RC India

This is my entry for the ‘Game Changers 2015’ competition.

I’ve built ‘X’ and ‘+’ type quadcopters before. They fly great but don’t provide enough space at the center for mounting stuff.  Another big problem is that the motor vibrations are passed easily to the flight controller as the arm axes pass through the center.

‘H’ and ‘I’ type quadcopters eliminate those problems to a large extent but aren't as agile as the 'X' type. Twisted-X type frames are the best but are complex to construct. Hence, I decided to build an ‘I’ type frame.

The ArduCopter 3.2.1 firmware has an improved code for the 'H' type frame. The hardware configuration is as shown in the attached pic.

I'm posting the 3D models of the quad and its various parts first as it is difficult to show some details of the actual quad.

The frame is made using 13mm square aluminum pipes bound by 2mm acrylic plates.

The motor mount plates (brown colored ones) are made of MDF for absorbing motor vibrations.

The motor-centers lie at the vertices of a square of side-length 264mm. This length was chosen after considering aspects like total weight, maximum possible thrust and volume of the components.

The motors are mounted upside down.

This configuration gives a small (1% - 3%) improvement in total maximum thrust. Many people argue that it's not worth the effort as it is impossible to land the quad without destroying the propellers. 
However, this configuration gives remarkably significant reduction in vibrations. It becomes obvious once you consider the fact that the prop-wash no longer hits the arm under it.

I've done multiple tests using multiple motor-prop combinations using different test platforms. Each and every time the vibrations of an upside-down configuration were much lower than the usual configuration.

To further reduce vibrations, MDF plates are added between the motors and arms as well as between the motor-bolts and arms as shown in the attached picture. 

Now, the main reason I chose this frame type was because of the ample space at the center. Also, this helps in getting the perfect CG.

The ‘electronics stack’ (pic attached) was designed such that the centers of gravity of all the individual electronics components are aligned with the CG of the frame. The aluminum pipes have holes at their longitudinal centers so that the quad can be balanced on a custom made balancer after assembly.

The most notable part of this design are the slabs of foam tape (yellow colored ones in the 1st pic) that separate the whole electronics stack from the frame. This is a lot more efficient in reducing vibrations than just isolating the flight controller board. The reason this works is that the weight of the whole stack and the momentum it gains is much greater than that of the FC and hence it doesn’t follow the vibrations in the foam.

Another noteworthy aspect is that the CG of the flight controller lies at the CG of the quad. The APM can work even if it is placed some distance away from the CG, but it is recommended to place it as close as possible to the CG of the model.

The LiPo has been deliberately placed above the FC to counter the weight of the motors.
The second attached pic shows the various mount plates and other details.
 

I’ve always been a fan of shrouds. Mostly, they just increase the weight and decrease maneuverability. However, for this flying blender, a shroud is a necessity, particularly while landing.

The propellers are 8 inches in length and so the shroud doesn't look ginormous. I can't imagine building a shroud for propellerss of length 10 in. or more.

This shroud is 20mm thick and the width of the rings is 22mm.
 

Here’s the final render of the 3D model. It was made using Sketchup 15.

seems really interesting. i recently designed a 250 size shrouded X quad.  shrouds are excellent for close quarter flying as well, and prevent a whole load of damage to machine and man/woman alike. best of luck for you design.

Thanks Yashodhan!
I wish good foam was locally available here. Then I'd make shrouds for all my multirotors.

Use normal thermocol. And cover with paper and/or cellotape. Works really nice.

You can apply a thin coat of fevicol-water mixture(80-85% fevicol and 15-20% water) to provide strength!
Anyways Great design! :thumbsup:

Yes himadri that's exactly what I meant. Thanks for completing my sentence. :) you could also apply some epoxy in a thin layer all over the thermocol.. That would yield a hard shell. But would be brittle.

I've been using epxoy on my foamies for about 6 years now. Nothing else gives a better finish.

I've tried a variety of adhesives from pidilite and araldite. I've also tried various tapes like plastic, vinyl, abro-tape with acrylcoat etc.

But, like I said, epoxy works best. I've used it on HD-foam and depron only. Thermocol needs 2-3 layers of tissue paper soaked in fevicol for best results and even then it deforms on small impacts.

Last year a friend of mine from Germany was here in Nagpur for some pre-college thingy and taught me this method her dad uses. Before that, I just used to do a single epoxy coat.
 
This is how she put it:

1. Coat the foam with a fairly thick layer of over-night epoxy. Next morning use a 150 to 200 grit sandpaper (silicon carbide) to get a smooth finish. Do this step again.

2.  Mix in equal quantites 'over-night epoxy' and '5 min. epoxy' and apply a 3rd layer on the foam. After 6 to 8 hours, sand it again with a microgrit sandpaper.

   

And now the build…

Build step 1: Frame

I got the aluminum square pipes from hobbypep (pic. 1). They varied in length by a millimeter or two. But that wasn’t a problem as I used a laser cut acrylic template for drilling the motor-mount holes precisely.

First, the acrylic template was used to mark the motor mount holes. A precision bench drill press was then used to drill the holes. I started with a 0.8mm drill-bit and then slowly moved up to a 3mm bit to get perfectly drilled holes (pic. 2).

Then I glued together the aluminum pipes and the acrylic binding plates (pic. 3) using standard epoxy. All the parts were sanded using a 120 grit sandpaper prior to binding. The frame was clamped overnight using nut-bolts which is why there are so many holes in the frame (pic. 4).

Build step 2: Electronics shelves

The base mount plate for the electronics stack and the mount plates for the APM and GPS were deliberately cut out of 2.5mm MDF to absorb vibrations. They were made waterproof and sturdier by applying two coats of black epoxy. The GPS plate had to be redesigned 3 times before the NEO6M could fit snugly.

Pic. 1: Base Plate before and after coating.
Pic. 2: APM mount plate.
Pic. 3: LiPo Plate.
Pic. 4: GPS Plate.

Before assembling everything, here are some snaps of the electronics used.
As can be seen in the first picture, the screw holes of the APM cover were enlarged to accommodate M3 bolts.

The BLDCs and props were dynamically balanced (on a custom rig) using average RMS readings of vibrations. The ESC-to-motor wires were braided to reduce local EM radiation.

Pic. 1: APM 2.6 + Neo 6M.
Pic. 2: 1050Kv BLDCs with 8x4.5 N+C props.
Pic. 3: QBrain 4x25 ESC.

Build step 3: Mounting the APM

The APM board is mounted on the APM plate with strips of foam tape sandwiched between them. Then, the rubber grommets are passed through the APM plate holes.

Before passing the rubber grommets through the base plate holes, stand-offs are added above and below the base plate for mounting other electronics. Passing the grommets through the base plate needs patience and steady hands as there isn’t much room to do it.

Pic. 1: APM 2.6, rubber grommets, APM and base plates.
Pic. 2: APM and grommets attached to APM plate.
Pic. 3: Stand-offs attached to base plate.
Pic. 4: Final APM assembly.

Build step 4: Mounting the base + APM assembly on the frame.

As mentioned before, there are foam-tape pads separating the frame and the electronics stack. Note that the foam pads are on the under-side of the bottom frame plate, and so, the APM assembly needs to be inserted from below the frame.

Next, the ESC is mounted and the ESC, Rx, GPS and compass wires are attached to the APM. Finally, the LiPo and GPS are mounted as well.

Pic. 1: Foam tape on bottom frame plate.
Pic. 2: APM attached to frame.
Pic. 3: ESC and wiring.
Pic. 4: Complete electronics stack.

Build step 5: Mounting the BLDCs.

As shown previously in the 3D model, the BLDCs are completely isolated from the frame vibrations using MDF plates. The motor mounts were given a coat of clear epoxy for water-proofing.

One important thing to note here is that, even though the motors are mounted upside down, the propellers are mounted as if the motors were upright.

Pic. 1: MDF motor mounts with clear epoxy coat.
Pic. 2: BLDCs mounted on frame.

Build step 6: Shroud.

The shroud was made using a template of a quarter of the shroud. A sheet of HD foam was clamped between two such templates and cut using a hot-wire cutter.

The 4 mono-shrouds were then put together using epoxy and bound by 4mm coroplast.
The completed shroud was then coated with black epoxy using the previously described method.

Pic. 1: Shroud template and cut-out.
Pic. 2: Shroud with coro clamps.
Pic. 3: Shroud coated with epoxy.

The first test flight revealed some minor flaws. As can be seen in the last part of the video below, the shroud catches the downwash during landing making it difficult to land precisely.


https://youtu.be/zmLICeYOMVA

Clean and nice build Swapnil :thumbsup: and very nicely documented.

Whats the AUW of this model? and you are using 4Cell or 3Cell lipo?

Thanks Satya! Really appreciate your input!

The AUW without the shroud is 911 grams. And the shroud weighs 177 grams.

It's pretty snappy considering it's an 'H' type frame. All thanks to APM. With other FCs there's a noticable decrease in maneuvrability when going from 'X' type to 'H' type frame.

The ST-360 I bought from you weighed 905 grams including the weight of my custom landing gear. I just had to increase THR_MID and the APM was ready to go.

I'm using a 3S 2200mAH 20C Haiyin LiPo. But, due to the increase in weight, I can't do crazy maneuvers anymore otherwise the battery heats up. I'll soon get a 25-30C LiPo for the wild stuff. 

After the first test flight I tuned some ArduCopter parameters, tidied-up the hardware and added landing skids under the shroud.

Pic. 1: New cover.
Pic. 2: Tidied-up under-side.
Pic. 3: Landing skids.
Pic. 4: Complete iQuad.

The new THR_MID value of 490 gives a much better hover now (around 6s in the video). And, adding skids below the shroud has made landing much easier.


https://youtu.be/i5wEmdk9_Jk

Perfecto-Mondo.

Really very nice and smooth hover.

Thanks Satya!  :)

All the time spent in designing and re-designing the suspension and vibration dampening parts has really paid-off!

I'll upload the accelerometer graphs during 'Loiter mode' soon. You'll be nothing short of amazed!

Awesome ! :hatsoff: Great build sir

Thank you!  ;D

The final and most important test was flying in Loiter mode and acquiring the IMU data to see how effective the design is.

Even though it seems that the quad is swaying after activating loiter, it’s really the camera that’s moving. My camera-man was a bit shaken after the quad rocketed upwards. Anyways, check-out the video! 

https://youtu.be/d_aeX4G1FeU

Edited: I posted data from a previous build by mistake in this reply. Please check later posts for the correct data.

I'd like to thank the following people who make this hobby great!

1) Satya "satyagupta" (quadkopters.com): For replacing a month old GPS module without asking questions and for being a really nice guy!

2) Vishal "vishalrao" (kinetichobbies.com): For introducing me to the super-awesome Haiyin LiPos!

3) Santanu "santanucus": Fighting for our hobby and for sending me a free prop-adapter!

And finally, RCM for organizing this competition!

Wonderful Swapnil! All the best!

And thanks a ton for sharing the epoxy tip. Will try that now on my planes :)

Wondering why the second coat uses a mix of 5 min and overnight epoxy?

Thanks Vishal!

The '5min + overnight' epoxy thing simply reduces the setting time of the 'over-night' epoxy. It was also suggested by a friend who owns a hardware store nearby. He said a lot of carpenters and masons do that to get 'best of both worlds'. Araldite sells a special 5min 90g package for mixing with the over-night epoxy.

It's not necessary to do that step unless you are in a hurry. I simply coat my MDF robotic parts with one coat overnight epoxy and another coat of 5-min epoxy for getting a smooth finish.

That tip really is going to help me a lot. Overnight epoxy has way too long setting time and 5 min, too less. Wanted something in between. And this seems to be perfect! Thanks again :)

You are welcome Swapnil, really its been a very long time since that happened let it go now ;)

I think the magic was in the frame that was able to remove the vibration and stopping it to reach the frame.

Okay, this sounds a bit like preaching, but time doesn't change the fact that you did a nice thing.
I know many sellers have a replacement policy and would have shown the same gesture.

But, I mentioned it here for a reason. Back then I was ready to quit on quads and go back to planes because every other part of the quad kept failing. I was going to sell the whole set. But, after you sent me that GPS everything started working.

So, even if it doesn't seem like a big deal to you now, that was the point where I really got into quads. The GPS module doesn't play a big role in this quad, but I wouldn't have reached here without it.  :)

I am very happy that this gesture turned out to be such a fun filled experience for you. Glad to be a part of it :)

Thanks :thumbsup:

So as another test, I went ahead and mounted a raspberry pi camera on a Pi 2 with a long range WiFi adapter (pic attached).

Note that I haven’t used any ‘gimbal’ or stabilizing system for the camera. There's no jello or jitter or shakes of any kind!  ;D

Please do not watch the following video in full-screen. It’s a 640x480 vid and looks crappy when maximized.


https://www.youtube.com/watch?v=qZtyAupA9c4

The IMU graphs I attached with post# 29 were actually for a previous version of this build. As you can see in the attached picture, I had placed the main plate on a slab of sponge which wasn't very effective against low frequency vibrations.

The IMU graphs for the current build are even better!

The Acc X vibrations are within + and - 0.075G! It's slightly higher than that for the Accy Y as the motor arms are along the X axis.

Pic. 1: Full graph of Acc X during loiter.
Pic. 2: Zoomed in graph.

The Acc Y graph, as mentioned in th last post, is the best! The vibrations are within + and - 0.05G!
If you've used an APM before and have worked with dampening systems, you know how seriously cool that is!

Pic. 1: Full graph of Acc Y during loiter.
Pic. 2: Zoomed in graph.

And here's the AccZ graph. My previous 360 sized quad used to bob up and down. This one maintains rock-steady altitude!

Pic. 1: Full graph of Acc Z during loiter.
Pic. 2: Zoomed in graph.

Here's the holiday special for iQuad! Do watch it if you like Japanese music!

MERRY CHRISTMAS AND HAPPY SATURNALIA EVERYONE!  ;D


https://youtu.be/LUAK1RHqA-Y