Homepage Knowledgebase GitHub

TalTech drone design considerations

This document contains some considerations related to a demonstration drone with one of the Telega based motor controllers. Two controllers are reviewed below:

  • Sadulli. Fits smaller drones with a payload capacity of 1-4 kg and flight time up to 1 hour
  • OpenMyxa. Intended to be used in bigger drones with a payload capacity up to 15 kg and flight time about 30 minutes

Small drone considerations

There are two options of Sadulli drive:

Sadulli characteristics

Parameter Sadulli Grosso Sadulli Piccino Unit
Mass 207 66 g
Optimal thrust 1500 1000 gf
Max thrust 3000 1800 gf
Maximum continuous DC current(with 8S battery) ~20 ~10 A

Sadulli Piccino thrust table


Let’s make some suggestions about Sadulli Piccino driven drone.
Sunnysky kindly provides the thrust table for V4006 motor with 15-inch propeller. Although this propeller slightly differs from one that is used in Sadulli Piccino, we will ignore this fact. We also have reason to believe that Sadulli can be more efficient than a Sunnysky controller. To maintain efficient flight the drive system should output no less than 10 grams of thrust per watt.
According to the table above this means that propeller should not be spun faster than 4000 RPM which conveniently can produce 1000 gf of thrust.
Sadulli Piccino itself with propeller weights ~150g(129g Sadulli + 21g propeller). Carbon fibre beam with 3d printed motor mount and motor wires adds 50 more grams. This leaves 800g/beam for the central plate + battery + payload.

  • Quadcopter - 800 * 4 = 3.2 Kg
  • Hexacopter - 800 * 6 = 4.8 Kg

Couple words about the battery. Each Sadulli consumes about 100w during an efficient flight(specific thrust > 10 gf/W). So at 8S (~30V) each Sadulli consumes 100/30 ~ 3.5 A(some margin added).

At maximum thrust power output is 3 times higher, so each motor will consume about 10 A In the case of a quadcopter, the battery should be capable of delivering up to 40A and 60A for hexacopter. There are two approaches to battery problem:

  • Creating a battery best suited for the application from single cells
  • Searching for a COTS battery

Battery capacity depends on the desired flight time. Let’s say we want 1 hour (this also eases the calculations). Each Sadulli motor consumes about 100W of power during an efficient flight. This gives the necessary energy of

  • 100 * 4 = 400Wh for quadcopter
  • 100 * 6 = 600Wh for hexacopter

At this stage it is possible to form the requirements for the battery for quadcopter:

  • Number of cells - 8 (nominal voltage 30V)
  • Discharge current - 35-40A
  • Energy 400 Wh
  • Mass < 2Kg
    Possible Candidate: MaxAmps 8S 16Ah 2.7kg $700

Requirements for the battery for hexacopter:

  • Number of cells - 8 (nominal voltage 30V)
  • Discharge current - 50-60A
  • Energy 600 Wh
  • Mass < 3Kg
    Possible Candidate: MaxAmps 8S 22Ah 3.4kg $900

In case battery construction approach is selected:
One of the best cells for medium power battery construction at the day is LG INR18650-HG2.

  • Battery Capacity (Mfg Nominal): 3000 mAh
  • Battery Chemistry: Li-Ion
  • Battery function: Rechargeable
  • Battery Form Factor: 18650
  • Battery Rated Voltage: 3.6 V
  • Max current output - 20A
  • Mass 45g

UAVCAN BMS are available or soon to be available from ARK Electronics, NXP, CUAV, and possibly others. Some relevant resources are linked below:

We will not be engaging in the design of a custom BMS.

Quadcopter
In order to meet the requirements (almost) 8S4P battery is needed, giving total 8*4 = 32 cells. This will be a battery with ~1.5Kg mass and ~350Wh energy capacity which is close enough to the desired specs.

Hexacopter
In order to meet the requirements(almost) 8S6P battery is needed, giving total 8*4 = 48 cells. This will be a battery with weight ~2.2Kg and capacity ~540Wh which is close enough to the desired specs.

With COTS batteries the results tend to be similar or worse.

Of course, this is not the complete drone. But it seems to be possible to design a drone with a one hour flight time based on Sadulli integrated drive. And it will have following thrust margin

  • Quadcopter: 3.2 - 1.5 = 1.7Kg for construction elements, on-board electronics and payload.
  • Hexacopter: 4.8 - 2.2 = 2.6Kg for construction elements, on-board electronics and payload.

The statements above are correct in case of efficient flight mode when hovering in place with no wind, the rotors are equally loaded, and normal atmospheric pressure. Real-life conditions are hard to predict.

From my experience, electronics and all the necessary construction elements for a drone should weigh less than 400g, which still leaves a wide range of payloads that will fit this project. For example, the body of Sony A7R(camera that is typically used in plane surveying) mass is 465g and average lens mass is ~0.5Kg, so in case some lightweight gimbal can be found this should work even with a quadcopter (still flying in efficient mode)

Sadulli Grosso thrust table


Sadulli Grosso propeller weight is 41g
Same considerations as above are applicable to Sadulli Grosso
Sadulli Grosso and propeller weight ~ 204 (Sadulli) + 41 (propeller) = 245g
Sadulli Grosso max efficient thrust - 1500g
Sadulli Grosso power consumption in efficient mode ~ up to 150W
This gives 1500 - 245 ~ 1250 g of pure thrust per motor in efficient mode (which is about x1.5 times more than Sadulli Piccino).
So, in theory, Sadulli Piccino hexacopter should have very similar characteristics to Sadulli Grosso quadcopter. Sadulli Grosso hexacopter should have about 1.5 times more payload mass which is around 4Kg. All stated above supposes that the aircraft is always in highly-efficient flight. So there is a good margin in thrust.
Sadulli Grosso powered aircraft may lift up to 8-10 kg of payload if efficiency and high endurance is not the purpose. Just as a reference, this bathroom sink mass is around 7 kg. Propeller efficiency with such a payload will decrease around 35% so it should be possible to hover with something like that for 60 min * (1 - 0.35) ~ 40 minutes.

COG estimations

Sadulli Piccino quadcopter

Name Price Qty Total
Sadulli Piccino €136 4 €544
Miscellaneous electronics €1000 1 €1000
Battery €750 1 €750
Frame €400 1 €400
Battery charger €300 1 €300
Tools, 3d printed plastic parts and other miscellaneous parts €200 1 €200
RC transmitter and other electronics for tests and debug €1000 1 €1000
Total €4194

Smallest possible option. Estimated flight time with 1 kg payload - 1 hour.


Sadulli Grosso hexacopter

Name Price Qty Total
Sadulli Grosso €157 6 €942
Miscellaneous electronics €1000 1 €1000
Battery €1000 1 €1000
Frame €400 1 €400
Battery charger €300 1 €300
Tools, 3d printed plastic parts and other miscellaneous parts €200 1 €200
RC transmitter and other electronics for tests and debug €1000 1 €1000
Total €4842

Medium powered option. Estimated flight time with 4 kg payload - 1 hour.

Big drone considerations

There is also an opportunity to build a bigger drone for much heavier payload. Although it is not absolutely clear what type of payload may need such a thrust margin (it should be made up).
Bigger drones need bigger propellers. These bigger propellers need bigger motors that can spin them. Bigger motors need a controller for higher power levels. Luckily Zubax has one suitable controller.

OpenMyxa

  • Continuous power output up to 1800 W
  • Supply voltage - 8-55V (4-12S LiPo battery)

This narrows the field of possible motor/propeller combinations that may be used. In the case of two blade propellers (which are the most efficient ones after single-blade propellers) its diameter should be around 30 inch.
EOLO CN28*9.5 looks reasonable.
Brief specs:

  • Weight - 132 g
  • MAX RPM - 4850
  • Folding design
  • Price $32 @ 1 pc.

For spinning a propeller of this size a reasonable big motor is needed. Sunnysky manufactures several motors that can fit:

Motor X8016S V8117 M8 pro V2
Mass (g) 560 N.A. 247
Max Power (W) 3025 N.A. 1100
Price ($) 163 193 179

X8016S and M8 pro V2 both seem reasonable, but the most comprehensive data is available for X8016S so let’s stick with it just yet

Although data provided is relevant for 30 inch propeller difference with 28 inch shouldn’t be huge and all the trends will be the same.

To achieve efficient flight with this motor and propeller RPM should be kept under 3000 which gives thrust of about 5 kg with power consumption of about 500 W. This still leaves the possibility to double the thrust which may be needed for maneuvering or during ascending but the craft will fly in less efficient mode. In the case of a hexacopter it may be possible to achieve a total thrust of about 30 kgf in efficient flight mode. As a rule of thumb, payload may take up to half of that number leaving 15 kg for the drone itself

Battery should be able to provide such current, which means in case of 12S battery it should be able to provide 1000/(12*3.7) ~ 22.5A peak with an average of ~12A per motor.

Needed battery capacity is determined by the desired flight time, flight mode and drone mass. Of course, all these variables can’t be predicted in this document, but for rough estimation, let’s assume drone mass is 30kg, it flies in efficient mode only (specific thrust 10g/w or better) and desired flight time is 30 min. This yields the required battery energy of

30 000 / 10 * 0.5 = 1500 Wh

To determine the last limitation on battery construction some basic mass calculations are needed.

  • Total thrust margin - 15Kg
  • Motor (M8 pro V2) + propeller mass - 330 + 132 = 462 g. Some nuts and bolts will definitely be needed, so this number can be rounded to 0.5 kg
  • As the proposed drone configuration is hexacopter all the motors and propellers mass is 0.5 * 6 = 3 kg

This leaves 15 - 3 = 12 kg for the frame, electronics, wires and battery.

The heaviest electronic part is OpenMyxa with its mass of about 100 g and as 6 of them are needed they will take 600g. There are probably no more electronic components of considerable mass, so 400g margin for all the rest seems reasonable.

So, the battery and frame mass combined should not exceed 11Kg. As the frame mass heavily depends on the construction, materials applied and additional features (like landing gear) it is hard to predict. As a rough estimation the frame mass may be somewhere around 5-6 Kg.

Battery requirements are:

  • Voltage 12S
  • Max continuous current - 22.5 * 6 = 135A
  • Energy capacity1500 Wh
  • Mass < 5 kg

At first glance, these requirements seem to be unrealizable. But this is only because they are. Still, some solution should be found and it should be as close to the requirements as possible.

In the image below some readily available battery assemblies may be observed.


It is obvious they don’t match the requirements a lot. Because of this and also because it may ease the frame construction a creation of a custom battery should be considered. And as with smaller drones, the most versatile building block for such a battery is 18650 cells. For example,

the aforementioned LG INR18650-HG2 or Samsung INR18650-30Q.

12S12P battery configuration that uses cells above should provide ~1500 Wh of power and be able to withstand 180-220A current consumption.

The only obvious downside of such a battery is its mass and size. Its mass cannot be less than 7 kg. Also its size will be so big that it definitely should also be used as a frame part.

COG for such a battery is around 1000 EUR (~600 for the cells).

Considering the usage of ready available hobby-grade LiPo battery packs (which is common in the industry), it should be mentioned that they are typically heavier. For example, Turnigy High Capacity 14000mAh 6S (https://hobbyking.com/en_us/turnigy-high-capacity-14000mah-6s-12c-multi-rotor-lipo-pack-w-xt90.html).

This battery contains roughly 310Wh of energy. 2 of them are needed to make a 12S battery. So in order to create at least half-decent battery 4 such packs are needed. Mass of 1 pack is 2280g, which means the battery of insufficient (1240Wh vs needed 1500Wh) capacity will have mass over 9 kg.

One more approach may be considered: constructing the battery out of rectangular LiPo cells. According to https://lpbattery.ru/ there exist 11122215-30000 type of cell which is a single LiPo cell with 30Ah capacity and rated for up to 5C (150A) discharge current. Its mass is only 540g. This enables to construct a battery comparable to 18650-cell battery.

COG estimation for OpenMyxa powered hexacopter

Name Price Qty Total
X8016S motor $163 6 $978
EOLO CN28*9.5 propeller $32 6 $192
OpenMyxa motor controller €220 6 €1320
Miscellaneous electronics €1000 1 €1000
Battery €1000 1 €1000
Frame €1000 1 €1000
Battery charger €400 1 €400
Tools, 3d printed plastic parts, fasteners and other misc parts €250 1 €250
RC transmitter and other electronics for tests and debug €200 1 €200
Spare parts €300 1 €300
Total €6640

Estimated flight time with 15kg payload - 30 min.

Conclusion

Zubax robotics off the shelf hardware enables the construction of very different UAVs with considerably different payload mass. Although power level may vary a lot, costs for one-off prototype are similar. Thus the most value represents the biggest UAV with maximum payload. This is hexacopter based on OpenMyxa.

Construction thoughts

Drone mechanical arrangement

Although there are plenty of possible drone frame arrangements, one should be considered for this project - the H-drone.

A good reference for such a frame can be found here https://pinshape.com/items/31107-3d-printed-quadcopter-h-frame

The main advantages of this arrangement are

  • Mechanical simplicity
  • High rigidity
  • The ease of extensibility
  • Low price and the lack of complex custom manufactured parts.
  • Is applicable for quadcopters and hexacopters

Drone logical organisation

Typical drone logical organisation scheme can be found in the diagram below

The flight controller commands ESC using some kind of PPM protocol over a dedicated data line for each ESC. All the ESCs are powered from one battery, that is located somewhere in the center of the drone. All this requires lots of wiring and what is worse - high power wiring.

A novel approach of constructing isolated thrust modules should be considered. These thrust modules should include the motor, the ESC and the battery in one package. The amount of external connections should be kept as low as possible. Something similar may be found at https://www.crptechnology.com/windform-3d-printed-tundra-drone-functional-prototype.

It is very likely such a thruster should be made in the form of a beam of multirotor, so there is enough space for everything. In the case of the OpenMyxa hexacopter described above, such a thruster should carry a 12S2P battery, which means a total of 24 18650 cells. Drone logical arrangement then transforms.

The key advantage of such an approach is that the power source and power consumer are physically located in one compact module thus eliminating the need in long high power wires(that are usually relatively heavy). Also, advanced data bus allows for better telemetry collection and makes Flight controller independent of particular propulsion system implementation as all the data is available on the bus in form of standardized messages and there is no need to support dedicated means of health and status monitoring for each ESC. It may be beneficial from a mechanical design point of view too as in this case the battery mass isn’t located in one place. Instead it is spread all over the aircraft reducing the load on the beam to the central plate joint (which is one of potential points of mechanical failure).

Ideally, such thrusters should have some kind of quick connection system so that it will be possible to construct any aircraft needed (with 3,4, 6 or 8 thrusters) in case the central plate allows for that. Though this is dependent on the mechanical design and is out of the scope of this document. As these thrusters are to be developed it would be reasonable to form some requirements for them

Each thruster should

  • Contain motor
  • Contain motor controller
  • Contain battery of reasonable capacity and voltage
  • Communicate with outer world via 2 UAVCAN interfaces
  • Use dedicated balancing/recharging power supply connector

All other thruster properties are to be defined during further development.

On-board electronics

Except for propulsion system each UAV has some on-board electronics. Only UAVCAN-enabled devices should be chosen (if possible). Please refer to the table below for possible candidates.

Flight controller GNSS BMS
Holybro Durandal Zubax GNSS2 UAVCAN BMS from NXP/ARK/CUAV

Important notice

As the proposed project is all about aerial UAV development, it should be noted that such mechanisms are prone to catastrophic disintegration as a consequence of even minor malfunctions and/or design flaws. This gets worse if the development team isn’t experienced in the field (which is probably the case with student teams). The final project budget for parts and tools should be arranged making allowance for these issues.

Please indicate in the text that the TalTech project will be focused on the large drone option, so the small drone considerations can be ignored.

Upon a closer look at the motors, I think there is no point staying with X8016S because at 3 kW it is grossly overpowered; M8 v2 is closer to the optimum at 1 kW @ 250 g. I suggest marking it in the text as the preferred option and updating the BOM table accordingly.

There are gross formatting issues in the text, please proofread.

Above?

At maximum thrust, the power output is 3 times higher.

Alex, you can’t write text without Grammarly. Please enable it and fix the mistakes. (Also, you and Dmitry have a weird habit of starting your sentences with “And”, perhaps you shouldn’t do that.)

The statement about reliability is untrue, I wrote about this in the original Google doc. I suggest removing this sentence.

Please don’t hide your links under here because it makes the text nonsensical when copied. Instead, you could write something like this:

Something similar may be found at CRP Technology.

Or like this:

Something similar may be found at https://www.crptechnology.com/windform-3d-printed-tundra-drone-functional-prototype.


The following notes from Trello have not been addressed in your text:

  • We are going to build the high-power option (at the end of the doc). — please indicate which option is the final choice.
  • The design will incorporate modular propulsor units: battery + motor + propeller in a single unit. Each unit has two UAVCAN and one balancing/recharging power supply connector. — please emphasize the interface design.
  • The electronics will be based on: Holybro Durandal, UAVCAN BMS from NXP/ARK/CUAV. — please mention that the autopilot is going to be Holybro Durandal running PX4 v1.11.

Below is the description of activities in the form of a plan that will bring us to the implementation of the relatively high-power drone. As mentioned before, this drone should have highly modular propulsion system with battery spread all over the thrusters. This allows for easing the testing process.

  1. Component selection and procurement for preliminary tests of the propulsion system.
    This step considers the procurement of all the necessary components to test one thruster unit.
    This includes:
  • Motor
  • Propeller
  • ESC (OpenMyxa)
  • Battery

All the components were in fact already selected (see post above), so in the absence of additional opinions on this matter, no SELECTION needed.
At this stage no mechanical construction needed. The goal is to connect everything and answer the following questions:

  • Can X amount of thrust be achieved?
  • How much power from battery does it need to produce X amount of thrust?
  • Can the battery handle it? And if yes - for how long?
  • What is the maximum practical amount of thrust that can be achieved?
  1. Thruster mechanical design

At this stage, some kind of thruster support structure has to be developed. It needs to hold the motor, the ESC and the battery together in a compact way. Several key factors are:

  • Mass
  • Manufacturability. As we faced difficulties with local manufacturing of precise parts before, the design should be as simple(or even primitive) as possible. The ideal design is manufacturable using hand-tools. Although, it is understandable that it most probably won’t be possible.
  • Cost. Exotic materials or processing techniques should be avoided if possible.

There are two design challenges except the actual housing for the components when designing the thruster. At the moment the best representation of the thruster is that it has the form of the detachable beam of the drone. A robust and yet allowing quick disconnection without any tools joint is needed.

The second challenge is about electrical connection. Each has at least one UAVCAN interface (two connectors) and one relatively low-power charging and balancing electrical interface. All these electrical connections should be possible to connect/disconnect at once when the thruster is plugged into the beam, without handling individual connectors. This feature may be omitted in the first design iteration, but if possible - it should not.

2.1 Battery design is related to this step. In the case, no suitable COTS battery is found (which is very likely) a battery made out of 18650 cells should be designed and described. Probably it should have some kind of BMS embedded. Battery assembly process should be described extensionally as it may be a relatively dangerous process and involves some special equipment.

  1. Frame mechanical design

This is where all the thrusters will be plugged into. This piece will also hold all the on-board electronics and this is where the payload holder is mounted. The key design factors are the same as for the thrusters.

  • Mass
  • Manufacturability
  • Cost

It may be a good idea to go for more conventional frame design, that consists of two carbon-fibre flat plates, that are somehow connected. All the electronics may be placed between the plates and additionally protected with 3D printed canopy.

The minimum set of electronics that has to be placed here includes:

  • Flight controller
  • Radio receiver module
  • Zubax GNSS2 module
  • Zubax Babel
  • Some kind of FPV cam with the transmitter
  • Probably one or several HD action cameras (one, probably, gimbaled)

For the ease of development and manufacturing, it is recommended to mount the electronics on 3D-printed mounts whenever it is needed.

  1. Payload holder

At the moment there is no clear view of the payload type and function. Thus it is needed to implement some kind of mechanical and electrical interface that will enable the connection of different types of payloads to the UAV. As this is a cargo drone the first type of payload should be some kind of hold and release device that allows robust holding around 20 kg of payload. The payload should not rely on some special packaging and in the spirit of the overall design should be able to use the simplest packaging possible - a plastic bag.
The second probable payload type is some kind of a heavy video camera mounted on a beefy gimbal. Although this is out of the scope at the moment.

  1. Validating the overall design

At this stage, all the concerned participants will have the last chance to express an opinion on impliable design issues.

  1. Actual hardware manufacturing.

This step includes the procurement of all the necessary parts and manufacturing of all non-COTS parts described above.

  1. Mechanical assembly and wiring

  2. Configuration of all onboard electronics

  3. Test flights

  4. Performance demonstration

A specific demonstration of the drone capabilities should be made up. This demonstration should be as vivid as possible and highlight the drone’s strong points:

  • High payload capacity
  • Flight stability
  • High endurance
  • The ease of configuration using UAVCAN
  • High portability and the ease of maintenances that is achieved by the modular design

This demonstration should then be uploaded to social media in video format.

1 Like

We should keep in mind that we have access to the metal SLS printing facility at TalTech. I think Oleg Sosnovski is somehow related to that department (not sure).

How about EPM v3? It happens to support UAVCAN, too.

Looks very cool indeed. I totally forgot about its existence.
Although max holding force of 200-300 N seems to be on the lower side. I mean, the payload may just fall off during maneuvring.
I wonder if it is possible to use two of them to increase the holding force.

Task 1: The tests can be done in TalTech laboratory. Our staff members could make that kind of tests if all devices that needed to be tested will be purchased. Time schedule depend on number of testing components.

Task 2: I see this chapter as thesis for the student. Deadline for that end of 2020 or Spring 2021.
Sub task 2.1: as additional thesis

Task 3: Same as Task 2 - theses

Task 4: TalTech staff is working on several test benches for drones, payload holder may be a part of this research.

Task 5: TalTech staff will be responsible, based on theses timeline

Task 6: TalTech labs equipped with 3D plastic printers and it’s also possible to use CNC for prototype production (not mass production)

Tasks 7, 8, 9: may be included to T2 and T3 or as separate theses

Tasks 10: TalTech media and social networks will be used (researchgate, facebook, instagram, etc. ). Moreover, scientific publications will be used for dissemination.

@antonrassolkin Thank you, I think it makes sense. Now, when the rough plan is in place, what do you need us to do to move this forward?

I believe Pirko can proceed with application draft, it would be nice, if you we will discuss application before submitting