Y6 Propeller testing

[Leonard Hall]

Hi all,

I thought I should post the results of the propeller testing I did for the 3dr Y6 copter.

The test:

1. I am using the default 3s 4200 mAh batteries that come with the RTF kit. I figured that as a starting point 3dr would want to see it fly out of the bag as well as possible.

2. My Y6 is currently completely empty and weighs 1.5kg

3. Currently I am using only slow fly props. I wanted to start with establishing what was needed for good control and efficiency. I assumed that 3dr would want to stick with APC Slowfly, the APC multirotor props don't come in a range of pitches.

4. I am measuring total current and power draw at 50 Hz, using the 3dr current sensor, and individual motor pwm outputs. I changed the code to increase the logging of the Current and Motors to 50Hz. From a control perspective miss-matched pwm output at hover results in less range on the throttle and potential loss of yaw control when you apply full throttle. In general it reduces the output range the controller has access two and introduces a discontinuity into the system quicker. So based on my current understanding of the problem I don't mind if the upper and lower motors are drawing slightly different power provided they are doing it at the same pwm. However, I suspect that matched pwm will also be generally good for efficiency as well.

5. Each test consisted in a take off and quick switch to AltHold at approximately 5m. I then hovered as stationary as possible for 10 minutes and landed. I have had trouble with weather while doing these tests, so I wasn't able to get 10 minutes of easy flying in. To try to counter this very significant variation in efficiency I measured the minimum power consumption during 10 seconds or 500 consecutive samples.

The Measurements:

1. Average power consumption is the average power used over the central 8 minutes of the 10 minute flight. This is not really a good indication of efficiency since the weather wasn't cooperating with me.

2. Minimum power consumption was the lowest power consumption over a 10 second period. This is a better indication of energy usage.

3. pwm miss-match is the difference between the average pwm commanded to the cw and ccw propellers. This is a good indication of how much yaw/lift coupling will be present. This is a result of throttle vs toque non-linearity and the discontinuity at maximum throttle.

4. Increased payload is an approximation of the additional payload that may be added before the throttle increases above 80%. This assumes that the approximate linearisation of the motor output in Arducopter is ideal.

5. I found that the bigger props were resulting in concerning temperatures on the motors and ESC's. So the later tests included these measurements. I took them using a Hobby King temperature sensor.

							Degrees C
Battery	Upper	Lower	Ave Power (W)	Min 10s Power (W)	Mismatch (pwm)	Payload (kg)	Lower Temp	Upper Temp	ESC Temp
3s	10x3.8	10x4.7	233	221	170	0.82
3s	10x4.7	10x4.7	230	218	187	0.87
3s	10x4.7	10x7	243	231	66	0.77
3s	10x4.7	11x4.7	220	207	116	1.00
3s	11x4.7	10x7	241	235	138	1.00
3s	11x4.7	11x4.7	222	210	173	1.18
3s	11x4.7	11x7	232	218	73	1.10
4s	10x3.8	10x4.7	243	228	154	1.63	50	35	60
4s	10x4.7	10x4.7	246	235	158	1.69	43	31	46
4s	10x3.8	11x4.7	238	222	163	1.58	43	32	49
4s	10x4.7	11x4.7	240	220	108	1.73	50	33	56

This is what I wrote after testing the 10/10 and 11/10 combinations.

I got to test fly both configurations tonight using two 4s 2200 mAh batteries in parallel. I am guessing this will be close to what you have there. I have got to say, I enjoyed flying it :)

So I did 10 minutes with each. The first 5 was spent doing an autotune followed by some yaw tests. The last 5 minutes I thrashed the crap out of it.

Autotune results in lower PID values suggesting that the 10/10 is sharper. This makes sense because of the higher rpm of the lower motor and lower inertia. So we generally expect it to be a little more agile.

So there is a general yaw/thrust coupling on both but I couldn't say which is worse without flying both back to back. I think the 10/11 was a little better, maybe. I don’t think this is too significant because you really need to be aggressive with the yaw stick to see it. Even then I didn't find it too annoying.

Both configurations performed very well in my thrash test. I was able to completely stall a pair of props once during each test. This may be both props stalling or I may have stalled an esc. In any case the copter got to around 60 degrees and recovered fine.

I found the 10/10 resulted in lower temperatures on the lower motors as was reflected in the results below.

Both configurations are very close but I will be putting 10/10 on the Y6 I am building (my own design). I think it is a safer combination that runs cooler and is a little more agile. However, I wouldn't fault 3dr for choosing either combination!

All in all I am very happy with the old Y6 frame and am looking forward to seeing the new version!

Leonard

[Leonard Hall]

Hi all,

Sorry, this requires some clarification.

10/10 refers to 10x4.7 on top and bottom and 10/11 refers to 10x4.7 on the top and 11x4.7 on the bottom.

10x4.7/11x4.7 had the best efficiency, closely followed by 11x4.7/11x4.7 and 10x4.7/10x4.7.

11x4.6/11/7 had the best pwm match, closely followed by 10x4.7/10x7 but both took hits in efficiency and lift capacity.

Anything with 11 inch props on the bottom suffered from significant heating of the lower motor and esc.

On top of this I suspect all these test would vary considerably as the copter starts moving forward through the air. I suspect that this would start to favor identical upper and lower props. The reason I say this is at some point the prop wash from the upper prop will flow back behind the lower prop and the lower prop will be in clean air. At this point the Y6 will effectively have the same efficiency as a hex. This depends on prop spacing and I don't for a second think we will get this complete situation in a normal Y6.

In the end it came down to a close call between 10/10 and 10/11 (all 4.7). I was burning my finger on the lower motors on the 10/11 after both a hover and thrash. On the other hand the 10/10 was pretty warm after a 10 minute hover but cool after a hard thrash.

Basicly the interactions between the upper and lower motors are so complex the best I think we can do is avoid the bad combinations and provide a simple, cost effective and, most of all, reliable prop suggestion.

Leonard

[Emile Castelnuovo]

Leonard, you should also try to use a higher (much higher) pitch on the bottom as well. I found with my coax X8 that this was the best combiantion.

On my setup I have 13" propellers and used t-motor slow flyer 13x4.4 on top (mostly like APC SF) and Graupner e-prop 13x8 on the bottom. You could use also APC electric prop with high pitch too.

This by far is the most efective setup. 

I don't have any bench data as yours, but only a throttle hover as a reference.

The combination of t-motor on top and APC slow flyer on bottom, hovered with almost 55-60%, while with the 13x4 and 13x8 combination it hovered at 45-50%.

This will also help in fast forward flying.

Best,

Emile

2014-01-29 Leonard Hall <leonar...@gmail.com>

--
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/groups/opt_out.

[john...@gmail.com]

From what I have read in forums etc, the propeller difference in a coaxial setup should be about 20%.
Also, I think the difference should be in propeller pitch only, not blade size. I have a suspicion that using different sizes can degrade the airflow between the propellers.

- JAB

[Leonard Hall]

Hi Emile,

I did test 11x4.5 on the top, 11x7 on the bottom and 10x4.7 on the top,10x7 on the bottom. Both felt very good and were also some of the best pwm match. However, they weren't as efficient.

Hi John,

My understanding of the literature is that the relative size and pitch of the lower prop depends on the distance between the upper and lower propellers. Then there is the relative kv of the lower motor. However, in the Y6 the yaw control changes the power sent to the upper and lower props to equal the toque. This messes everything up. Then things get different again when hovering in stationary air or moving forward, yawing or depending on the disk load.

Hi all,

The other consideration I didn't mention was lost propeller or motor performance. While I didn't test that personally others reported much better control using identical on top and bottom. This is also makes me worry about using XXx7's on the bottom as they will stall much easier than the XXx4.7. Using the same props on top and bottom should result in similar recovery performance no matter what prop is lost.

[stdio9]

There is a NASA document from the 80's online that discusses coax & efficiency, with all the charts and tests. In fact, there are a few, if you google "nasa coax helicopter". 

I believe the "20%" number comes from the Nasa publication, and has been (mis)quoted all over the RC forums for years. A quick review of that document may yield some other interesting bits. 

That being said, it wasn't the Americans, but rather the Russians, who actually built and used a lot of coax heli's, so they might have more research/data. I'm not sure how well the conclusions scale down, but I'm very glad to see someone iterating and producing data for this question. 

-Mike

[Craig Elder]

Just adding the Nasa paper to this discussion

[john...@gmail.com]

Emin Bu posted this in the Some MultiCopter Design Thoughts thread. It's the same post I remember reading a long time ago and where I got the 20% from. I've highlighted the conclusion.

See if this helps clear up some of the questions. Here's a dissertation from Don Stackhouse:


OK, so we've learned that pushers are usually a detriment unless you really
do your homework, contra rotation is not generally worth the trouble on
models, but if we're going to do it anyway, we should try to keep the
airflow into both props as clean, smooth and uniform as possible. What's
that bit someone else mentioned about different diameters due to
"slipstream contraction", and what about the need for different pitches
and/or rpm's for the two props?

A prop makes thrust by grabbing chunks of air from in front of it, and
accelerating them out behind. About half the acceleration occurs in front
of the prop, and the other half behind. The reaction to the force required
to accelerate the air's mass shows up as thrust. Because the air has to be
accelerated to make thrust, the velocity of the air behind the prop is
faster than the velocity in front of the prop.

As the velocity changes, the roughly cylindrical stream of air flowing
through the prop has to obey Bernoulli's principle. If its airspeed
increases, then the cross-sectional area (and therefore the diameter) of
the stream has to decrease in proportion to that in order for the volume of
the flow to remain constant. If this were not so, the flow through the prop
would violate the law of conservation of mass and energy, which happens to
be one of the most inflexible laws in all of Newtonian physics. Thus, the
diameter of the inflow to the prop is actually larger than the prop at some
point upstream of it, then contracts during that first half of its
acceleration until it is equal in diameter to the prop when it reaches the
prop disk. It continues to contract after it passes through the prop,
during the second half of its acceleration. This is that "slipstream
contraction" that some other posters to this thread have mentioned. This
means that a second prop, aft of the first one, that is supposed to be
working with the slipstream of the first prop, needs to be a little smaller
in diameter in order to match the boundaries of the now-contracted
slipstream.

Just how much faster (and therefore how much smaller in diameter) depends
on a number of factors. For the ratio of slipstream dynamic pressure to
free-stream dynamic pressure, Daniel E. Dommasch's "Airplane Aerodynamics"
suggests an equation, which with a little algebraic juggling gives us:

Qt = Q + [(4 * T) / (D^2 * Pi)]

where:
Qt = dynamic pressure ("ram air pressure" minus the static pressure) in the
fully developed slipstream well aft of a prop
Q is the dynamic pressure in the freestream well ahead of the prop, and
outside of the propwash
T = thrust
D = prop diameter
and of course "Pi" is 3.141592...

Dynamic pressure ("Q") is equal to one-half the air density, times the
velocity squared. If we plug that back into the formula and do some more
algebra, we get:

Vt = SQRT [V^2 + (8T / rho * D^2 * Pi)]

where:
Vt = the velocity in the fully developed freestream in feet per second
"SQRT" means you take the square root of the result of the formula inside
the [ ]
V^2 = the freestream velocity squared (velocity in feet per second)
T = thrust in pounds
rho = air density in slugs/ft^3 (.00238 at sea level standard day
conditions)
D^2 = prop diameter in feet

Other units will work as well, just make sure that you use the same system
of units throughout (no fair mixing feet in one variable with inches in
another, or metric units with English, etc.!).

Ok, now that half of you are getting glassy-eyed and most of the rest are
running for cover in a mad panic, let's clarify that terrifying blast of
algebra with a practical example:

Suppose we have a twin-engined model that weighs 1 pound, and we're
planning to modify it into a twin contra-rotating arrangement. Let's also
assume that the L/D (essentially the same as the glide ratio) at our
expected cruise speed of about 25 mph ( multiply by 22 and divide by 15 to
get 36.67 fps) is about 4:1 (I know that sounds low, but remember, typical
cruise speeds are higher than best gliding speed, and besides, this
airplane has a bunch of extra stuff hanging out in the breeze). This means
our drag is equal to the weight divided by the L/D, or 0.25 pounds. In
level flight, that is also equal to the total thrust.

Let's also assume the front prop is doing about 55% of the work (0.138
pounds of thrust) to allow for the lower efficiency of the aft prop. We'll
define the prop as having a 6" diameter (0.5 feet).

Plugging all of that data into our formula:

Vt = SQRT [36.67^2 + (8 * 0.138 / .00238 * 0.5^2 * 3.1416)]

which is equal to 43.99 feet per second, or 30 mph. That's a velocity ratio
of 1.2, or 20% more than the freestream velocity.

This means that if the aft prop is far back enough to sit in the fully
developed slipstream from the forward prop, it will need either 20% more
pitch (the preferred solution) or 20% more rpm (which opens several other
cans of worms). In addition, the slipstream contraction will be SQRT
(1/1.2), or 0.913 . That means the aft prop should be 91.3% of the diameter
of the forward prop, or just a little less than 5.5" diameter. See, that
wasn't so hard, was it?

If you plan to do this a lot, I suggest coding these formulas into your
favorite spreadsheet program, such as Excel.

I helped advise a guy recently who scratch-built a VERY giant-scale
electric model of the Voyager. As I recall, his original setup used the
same size props on both ends. It flew much better when we put a prop with
more pitch on the aft motor.

So, that's all there is to it! Just correct for slipstream effects on the
rear prop, and keep the inflow into it as clean and undisturbed as
possible. You will probably not have as much prop efficiency as a pair of
tractor props with nice clean inflow, but it shouldn't be too bad."

[Marco Robustini]

nteresting discussion, I follow with interest.
Personally I abandoned coaxial configurations because they have only an advantage in my opinion, good thrust considering the small size compared to a flat, all the rest are only disadvantages.

Marco

[Leonard Hall]

Hi all,

My interest in coaxial props is because I am building a compact camera ship for my 4wd trips. I like the ease the Y6 is folded and it's ability to handle lost motors or props.

A Y6 can loose 1 motor or two opposite rotation motors on different arms and still maintain full control. I can also loose one motor on each arm and still maintain roll pitch control.

The obvious downsides of the coaxial based frame is the lost efficiency and yaw/lift coupling on the Y6 frame (X8 doesn't have yaw/lift coupling). However, both of these are a lower priority in the application I have in mind.

Leonard

[Jonathan Challinger]

Those pwm mismatches are huge! Going to have bad yaw performance etc... Maybe people should be steered towards an X8 if they don't have a very specific need for compactness?

Why is there still such a big pwm mismatch with a 10x7 on the bottom?

[Leonard Hall]

Hi Jonathan,

Yeh, the mismatch is large but I didn't find it noticeably different going left or right. The only time I really saw it was when I was doing nose in spirals. One direction was definitely slower than the other then, but in this case the throttle is pegged.

I was surprised that I couldn't make the toque of the upper and lower motors match at closer pwm using XXx7's on the bottom. It is a much more complex problem than a google search might suggest. I suspect that this may be partially because of the size of the props and the turbulence over the frame.

Leonard

[Romain Chocart]

Hi,

To Marco:

You did not notice a better resistance to wind gusts on coaxial configs?

I built my Y6 only for that. And so far it's hard to conclude, it looks like it moves on gusts, but does not take big horizontal changes, like "big pancakes" can do. An the other hand, non coax configs looks to have a more precise and soft control in flight.

My Y6 is compact (good for wind resistance), but it has big props (14 to 16 inch) which may be worse than smaller props dealing with stability in wind. Another subject to be studied..

General discussion:

Conclusion: instead of bigger props downside, it is more efficient to have bigger pitch.

I will make some tests on my "big" Y6 config and let you know.

Now I am with 14x5.5 on both sides and plan to swith to 15" props. 

I will try 15x5.5 up, and 15x7.5 down.

But the RCtimer 15x5.5 props are much stronger than the very thin 15x7.5 ones, so I will probably go for everyday flights with 15x5.5 on the 6 motors.

Question: how do you measure PWM mismatches? Is it the average difference of PWM command sent to ESC between top and bottom? This can be read in telemetric datas?

We could consider a way to increase efficiency on configs with only one kind of props: have a param in the soft for bottom / top power ratio, as spoken there. 

But the conclusion is maybe that Y6 structure is by nature not symetric and hardly can lead to perfect behaviour without specific tunnings, as X8 can do (the power ratio would be also efficient for X8 configs, allowing bottom props to run faster)

[Ian Ray]

Hi Leonard,

Just to clarify, were these tests done with the original Y6A rotation pattern, or the newer Y6B?

http://planner.ardupilot.com/wp-content/uploads/sites/5/2013/12/APM_2_5_MOTORS_Y6A_Y6B.jpg

- Ian

[Jonathan Challinger]

Newer Y6B.

[Jonathan Challinger]

Hate to necro this thread but I would suggest that these test results are not valid for an X8, and that we possibly have an extra degree of freedom to optimize X8 efficiency since the top-bottom throttle ratio is potentially free to change.