Starting a new thread to show what I'm working on.  I'm building a new Octocopter of my own design.  The goal is to use it for Aerial Photography, lifting a mid-weight camera such as a Sony Nex5.  It will of course be using Arducopter for control.  ;)

I started off researching the various kits on the market, and starting getting analysis paralysis.  I looked at all the options such as Droidworx AD8, Cinestar, CarbonCore, and SteadiDrone.  But I just wasn't happy with the design on them, particularly for the price they ask.  Combined with the fact that I can be pretty frugal, and have a strong DIY ethic, I decided to just design my own.

The basis of the design relies on many of the Hobby King quadcopter parts.  I liked the design of the motor mounts, they are very professional looking, even compared to some of the high end machines.  And I liked that the boom mounts are blocks of aluminum instead of plastic.  Very rigid.  I also liked the look of the dome on their H.A.L. quad, but since they don't sell it seperately, I bought the whole kit, it's only $34, amazing!  I might end up using the rest of it someday, who knows?  So, I basically emptied out their stock of Talon parts, so if you need to fix your Talon quad and the parts are backordered... sorry!

Once that was settled, I set about designing the center frame.  My design required a few things, namely I want all the wiring hidden.  I don't want any spaghetti showing.  Particularly with the ESC's, while still allowing cooling airflow.  This required the center hub to be larger than is typical.  I then needed a smaller subframe to house the avionics.

You can see these plates below.  The larger one houses the HAL dome. You will notice the 3 blue anti-vibration grommets in the middle.  This is the APM1 pattern.  I did that just because.  Maybe I'll use an APM1 as a stand along gimbal controller.  The second plate had grommet mounting for an APM2.  The final plate is the top plate, and has bolt patterns for the Ublox GPS and the magnetometer.

This next photos shows the avionics frame built up with an APM2 mounted.  25mm aluminum standoffs are used. I don't like the plastic standoffs typically used.  They get loose, and lead to vibration.  They also break easily.  This structure ends up being quite rigid.

Here is the avionics frame mounted on top of the main frame.  Yes, the main frame is HUGE.  I actually don't see the point in having a tiny center frame, and then long arms.  This seems like it just leads to flex, and doesn't leave you with any real estate to mount your avionics.  

This photo shows one of my design features.  I put some cutouts on the bottom frame for weight savings, and then made matching cutouts on the top frames.  This creates some nice conduits to run wires neatly.

Here it is with the HAL dome mounted and one of the Talon arm mounts bolted in.  I'm really happy with how stiff the assembly is already.

Here it is with one of the arms mounted.  These are the short 220mm arms, I also got some of the 320mm arms.  One may wonder how I'm going to get away with such short arms and 13" props....  Astute readers will know what I'm planning already. ;)

This last photo shows one of the ESC's in position. This is a Hobby King F-40A, it's somewhat larger than typical 20A units which is partially why the frame is so big.

That's it for now.  I had the plates cut out of aluminum to start with because it's cost effective.  If it all works out, I'll probably have it redone in carbon fiber plate for the weight savings.  I was a little uncertain about the weight, but it looks like the frame will come in at 1070g all in aluminum.  That's about 300g heavier than a Droidworx AD8, but not too bad. It will drop about 1-200g if redone in CF.

Views: 81420

Reply to This

Replies to This Discussion

Nice and sturdy! keep it going Robert, I am subscribed.

I set up to do some thrust testing today.  It's taken more time than I thought as I had to build a test stand first.  But that's done and I'm glad I did as it shows some interesting results.


Here's what it looks like set up for single motor and coaxial testing.



First, the eCalc didn't work out.  The motor/prop combo I chose doesn't work out as it predicts.  It's over-propped and I can't safely go above 1800 PWM on the test stand.  Above that and the Watts and Amps go too high and I risk burning something out.  So I limited testing to 1800 PWM.  I'm going to have to try and exchange my 13" props for 12".  :(


What's interesting is if you look at the results of the coaxial testing, it reinforces the common statement about this setup being less efficient.  Don't just look at the thrust vs. PWM graphs, but the specific thrust graphs as well.

Then I tried props overlapping.  This is something Brad Hughey has been a proponent of.  There is some research that shows it is actually *more* efficient than motors fully spread apart.  Here is the test rig:


I only tested one D/L so far.  330mm props on 280mm spacing.  Check out the data in the spreadsheet.  (I doubled the numbers from the single motor test for scaling.

The theoretical testing shows the "sweet spot" for the D/L efficiency bonus is at 20-30% overlap, and I only had 18%.  It should also be noted that my test rig does not really do the best job simulating the test, since there is a huge square pipe right in the overlap zone.
Regardless, I'm sold.  I'll be building the Octo with overlapping props.  At the least, it will be more compact than a normal setup, while not losing much if any efficiency.  At best, it will be more efficient than a normal Octo.

Oh man, one of my ESC's started outputting 12V out of the BEC (WTF?) and it seems to have fried my nice digital servo tester.  :(

How is that possible?

I did another round of testing.  I used a longer bar to spread the motors fully apart so D/L was 0.83.  I ran the test again, and there doesn't seem to be an advantage or disadvantage to the overlapping props.  Of course, overlapping them means I have to mount one motor up, the other down, which puts them closer to getting in the view of the camera...  So now I'm wondering if it's worth it, with no efficiency advantage.

I wonder if you dont get "false" readings with your setup. you are not far from the wall and shelves... some kind of ground effect could be making your readings wrong?  

You will have a margin of error on your setup, as long as you can log many calculation and be almost all the time the same you can really predict how it will affect the efficiency.  The "only" concern I have with this setup is only because I dont know how it work when the auto pilot is changing direction.. will the 2 motors still give the same efficiency when one of them goes slower?  for example on the overlap now you are comparing them both at same speed, but if one needs to slow down 50% will it really be  50% and not actually influencing it to get 55%? then the result will not be the same... 

Anyhow, Build it and fly it! that is the only way go see!  :) 

@Robert,   Your new octocopter is looking really good. And thanks for the tips about aluminum vs. plastic spacers.

I am envious of your work space. Especially your test space. The photo below is of my indoor test space. That is, if I hang heavy pads on the walls and ceiling, and put a heavy board on the floor (with quadcopter tie-down rings on the corners). We don't use this space very much (we have a similar space upstairs). Needless to say, my plan will be easier to execute if my wife is not in the house during the test periods. LOL


I'm kinda of thinking now that I'm just going to build it conventionally with long arms.  There's no efficiency advantage, and a pretty big disadvantagewhen I want to sling a camera under it and keep the props out.

There's also a stability advantage with the wider set motors.

It's too bad the real world testing doesn't seem to prove out the theory.  Maybe there's a scale factor that comes into effect in larger sizes?

That's a nice set-up.  My much more amateuristic set-up bounces around quite a bit even when i set to an unmoving pwm value.  I'll bet yours doesn't do that much.

Could you add the pwm axis to your graph as well?

Yeah, the force reading was fairly steady, except at 1700 and 1800 PWM, I found the power reading was dropping.  So it was a bit of a judgement call to decide when to "call it", write down the numbers, and drop the power.  I think the motors were heating up and the increase in resistance dropped the amps and thus watts?  They got alarmingly hot which is why I didn't go past 1800.

I should get a temp gun.

I tried to add the PWM values on the Y axis, but couldn't figure out how.  Any idea?  The test starts at 1100 and ends at 1800, you can see many of the points. 

When I ran the second "independent motor test" but using two motors instead of just 1, but spaced apart by ~14.5" (13" props), I found the thrust/watts almost exactly matched the overlapping prop numbers.  I also re-ran the overlapping prop test, with the backwards motor flipped around facing forward, but using 30mm spacers between the motor and mount.  This offset the blade disks exactly 30mm, and got the square tube out of the zone.  The numbers again, were very close to the old test, and the dual-spread motor test.

I think the test when I ran the single motor alone might be a bit of a throw-away.  I think that the battery supply, voltage drop through the wires, across the power meter, etc... I think it put in an inaccuracy.  Because the results of the dual-spread motors didn't match.

Basically, I can see not much difference one way or the other. The only clear loser was the Co-axial setup.

 Robert that is quite a project you have going! It has been my experience as the voltage drops from usage the amperage goes up more ,resistance , more heat   voltage lowers .(vicious circle ) eventually something fry's.  Unless you are experienced working with  composites , I would stick with the aluminum you could add lots of lightening holes  without jeprodizing strength  and shave any access weight off

Looks Great! I like your load tester  Have a great day!

Actually I recharged the battery between runs to try and ensure that I had as accurate of an A-B comparison as possible.  And the battery, I don't think it got below 80% at any point.

As far as composites are concerned, I would just send the files out to be cut by a professional.  It's not a big deal.  I wouldn't have to do anything, just as with this.  These were simply waterjet cut.

Robert, you are way ahead of my next build and have noticed the same issues with wiring congestion. Why the small center section when just slightly larger will help several issues? I also have a strong DIY slant (1st quad frame was all scratch built). To that end here is a teaser of where I am going...


The larger base/top plates will allow:

  • ESCs to be mounted on more real estate (not scrunched with wiring on the arms)
  • Integrate my arming plug strategy closer to the PDB
  • Use less aluminum tubing in the arms (0.75 sq 1/16 wall is 10 times heavier per area)
  • Still use the current 3DR stackup arrangement.

The material will still be 1/8 aircraft plywood. The stuff is about 10% lighter than the FR4 material. Another goal is to allow someone to purchase most of the raw frame material at a DIY store, hobby shop, or craft shop. A dual battery box under the base plate ads rigidity and protection for the $$ LiPos.

By adding a new work bed to my home brew CNC, I think I can just cut these plates. If not, that is what other woodworking tools are for.

Oh, before anyone comments, I have three CAD programs and still keep going back to paper and pencil. I know, Caveman methods..but it works for me. G-code for the CNC machine...I still do that manually too. I'd rather spend the $$ on flying stuff than some software that will go obsolete in a few years.

Just to be clear, the whole idea behind overlapping rotor disks was to increase the effective disk area in a more compact footprint.  Dr. Leishman's efficiency "bucket" prediction is only ~5% of unity, so I would expect to find no difference at 20% overlap unless the lab conditions were very tightly controlled (which is what he said in the paper).  It may very well be that a grid layout is best for this approach versus a classic multicopter "star" pattern.

However, the efficiency cost of a coaxial configuration is easily discernible in testing, at a range of 20-40%.  To be fair, the downstream rotor should have a higher pitch (which you can calculate via the induced velocity of the air stream of the top prop) for it to work the best.  That said, the whole disk-loading vs. thrust-spacing vs. control-response vs. center-of-mass conversation is a whole other kettle of fish.

The efficiency of motors peaks at about 1/3rd of rated power and falls off rapidly after that.  A unit rated at 85% peak efficiency might only be about 60% at full power (remembering too that our friends in Asia tend to play fast and loose with rating numbers).  In other words, a 300 watt motor is trying to scrub off 120 watts of waste, yeah, it will get hot.

Anyway, nice work Robert.  I'm sure the finished product will exceed expectations.

Reply to Discussion


© 2020   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service