Run 5 motors off 4000 mAh battery and Matek F765 flight controller?

PsyBorg

Wake up! Time to fly!
Honestly if I were going to attempt something like this in a model I would use a differential from an RC car to drive the props using gear ratios for RPM and keep the motor higher voltage lower kv for the torque and efficiency. Just having the bulk of weight on center line alone would save tons of power wasted to compensate for the extra weight out over wings and such places.

Same thing for the other person working on an Ospry.
 

OwenN

Active member
Sounds complicated? How do RC helicopters work? they need quite low rpms.

I think the 2805.6 motors are quite high efficiency, but twice the weight of a 2200 motor.

I think the reasoning there is that the motor weight is small in comparison to battery weight.

You really need a fully contained ready-made gearbox with all the bearings and output shaft, retaining thread, and nut, needed to support the prop and resist thrust.

Thrust efficiency is 32 % of power, but most of the losses are due to the static thrust losses of the prop. I think the motor should be over 60% efficient at this loading, though I can't find any actual specs- maybe you can?

At 60 % motor efficiency, torque = 5 N-m or 3.7 ft-lb.

Compared with actuators, this is not a lot.

Maybe RobotShop has something, though step-up gearboxes are not often required.

I think you would get better efficiency running the motor faster with less load, and then stepping down the RPMs, as per helicopters.

The model could cope with 12 inch props -up from 9 inch props, with very little modification, and that would give big thrust efficiency savings.

The motor needs to produce less power for the same thrust.

At equal thrust, doubling the diameter halves the power needed. In this case, it would be 76 % of the power, and current.

If the motor was run faster at lower load, and factoring in improved prop efficiency at lower air velocity,

You could easily double the power-thrust efficiency!

With very little increase in weight.
Or use a lighter, lower efficiency motor, don't gain any weight, and still save a lot on power-thrust
efficiency.

Here is a 2.5:1 reduction- I got the tip from a plan by Jack Lynn Bale of the Chance-Vought V-173 (flying flapjack).

Master Airscrew 3025K AS Gear Box 2.5:1– Hobby City NZ
 
Last edited:

OwenN

Active member
That looks a bit expensive, plus I lose the pitch speed I need for forward flight.

That raises another possibility, however . Does the 12 inch prop fit on the motor torque-rpm-current line?
It won't make as much power, but it only needs 75% of the power to make the same thrust-probably less, due
to greater prop efficiency.

You need the right kind of motor for this-though.
Some will emit the Magic Smoke with the wrong prop.
Others, like quadcopter motors, are only bothered by going outside the voltage rating.


for the motor, 75% output power = 75 % rpm.
Prop:
V : m/s
dens = 1.2 kg/m3
T=2xdensxVsq x A (Newtons) : Kg x 9.81 (G) m/s2 : final units = kg-m/s2
Power=TxV (Watts)

Motor:
Power = Torque x revs ( Watts N-m, rad/s)

For the prop, if old speed = 10,000 rpm, to make 2016 g thrust vs theoretical 3600 g thrust, then the new speed is 6810 rpm.
old V = 21 m/s, new V = 14.3 m/s.

The old prop produces 36 N in theory, 21 N in practice at assumed 10,000 rpm.

Motor consumption is 867 W, makes 756 W (from theoretical thrust calculations) real, so it is 87% efficient.
(Based on rpms assumption).

Torque then becomes 756/(10,000 x 0.105 rad/s per rpm)
= 0.72 N-m


Then power needed is: TxV = 514W at the prop at 6810 rpm.

Thus this motor can make the same thrust with the bigger propeller.

Torque would need to be 0.72 N, identical!
Motor current should then drop to (6810/10000) = 68% to 37 amps.

Initial prop efficiency seems to be around 55 % at 10,000 rpm (assumed), for static thrust.
At torque needed of 0.72 at 10,000 rpm, assume blade drag acts at 64 % of blade length,
and CL of 0.03, drag torque would be 0.4 N-m, = 55 % of required torque.

-this approximation is probably out by miles, due to the Vsq relationship.
Breaking blade area into slices and adding drag from each slice would be closer.
-likely 2/3 of this value. 38% of required torque.
- still
This drag addition explains excess torque requirement well,

and actual rpm would be closer to 8,000 static, I suppose.

Very little inefficiency can be left to attribute to using only 2 blades, and for tip circulation.

More blades only picks up a little thrust, and adds quite a lot of drag.

(Rpm could be as much as 12,500 rpm for a 4S, 1800 KV motor, and this kind of load.)

Prop efficiency will be better at 6810 rpm static thrust, 12 inch prop.

This is where you could pick up some lower amp use, by dropping rpm a bit more to take advantage of better prop efficiency.



If efficiency varies by V squared for static thrust, then the new efficiency would be 65 %- this is in line with drag.
(This doesn't account for tip circulation, which would be more complex).

New thrust = 1.1 x old thrust, up by 10%

if we reduce thrust to 91% of the new value (100/110) , velocity drops by 30 % sqrt(0.09), rpm drops by 30%,
and power need varies by 30% down also; (TxV).
Torque is the same, so current and rpm drop by 30%
to 4767 rpm and 26 amps.- a good saving if I start at 54 amps.

Very likely the top speed would not vary much either.

The drag thrust requires a smaller drop in rpm if we regard the prop will still spin a similar rpm at speed.

This is not strictly true as prop surface drag rises with diameter at the same RPM, plus we have profile drag, of course.


For practical use as a VTOL, tilt-prop, initial power consumption doesn't matter much, because you generally don't do a lot of hovering.

During normal flight, current drops to more reasonable levels, plus you have the 70% cruise speed option.
 
Last edited:

OwenN

Active member
Metal screw heads/horns look like a good idea.

Unfortunately not available for the emax ES09md servo.

I shall have to epoxy a 1.5 mm aluminium washer onto the cross-shaped arm-head. 20mm dia. should do.

If I sand the washer, degrease the plastic part, and slather the epoxy on, it should hold.

Then I can use very short self-tapper screws plus loctite to hold it to the motor bracket.

The servo can also be epoxied onto the main mounting channel. If it breaks, it can be chiselled and ground off!

The ply main bearers can also have aluminium plates bonded to them for the mounting channel screws to lock against,
otherwise they will shift under vibration.

At the other end, the stub-tube inner shaft can be bolted to another washer, then self-tapped onto the main
motor mount bracket, with loctite.

Otherwise the assembly cannot be assembled! :)

I have found some small plastic shoulder-bushes, or shoulder-washers, which can be fitted to the brass inner tube shaft.
The shoulder bushes can be epoxied in place to stop them spinning.

Drawings to follow!

Attached- picture of some metal servo horn-tops.
metal round servo horns.jpg
 

OwenN

Active member
Rework of prop efficiency
assuming rpm = 8000.
Summing slices method:

sum of Vsq x A x r = 0.795 m^5/s(sq) over 8 stations x2, v1 = 30 m/s,39,47,56,64,73,83,89.
where v1= rad/s x r1 = 832 x 0.036m = 30m/s
Vsq = 900,1521,2209,3136,4096,5329,6889,7921.
VsqA =.13, .26,.4,.53,.7,.8,.96,.95 where A=(x10^-4) 1.5,1.7,1.8,1.7,1.7,1.5,1.4,1.2.
check: 900 x 1.5 x 10^-4 = 0.13 - good so far.
Vsq Ar = 4.7,12.2,22.8,36,53.9,70,96,102
Check: 0.13 x 0.036 = 0.0047
so total should be multiplied by 10^-3 -OK
grand total = 397.6 x 2 = 795.2 x 10^-3 = 0.8
multiply by 2xdenxCa should give torque. (x0.072)
= 0.0576 - this seems to be out by a factor of 10

a rough calculation gives 50 x 15 x 10^-6 area = 7.5 x 10^-4 (x2) =15 x 10^4
and r=0.09m
F = 2xdenx Vsq A x 0.03
V = rad/s x r = 832 x .09 = 75m/s
F= 6 N
Tor = 6 x .09 = 0.54 N-m
So calc with slices must be a bit out somewhere.
-------
say F = 0.58 N
then Tor seen by motor =:
---------
Thrust = 28 N in theory.
Thrust = 20 N actual - attribute to tip circulation, dead area on disk, variation due to only 2 blades.
motor power needed = 476 W
torque needed = 476/832 =0.57 N
add drag = .57 + 0.58 = 1.152 N
and power needed = 1.152 x 832 = 958 watts.

so-
*** drag coeff Ca is more like 0.01 for this propeller ***

This is not unheard of-possibly due to small bernoulli number...
Largish velocity but small active length, or blade width in this case.

drag torque = 0.19 N,
---------
Torque needed = 0.78 N, and power needed is 632 W,
200/632 = 31.6%

Power use is 1.32 times expected. 1/1.32= 0.76
giving motor efficiency of 76%

and prop efficiency of 1/( (8/28) + 1) = 78 % (prop thrust is 8 N less than expected.)

Both those figures sound quite reasonable!!! :)
My guess of 8000 rpm must be close.
<edit>
Another factor to add: Transmission, or torque efficiency.
= 1/((drag proportion of theoretical) + 1) = 75%

Now, if I multiply all these together as decimals, I will get actual thrust efficiency as a proportion of ideal (frictionless).
= 0.76 x 0.78 x 0.75 = 44 %

I think I have calculated all these decimal fractions in such a way that they can be multiplied like this.
This method expresses loss amount as a proportion of the theoretical amount.
Any comments from math or statistics experts?
 
Last edited:

OwenN

Active member
I have been watching the videos on bi-copters, and I think I can trim my design to work exactly like a bi-copter.
at the moment, the MAC is aligned with the props when tilted, and the props are located well above the COG .
It I trim COG exactly on the MAC, it should fly the same. There is a Bi-copter program in Ardupilot, I think.-It does several configurations.
It needs 3 mode signals; bi-copter, transition flight, and normal airplane mode. On approach, I can throttle back, change to bi-copter while
still flying above stall speed, then fly pitched up until speed drops into proper bi-copter range.

Pitch sensitivity due to the rearward COG can be countered by gyro pitch control.
Thus, the front thrust tube is not needed at all, which will save weight, and cost.
Also, no extra control routines are needed to run it.

Transition flight is also implied if switching from bicopter to normal flight. I will check if the motor tilt sequence can be programmed in and varied, to suit the drag-induced pitch-up on a full throttle launch.
-I will look for Ardupilot setup videos and websites.

Wing construction method:
It occurs to me that each wing can be built flat, for the wing frame, anyway.

The main ribs can be set in at an angle, using an angle template. This simplifies building in the mating face for the wing root,
and attaching the motor bearing plates.

I still need to build the wing root with gaps, to slot the angled spar lap plates in.
Maybe use top and bottom gussets to hold the rib assembly together, and make the lap plates less than full height.
I can also add location guides for the lap plates.

Any doubled ribs can still be installed square to the skin, to remove the need to sand them at an angle.

The topping strips for ribs and web spars can be cross-grain 1/16 balsa, and fitted up without sanding angles.

They can be split and attached each side of ribs and spars. Cross-grained balsa should adapt to curves in 2 dimensions.

They can be attached with balsa cement/pins, or CA glue/pins, then any V-gaps back-filled with epoxy glue.

The wing is thick enough that this can be done for the "top" joins once the frame is lifted off the building board, and before the skin is attached.

The "topper" strips can also be used for lap joints in the skin, so the skins can be put on in segments, to suit the 15 minute epoxy
glue.
 

OwenN

Active member
https://discuss.ardupilot.org/t/bicopter-ardupilot-firmware/39651

Bicopter transition to VTOL may be a little tricky to get working, using Ardupilot on Matek.
This looks like the correct form to discuss it.
I will watch other videos using the bicopter-vtol transition.
<edit>
This guy, Tom Stanton, used an arduino board as a back-end controller.


It needs 2 lines in from the receiver, though.

I can link the prop tilt servos to the elevon servos, but at a different angle rate and stop points.
Then roll-input converts to yaw-effect at the props.
Elevator rear-up, front-down converts to motor tilt-up.
This allows me to fine-tune prop tilt to keep it facing up as the plane pitches up.
As seed reaches flying speed, elevator goes to center, so does the prop tilt. I can disengage tilt at this point.

This would work well for the launch phase, but not so well for landing.
I would want full motor tilt at speed, then link back to the elevons once forward speed was zero.

The craft should be pointed into the breeze, and this may act on the control surfaces , as well.

In the event that Ardupilot doesn't allow this level of programming, this is an option to consider.
 
Last edited:

OwenN

Active member
Quote:

From RC-Groups, 2013. I think there is a Flite Test post as well?
https://www.flitetest.com/articles/semi-scale-vtol-v-22-osprey

Originally Posted by jk6stringer
Thank you. Something to think about. At least with math I can use a calculation vs a pure guess.
In regards to the hinge, is that to say that by securing the hinge to the plane that much of the load is taken up, and less impact is placed on the servo directly?
If seen so many were they are direct mounted, but this stands to reason as well.
Thank you
There are multiple forces involved and as Parky said, it is a messy calculation, but the essential issue is that when in forward flight the advancing blade has more airspeed than the retreating blade and generates more lift. In theory this would generate a roll force, but since the tilt hinge only allows the motor to tilt in pitch it is not a problem for the tilt servo. However, it is not that simple. Since the propeller is spinning it is a gyroscope and the force is translated 90 degrees in the direction of rotation by gyroscopic precession. That would turn a roll force into a pitch force and make it a problem for the servo. However it is not that simple either. If the propeller blade were perfectly stiff then this 90 degree precession would not occur. If the propeller blade were perfectly floppy it could not transmit a force to the hub. Somewhere in between where real propellers live only some of the force gets translated to pitch and in my personal experience it can be enough to stall a servo under certain flight conditions.

The following text is taken from the VTOL-Trainer Build instructions. It uses 4 each GWS 9x5 3 bladed props. In this case each servo was tilting 2 of the 4 motors-propellers.

Servo Upgrade

The torque provided by the originally recommended motor tilt servo, Hitec, HS-85MG+, is insufficient in some flight conditions and can result in a crash.
I recommend that the servo’s be upgraded to HS-225MG. I also recommend that the Castle Creations SBEC be programmed to provide 6V to the tilt servo’s and RX. The Flight Controller and the other flight servo’s can still be powered with 5V from the BEC’s within the various ESC’s.

The HS-85MG+ is rated at 48.c oz. in. of stall torque at 6V
The HS-225MG is rated at 66.65 oz. in. of stall torque at 6V

If the aircraft is placed into a shallow dive prior to the transition from hover mode or slow forward flight (SFF) mode to fast forward flight (FFF) mode, the aerodynamic forces on the rotors can prevent them from tilting forward as intended. This can result in an asymmetric motor tilt from side to side, and a loss of control. In some cases the rotors on both sides can fail to tilt forward and the aircraft will remain in hover or SFF configuration longer than intended. If the flight controller used disables stability feedback to the motors, this can result in a multi-copter without gyro stabilization, which is very difficult or impossible to fly. If the aircraft is built with the stock KK2 firmware, the stability feedback to the motors is never disabled so the aircraft will remain flyable if the rotors on both sides fail to tilt. However, the flight controller will not be able to fully compensate in the event of an asymmetric tilt situation.

This issue only impacts outbound transitions from hover mode or SFF to FFF. Inbound transitions from FFF to SFF or to hover are not impacted. Outbound transitions at normal (slow) flight speeds are not impacted. That is partly why this problem was not noticed for a long time.[end quote]
It looks like I need a much bigger servo.
66 oz-in = 4.7 kg-cm
I will see if I can find one with the double ball-races.
 
Last edited:

OwenN

Active member
Here is an EMAX one that would do. It is a little expensive at $42.45 US each plus freight.
I need to find if I can get the internal spec in a plastic case design? - needs 6V to do 5 kg-cm,
but goes right up to8.4 V, so it should be pretty rugged.
I will redo my drawings to suit this one.
EMAX ES9252 ex emax-us 42-95.jpg
 
Last edited:

OwenN

Active member
Here is a similar one, available locally, $49.95 NZ , so quicker to get, less freight. Slightly heavier, bigger, so I will dimension to this one.
savox mg 2 ballrace servo 1.jpg
savox mg 2 ballrace servo 2.jpg
 

OwenN

Active member
It strikes me that a nice addon for my model would be a pair of wingtip cannons off a star wars x-fighter. Anyone got a pdf of the plan drawings for one?
<edit> never mind, I have found some. Probably leave that reflector-bit off, change the tip detail a bit?
x-wing 2.jpg
 
Last edited:

OwenN

Active member
here is a version with ducted props. Probably not suitable for STOL operation, though. It may work as the props are better aligned
with the front wing, and the front wing elevators may generate enough lift.

However, the thrust line is away from the COG, and the props are back further, so it is dependent on blow-over
on the control surfaces for stability. A bit of gyro on pitch is needed.

The little gap on the upper-inner prop duct should be closed and faired in.
A construction advantage is that the duct is built into the wing, so it needs no locating webs, and also can be built lighter.
Would any efficiency gains offset profile and surface drag?
You would need to run it through a computational aerodynamics simulator to find out.
The thin aerofoil-shaped casing wouldn't generate much Coanda thrust.
Ducted Prop Mod 30-12-20.jpg
 

PsyBorg

Wake up! Time to fly!
I use various ones as I have been in the hobby long enough as well as learned stuff from all the incredible people here and reading pretty much every post whether it interest me or not.

I basically first search the actual manufacturer of the motors which in this case came up with squat. Usually that is a sure sign of a Chinese knock off of something so I drop the name and just search the motor specs and that usually pops up several makers that actually do make their product and do the testing and research.

Then if I want more info specifically I search RCtestbench and various guys pop up that do similar testing with various names. Browse those for the info I want until I find it.

Then as a last resort I go to youtube university and search the brand and again working backwards to just motor specs. There is usually always some kind of review and there you can tell if what you are looking for is junk by the comments. Content providers portray the product as best thing since sliced bread as it is their job as a presenter. The comments are from the real people who actually used the products and tell their real life experience and not just a use or two to make the video.