stable and you sure couldn't see anything wrong with the video.

Rob

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[john...@gmail.com]

The traditional solution in R/C for coaxial propellers is to have a ~20% pitch difference on the propellers.

[Robert Lefebvre]

I'm not sure that's well supported by any physical concepts.  At best, it may be empirically true in some case.  Many people think that the air above a prop has no airspeed, and the air below a prop does.  Therefore the bottom prop would need more pitch to account for the fact that the air is "already moving".  But the fact is, the top prop, or even a prop in a single prop system,creates induced velocity above itself, and it also operating in an air column which is already moving.

If there's a benefit to props with more pitch, or higher KV motors, is probably dependent on a lot of factors, such as prop separation, etc.  It may really be not worth worrying about.

[Andrew Chapman]

I did the tests, and for me having matched 18x6 props top and bottom gave the same result as a 18x6 on top and an 18x8 on the bottom. 

However, that was using the T-Motor beechwood props, which while good, are heavier and not as efficient as the T-Motor CF prop, which only comes in a single pitch of 18x6.1

The theory makes sense and perhaps if the prop discs were closer together it would pan out as a benefit, but given the >=10cm distance between there wasn't a measurable difference in my testing. I'm not ruling out some small percentage gain, but not large enough to notice on the static test, perhaps only over the course of a 15 minute hover flight as comparison, but that involves ordering a full set of 8 props instead of just 2, so it's an expensive test.

Another thing I noticed is a big efficiency gain in having the props underneath the boom rather than above (in a non-coax setup). So any long endurance quads should consider that option (on the high end, Aeryon recently switched to underslung props in their SkyRanger).

AC.

[Andrew Chapman]

And hopefully this is a moot point in a few years when variable pitch props become standard on multi-rotors, with the flight controller dynamically adjusting the pitch on the fly.

AC.

[Robert Lefebvre]

Which theory makes sense?  There is no scientific theory I have ever seen, which explains that higher pitch props on the bottom would help. The fact is, that the closer the props are together, the less difference there is between the flow velocities.  When the disks are very close together, the induced velocity is exactly the same.  There's no reason to use more pitch or higher KV on the bottom.

[Andy Lee Robinson]

It seems intuitive to me that the lower propeller would give more thrust if the pitch angle was increased to cope with already accelerated incoming air - think how the compressor blades on a jet engine gradually increase in pitch through each stages. An awful lot of time and money has gone into squeezing every last drop of efficiency out of them!

I think I read somewhere that contra rotating blades are less efficient and noisier as the lower propeller smacks into the upper's bow waves.

I would be interested to see what the figures would be for a lower blade (or blades!) running with higher kv, and/or higher pitch, and rotating in the *same* direction.

[Leonard Hall]

Hi Andrew,

I think you are on the money here. Unless the benefit is obvious, go for simple matched upper and lower props. The maintenance benefits will be much more significant than the last fraction of performance.

The other thing to remember is the copter is constantly adjusting the upper and lower motors to balance yaw so you never sit in this perfect world where both props are wonderfully balanced and efficient. The dynamic nature of multirotor thrust makes the optimisation of coaxial props very difficult to impossible. My attitude to coaxial props on multirotors is to make sure you avoid the bad combinations (can easily loose 20% flight time) rather than try to optimise for the last 2%. I then work on making the copter easy to maintain and fly well.

From a control perspective, there is nothing you are doing here that will cause any problems for the X8. The Y6 is a much more complex problem because you also need to match the pwm outputs to each motor (yaw pid integration) .

Hope this helps,

Leonard

[Emile Castelnuovo]

My experience with X8 is that the best result I've obtained is with higher pitch on the bottom ones.

Props have a stall velocity and this is influenced by the pitch. If the air above is already moving the effort to push it is harder, thus a higher pitch will be more efficient.

In planes variable pitch propellers are used with low pitch for take off (more power) and high pitch for cruising (more efficiency at higher speeds).

I have not bench tested many propellers, but I was lucky enough to find my sweet spot with a few sets.

I could not fly, for example, with two 13" APC SF, neither with T-motor CF and APC SF. When I used graupner e-prop on the bottom 13x8 and 13x5 on top I touched the sweet spot and never went back.

To note though that I use 700Kv motors in 4S, so probably this setup is good for only this set of motors.

But then I totally agree with Leonard, when he says that with bad combinations you get really bad flight times.

Emile

[Ian]

Chiming in with my experience: I've been flying an x8 built on the 3DR quad frame with 900kv/10x4.5 props on the top, and 1200kv/8x4.5 on the bottom; this wasn't by design, those were just the motors/props I had kicking around from my two previous quad builds so I thought I'd give it a try and see what happened. Seems to work fine; I'm by no means an expert, and haven't had time to conduct in depth testing yet, but nothing sounds or behaves "off" so far.

My thinking was to put the faster motors on the bottom, as the air would already have been accelerated by the top motors. I'm going to try 10 x 4.5 props on the bottom motors over the holidays and see what difference it makes. Will probably swap to identical motors eventually, but it was still interesting to see it working!

[Robert Lefebvre]

Your question about jet engines made me question what I believed, because I didn't immediately have a good answer for it.  But after thinking about it for a while, here's what I have come up with:

1) Jet engines have only a single, or at most a few shafts.  Thus, each stage is constrained to rotate at the same RPM.  For each stage to do work, it must have a higher blade angle so that it does work on the fluid.  If a further stage had the same blade angle and RPM, it would just be "coasting" and not doing any work on the fluid.

2) While each stage does work on the fluid, the cross-sectional area at each stage is reduced (speaking of the compressor, the turbine is the opposite).  The volume of air is reduced through each stage, thus compressing it to higher density.

3) This is not true at all for a coaxial propeller system.  There is no volume reduction, and I don't believe there is a pressure increase.  There is a velocity increase.  Also, the propellers are not linked.  

Point 3 is a bit fuzzy, and maybe leaves more questions than answers, but I think it does demonstrate that the comparison to a jet engine is not accurate, since we are not dealing with compression.  Part of the problem with this discussion, is we are always using a very simple model, which actually neglects compression, and assumes this perfect stream-tube, such as here:

http://www.mh-aerotools.de/airfoils/propuls4.htm

The truth is, the stream-tube is far from clean as shown.  

[Andrew Chapman]

Hi Andy, I did that test, and it seems that having the prop pair rotate in the same direction is both quieter and a little more efficient. 

Here's the total amp consumption over three tests (logged in the Taranis), I had the pair pulling a total of 1.5kg of static thrust over a few 3 minute runs:

With the differential motor mixing I found to be most efficient (about 4% faster on the bottom) these are the results with:

 - 380kv on top, 400kv on bottom, contra-rotating: 392 mAh consumed

 - 380kv on both top and bottom, contra-rotating: 406 mAh consumed

 - 380kv on both top and bottom, rotating the same direction: 386 mAh consumed

With contra-rotating props it was making 87.8 dBA of sound, with non-contra-rotating (what's the term for that?!?) it was a much quieter 82.7 dBA.

Also, there was a slightly imperfect variable in the non-contra-rotating prop test, in that I only had a CW/CCW set of the most efficient props, so for this test I used a Foxtech Supreme CF 1865, which I found in my single prop static tests to be marginally less efficient. So there's probably another percent or two in there if I had a perfectly matched pair.

So... there looks to be some benefits -- is there any downside from a flight control perspective to having each pair of motors in an X8 configuration spin the same direction?

AC.

You can see that having the slightly higher kv motor on the bottom is slightly more efficient than a matched pair of 400kv motors, and 

[Robert Lefebvre]

Well that's an unexpected result.  I wonder how the contra-rotating would perform if there was a set of stator vanes between the two sets of props?

In regards to flight control, if each pair of motors turn in the same direction, you would basically have a quadcopter.  Actually, you would have to fly it as a quadcopter.  I think this might actually be fine.  You would still get the benefit of redundancy. 

[Andrew Chapman]

Yep, in terms of yaw contributions it would be a quad, but you'd still want an octa mix for the extra efficiency of driving the lower motors faster than the top ones.

One other consideration if this were to become a supported config, is that both my X8 frames have laterally folding arms. One has very solid locking pins so I'm not worried about that, but the other is more friction based, so it would be really cool if the yaw directions worked out such that the arms were trying to keep themselves locked in place rather than folding up in flight if the two motors were both pulling with the same torque direction ;)

AC.

[Andy Lee Robinson]

Rob, thanks for your reply - I'm always questioning my beliefs!
I wasn't drawing a direct comparison, but merely indicative in that there could be some properties that could be transferable.

For point 3, yes the propellers are an open system, but I don't believe we can say there is *no* volume reduction or *no* pressure increase - small or insignificant perhaps!

The 'walls' of the compressor are the atmosphere itself which has non-zero mass and inertia.

Looking at the stream tube diagram, it suggests to me that smaller 2nd stage props would help, resulting in a higher velocity and even smaller diameter stream tube (mv²)

Contra rotating blades correct or recover energy lost in swirl, but the 2nd stage blades rotate in a turbulent airstream and hitting the 'swirled' prop wakes at twice the frequency and seems that more energy is lost as noise than is recovered from the swirl. (I had a ride in an airboat in Florida once that used contra-rotating blades, so can attest to that!)

So, my hypothesis would be to favour co-rotation - may result in more swirl, but also produce higher velocity and therefore more thrust.

[Andy Lee Robinson]

Andrew, thanks very much for your tests - looks encouraging!

Opposite of contra-rotation would be co-rotation, and yes the noise difference is significant!

Would be nice to see the energy consumption of each motor, Perhaps use a battery each for top and bottom motors, and then measure charge state after long tests?

I'd guess an optimum configuration would be when top and bottom power consumption become equal, perhaps with bottom blades 50% extra pitch, or 50% extra kv?

Also make bottom blades smaller diameter to fit entirely in the top blades wake.

Lot of dimensions to plot here!

Cheers!

[Brian DeBusk]

Results reported by coaxial props that are matched, or vary by pitch, or vary by KV,, or vary by diameter will suggest an optimum configuration -- but that probably only applies to that specific scenario.  I personally believe the airflow is so chaotic in a coaxial setup that if you've optimized one specific scenario (i.e. and found a 20% greater pitch on the lower prop), it probably won't hold after an even slight equipment change..

That doesn't mean anyone should give-up, but I do believe everyone should be clear not to over-generalize the observed results.