8" Sheep herding quad build thread

[evranch]

Well, all the parts are on order and some have started to show up. So time to put together a build thread I guess.

Components:
- TBS Source One frame
- TBS Source One 7" arms
- HGLRC XJB F438-TX20 V2 Stack FC/ESC/VTX
- Foxeer Monster Micro Pro camera 16:9
- EMAX Stubby pagoda antenna
- FrSky XM+ RX
- DJI Mavic Pro motors, used
- DJI 8330 folding props
- 3s battery TBD as thrust and stability permits

Reasons for selection:
- TBS frame is tough and cheap.
- Camera: good colour and resolution for spotting sheep. 16:9 as I need to see a wide field rather than spot gates at high speed and angle.
- Motors and props: large props for efficiency, folding for pack size. DJI was the only way to achieve this. Used motors are easy to come by in good condition.
- 3s battery: the Mavic Pro ran a 3s battery so it's a no brainer. I already own a wide selection of 3s batteries so testing to find the max weight will be easy.


The motors and props arrived from Ebay. Pretty much as new with a couple scuffs. I wanted to see how they would run with a regular 30A ESC. The answer: Very well. There are no published specs at all, and once secured to the quad I want to gather a bit more data on them.

Remember to always wear goggles and heavy clothing while performing this sort of test. Also, this vise sucks and always comes unstuck from the bench.

 

[evranch]

The main issue on this build was - how to fit the nonstandard DJI motors onto the standard mounts of the Source One. The main issue: DJI motors have a 3-bolt pattern and everything else in the world has 4.

So I took some thin hobby plywood and made a little test mount with one of the spare 5" arms.

Clamped in a sturdier vise, I was able to bring it up to full power to check that the plywood would not flex. 110W per motor on 3s, and the plywood held up just fine. I figured, why not build 4 of these, check for prop clearances and maybe do a test hover...

Oops, I did something stupid. I tested with a 2200mAh which hovered nicely at low throttle with tons of excess lift. Then I decided, instead of adding a little more weight (or replacing the adapter plates as planned) I would see if I could lift the heaviest battery I had (7700). Worst case, it just doesn't fly, right? Wrong. It got off the ground, but then the plywood did not flex but failed catastrophically. Wrecked one prop and damaged the leads on a motor, which I repaired with no issues.

Stepping up to aircraft aluminum mounts. Did I say aircraft? I meant shovel. This is a farm. I found an old aluminum grain shovel that I had run over with the truck, and it was nice and flat. The four holes are off center... Yes, but they fit easier against the slots this way. A perfect square only fits in one perfect location, and I didn't feel like doing precision machining today. I just marked them with a pencil and drilled them where they would line up, then made 4 identical ones. All fit nicely with no adjusting.

All four motors mounted and waiting for more props (Originally bought the props just to see if they would fit... that's why I only had four). I carefully spun them up in pairs though, and the high/low mounting on the thick arms appears sufficient to avoid collisions between the prop disks. The 7" Source One arms are about twice as thick as the 5" arms.

Technically the props would be OK in the same plane after the arms got extended by these adapter plates. However, they only clear by about a millimeter... so in case anything flexes, I decided to put them up and down as originally planned.

 

[evranch]

Now that I know it flies, time to do some soldering. M8N GPS and Foxeer Monster cam. I didn't realize the FC did not include a barometer, so I will have to order one and add it on. I configured UART3 for I2C and hooked the compass to it, works fine so that's where the baro will go as well.

This means I've had to give up control of the VTX from the FC, but that doesn't really matter. I'll be flying at max power out here in the hills on a channel all to myself.

Great picture on the cam, terrible VTX that came with the stack. Video is ripply and frequencies are way off, I'm going to try to warranty it. Hopefully they will not insist I return the entire stack, as the rest of it is kind of soldered to the quad...

 

[evranch]

Maiden videos. I did a hover test with a standard 2200 3s battery and made it to 13 minutes before I quit. AUW 550g.

My hovering skills definitely improved towards the end of the video.

At one time I kicked in HEADING HOLD to test the compass, which resulted in the quad spinning on the spot. Not sure what went wrong.


Then I decided to try an FPV flight despite the awful VTX. I pointed the camera towards the sun which grayed everything out. I thought it was drifting off channel again and decided to panic land. Can't wait for that new VTX to show up!

Finally, let's go back to lifting big batteries. I was scared off by the 7700, so why not try this 6400. AUW 950g which was my target weight for the project:

Really happy with this test! It seemed more stable with that extra weight underneath it, and didn't require much higher of a throttle setting. I think the drift is all me as I've only flown sim before. I want to test this outdoors for climb performance as well as try out those 7700s just to see how they perform!

 

[IcedStorm777]

Cool!!

 

[evranch]

My parts arrived and it was snowing, so I built the quad, thread updated! I had to do some unusual mounting to get the DJI motors attached, but it does fly!

 

[FDS]

Awesome agricultural engineering! My scrap bin has saved me on plenty of occasions.

 

[PsyBorg]

This has flown already with them extended motor mounts? I would think as thin as they are they would create insane noise harmonics and drive the fc bonkers.

Be interesting to see video with both onboard and external sound examples.

 

[evranch]

This has flown already with them extended motor mounts?

It was tested with the plywood ones which are considerably thinner and more flexible than the aluminum. Just waiting on props to test the aluminum mounts, and then I will probably do a video if the weather is good enough to fly outside. Also waiting on a new VTX as GetFPV has agreed to warranty this one, then will post DVR video.

It was just a short indoor hover test due to a blizzard outside (yep, love this country). Used an arm toggle switch in case things went crazy but it was very well behaved. It hovered very cleanly with minimal drift and no "toilet bowl", I commanded it in the four directions and spun it on the spot. Then I loaded it up with too much battery and broke the mounts.

I did tune it conservatively, using the default settings for a general purpose 10" and the FC is soft mounted on O-rings. Also, maybe the flex in those long arms is enough to take out some vibration?

It's very quiet for a quad of this size. The DJI motors are nearly silent without the props and the folding props seem to cut the air very smoothly.

 

[ElectriSean]

Glad it flew for you, I have the same concerns as Bill about those mounts. Are you running iNav?

 

[evranch]

Yep, iNav seems very simple to set up, calibrate and be ready to fly.

I was originally coming up with a more complex plan involving rigid stacks above the mounting pads on the arms, but then I thought... for a first test, why not just try a simple plate? And it worked, with a lot less engineering. An issue with the DJI motors is they come with very short screws of some odd size, the original quad must have had them mounted to a very thin plate as well.

I hope the aluminum works as well as the plywood and does not induce more vibration, since it's definitely stiffer. The plywood may have been acting a bit as a soft mount, absorbing vibrations.

An interesting note is that when clamped in the vise for the full throttle test, the entire arm flexed slightly but the plywood did not bend at all. That stuff is pretty tough.

 

[evranch]

Alright update time with successful flights! I got props in the city at Staples (the last of them, they aren't carrying DJI anymore) and got it into the air. Got a bag of aftermarket props on the way ordered from eBay too, for breaking while flying aggressively. See edited post for videos which may be still uploading.

Now I'm looking for an ultra-loud piezo that I can trigger with the momentary switch on my transmitter to irritate the sheep.

Question for iNav gurus - why can't I find NAV_ALTHOLD on my list of modes? NAV_POSHOLD is there as well as NAV_RTH, but NAV_POSHOLD isn't much good without ALTHOLD? Too scared to enable it.

Is it due to my lack of a barometer? I did some reading and supposedly the GPS can be used to hold altitude all by itself - presumably fairly far from the ground, I would think.

Also, how to check compass calibration? It's hard to tell if the compass is actually working as there doesn't seem to be raw outputs for it in the configurator as there are for the gyros and accelerators. It claims it is healthy and set some calibration values, but HEADING HOLD totally failed in the test.

Edit: Here is a quick DVR video from an indoor hover with the 6400

 

[evranch]

Flies great with the 7700 and I'll be making that target 30m flight time with no issues. I called off the test flight today at 26 minutes as the gauge ticked down to 10.7v to see what an actual meter read. At 11.1v resting (3.7/cell), there is still quite a bit left!

I also went and bothered the sheep in the corrals to see what they thought. They were curious at first, but after I blew a little propwash on one they decided it was worth paying attention to. I'm happy with the initial results:

Also, this being my first real FPV quad flight, 25 minutes is a long time to fly. Hands were a little shaky I'll admit...

 

[ElectriSean]

Nice

 

[evranch]

I've been too busy to post or even read the forums, but I have been working on my quad skills. Today I finally did a real sheep roundup with the quadcopter with great effectiveness!

Unfortunately, the bulk of the footage was lost due to a DVR mishap where I almost lost the quad as well! I launched without pressing the record button. I decided to press it in the air, but instead somehow got it into a mode where it was playing back old DVR footage... while I was in the air in ACRO... uh oh.

After a crash course in flying ACRO LOS with one hand while fussing with the goggles with the other, I decided to power cycle the goggles as the quad drifted out over the pond. They came back in the proper mode and there was no way I was pressing that button again in the air. I proceeded to search the pasture and gather sheep and it worked great. All sheep were gathered from the home quarter and put into the corrals within 12 minutes of flight time.

All I can say is that if I was flying anything other than Vipers sans shroud it would have been the end of my quad. I will never touch my goggles in the air again.

At the end of the day though, 3 sheep made a wrong turn and missed the gate. This time I activated the DVR on the ground, and I got this footage to share of bringing them back! Being able to bring these sheep back in only a couple minutes is amazing!

Note how the llamas are not intimidated in the slightest. I decided to land and shoo them in with my arms. I have yet to add the loud siren I've planned on, and could have used it a couple times to dislodge a couple sheep that realized they could hide under the shell of an old combine to stay safe from this unusual bird.

 

[evranch]

Flight notes:

- Flying in ACRO greatly reduced the issues with the VRS/buffeting, but it's still present. I bounced the quad off the grass once because of it while trying to descend in close quarters beside the scrap combine. As such, I've decided to try to avoid flying above the water unless I'm transporting at high speed and altitude.

- you may have noticed the 0 degree camera angle. It doesn't work for racing, but it works great for this task as it lets me look down towards the sheep while flying forwards. I also find I have a very hard time hovering on the spot with even a 20 degree angle - I always end up with forward motion. As you can see in even the short clip posted, being able to hover and inch forwards is essential.

- Travelling at high speeds with a 0 degree camera, you obviously can't see in front of you. I tend to fly high for rapid transport and then bring it down.

- The Foxeer Monster cam is absolutely beautiful and I can pick out sheep against anything from high in the air. For some reason, the captures end up in 4:3, but the 16:9 in the goggles is just great. I spotted a sheep who had tripped and gotten her leg caught in a hole, and was able to go out and rescue her! Worth it right there!

- I still don't trust myself to land in FPV, and always fly over my head, spot the quad LOS, and bring it down. I still find LOS ACRO very challenging to fly, but I've at least gained enough skill to land without switching to ANGLE.

 

[servantelectron]

Hi

I just found this site/thread. Since January I have been designing/building a drone specifically for sheep herding. My son has a 140 acre farm in central Virginia with ~250 sheep, ~600 chickens, and ~50 goats. He has 5 Great Pyrenees for guarding the sheep, but no herding dog. Moving the sheep from one pasture to another is a serious problem, thus the drone effort. I am a 70+ retired electronics designer living with my wife of 49 years on a 50' sailboat, so building a drone in this small space presents some challenges. My first shot (flown 6 times so far - crashed 3) is a Tarot Ironman 650 quad:
Pixhawk 2.1 cube with Here 2 GPS
Multistar Elite 3508- 268KV motors
17 x 5 CF props
Bardwell 4in1 ESC
Frsky 8XR RC
Maxbotic Sonar altimeter
Lidar
3 cameras on a video switch Wide Angle Down, Narrow Tilted, Wide Forward

The AUW with a 6s 3000mAhr LiPo is ~1200g, and it hovers at 50% PWM with ~ 10A total battery current.
I am constructing 6s Li-ion battery packs , the 6s4p is done and weighs in at 877g, and I am beginning construction of a 6s3p.

I am gathering parts for the next versions of the craft.

I have as goals:
The mission is sheep herding, not flying, it must be easy to fly.
Precise altitude terrain following and geo-fencing is a must.
Long duration flight is essential - one hour is the target.

My next step is to read all of your posts.

Phil

 

[servantelectron]

I have been focused on the landing gear for the sheep herder. There were problems with the landing gear supplied with the Tarot Ironman650. It is heavy, not very tall, tippy, and it has no shock absorption. I came up with the solution pictured above. The length of the CF sticks can be tailored for the expected max grass height so the craft doesn't have to mow its way to an unplanned landing. The sticks are held in the polypropylene Ts with a 2cm segment of 5/8" x3/8" latex tubing. This gives very effective shock absorption and enough rigidity for a solid landing. The whole landing gear set with 6 feet (2 amidship) weighs 45g, ~ half the weight of the supplied manual folding Tarot rig. The legs are firmly held, but they can be easily removed if necessary.

These are the T fittings that I modified to fit the 16mm arms:
https://www.ebay.com/itm/10-Pack-3-4-Barbed-Plastic-Tee-Hose-Fittings-Polypropylene-150-PSI-NEW/233185427145

I used a heat gun to soften the Ts so the barb ends could easily be cut off with a packing knife.

I used this reamer to match the ID of the thru part of the T to the 16mm arm OD.
https://www.ebay.com/itm/uxcell-Straight-Shank-H8-Chucking-Reamer-16mm-Cutting-Dia-8-Flutes-Hand-Milling/323804254663

This is the CF tube I used for the legs:
https://www.ebay.com/itm/4-pcs-x-Glossy-10mm-OD-9mm-ID-500mm-3K-Roll-Carbon-Fiber-Tube-10-9-500/283105539836

I cut 2cm lengths of this tubing for the leg mount/shock absorber. The latex is slipped over the end of the CF tube with about a 3mm overhang, then with a drop of water it is pressed into the T with a slight twisting action.

https://www.ebay.com/itm/4-FEET-3-8-I-D-x-1-8-wall-x-5-8-O-D-Surgical-Latex-Rubber-Tubing/191923640851

I used a hole saw to cut the landing feet from a block of closed cell packing foam.

It is easier to slip the T onto the 16mm arms if it is heated first. A small spot of hot glue keeps it aligned.

I will report how this performs in landings and crashes soon.

Phil

 

[servantelectron]

The new landing gear works extraordinarily well! It has great give and shock absorption in crashes. In a normal landing the forces are distributed over a much larger base than the landing gear supplied by Tarot, and it is much lighter. One plesant surprise is that it is easily removed for storage or service, and it is just as easily reinstalled.

 

[servantelectron]

Here is a photo of the attachment end of a landing leg.