Considering repowering from pusher to twin. Efficiency loss?

[evranch]

I do aerial mapping on my farm with a large, heavy, underpowered plane, the Bormatec Maja. With the change in regulations this year I plan to branch out and do some contract work. Trouble is, the Maja is a great flyer that's hard to get into the air. With a takeoff weight of 3300-3900g (7.2-8.5lbs) and a 30A motor and ESC, my takeoff roll is realistic to say the least. A couple hundred feet without wind. I yardsaled it last week when I clipped a hard snowdrift and was bounced into the air prematurely. Not enough airflow over the control surfaces = lots of pieces to pick up.

It can be hand launched with a hefty two hand toss into the wind, but with the pusher motor I am relying on my FC to sense the toss and spin up the prop after it's away - and the time it takes to spin up means every launch barely misses the ground (or sometimes doesn't, if the wind isn't strong enough...)

Flaps are counterproductive as the added drag cancels out the improved stall speed, and I need even more runway.

Current power package:
- 30A ESC
- Pichler Boost 25: 980kV 30A max
- APC 12x8E

With this setup I can cruise at 8A on a calm day, which gets me the better part of an hour in the air with a 7700mAh. I used to swing an 11x5.5E which could achieve the same cruise at 5A due to the better pitch speed, but takeoffs were even more marginal.

I'm seriously considering ditching the pusher for a pair of power pods on the wings, swinging the same 12x8E or the more efficient 11x5.5E.
- double the static thrust
- better weight balance, more options for camera and battery placement
- differential thrust for better yaw control while mapping

But everyone knows that in the real world, twins are fuel hogs. Does this carry over to RC scale?

Other options?
- More powerful pusher motor
- Keep the pusher motor for cruise, and only use wing motors for takeoff?
- Add a tractor motor and folding prop for takeoff, which will give me more thrust and more airflow
- Quadplane VTOL/STOL? Too heavy to be practical, I think.

 

[Merv]

Other options?

There are many bungee launch systems that may be usefull

 

[evranch]

I tried building a bungee launcher in the past, but had some issues. A bungee off the ground required a very sturdy anchor, was about 75' long when stretched, and still could barely get the plane above its stall speed. I tried to make one that launched the plane off rails, with some pulleys, cable, and a hefty spring from the junk pile, but it had too much "snap" rather than push - it was hard on the airframe, and the attitude of the plane was fairly unstable as it came off the rails. Speed was still lacking, and even on a successful launch the plane would be wallowing badly until it picked up speed.

Stall speed is around 30kph but from reviewing my logs, Vr is closer to 50kph for a reliable takeoff. Otherwise, it's too easy to pull a little too much elevator and bring on a power-on stall.

I'll look into whether catapults have changed in the last couple years. The pneumatics looked the best, but would be expensive to buy or somewhat challenging to build.

Ultimately reliable hand launches is my goal, as I want to be able to show up on site and get into the air without a lot of setup.

 

[quorneng]

evranch
For the same total power a twin is almost bound to be less efficient than a single simply because you have the losses of two motors.

My suggestion would be to use a more powerful pusher motor and simply throttle back to your current power level after take off. This would also allow you to safely use an efficient 'cruise' prop for a long duration but still achieve a reasonable take off performance.
Brushless motors do have a reducing electrical efficiency as you throttle back but the prop efficiency usually rises by a bigger margin.

My "endurance" pusher has a minimum cruise to full power amps ratio of just over 1:10 due to the fact that its big slow revving prop is set for cruise conditions with the result the motor is significantly over loaded at full power but then full power is rarely used for more than a second or two.

 

[clolsonus]

I'm happy to answer questions about my mapping setup if you are interested. I'm pretty pleased with how it performs. I have been flying an X-UAV Talon (with the wing extensions.) With a Sony A6000 camera and an 8000 mah battery it has an all up weight of about 7 lbs. I've flown a 77 minute flight without trying too hard. I've captured 2450 (24megapixel) images on a single flight which is almost too much to throw at my computer all at once! I typically cruise at 25 kts and estimate 6-7 amp current draw (although on my 77 minute flight I estimated more like 5.5 amp draw.) Hand launching has been pretty easy for me. I've run my launch video in slow motion and the clearances look ok. In the summer I'll wear a soccer shin guard on my forearm and a leather glove. That said, it basically flies right out of my hand which is good because I've never been much at throwing type sports. I'm not the most experienced and I'm not the least ... so in my experience I'll take a solid flyer that is easy to launch and land any day, even if I have to give up a few percent on the numbers printed in the brochure. The stress reduction is worth every bit I think!

 

[evranch]

Now that's a nice takeoff, @clolsonus! You are around the same weight and cruise speed as me, so what motor/prop/ESC are you running? Your toss is much gentler than my javelin pitch. Do you pull any flap on launch? It doesn't look like it.

I bought the Maja on consignment and have always ran what it came with, but when I did my research I found out that Bormatec doesn't supply the power package... the previous owner wimped out obviously.

Off topic: Cool to see another guy doing mapping here! I'm still using an old Canon point and shoot, only 4mp. Mostly because it fits in the Maja bay very well and has very little shutter lag with CHDK, compared to any camera I've considered replacing it with that would be economical enough to wreck up in one of my many takeoff-related crashes. Also, at legal altitudes I am getting plenty of ground resolution with it for my purposes, so more megapixels is just more work for my computer. Of course, higher altitudes and more megapixels would let us cover a LOT more ground in the same time...

You must have a pretty good computer setup to stitch 24mp x 2450! I'm running Micmac for stitching, 12GB ram on an i5-3570. Machine is about 5 years old. I can do a quarter section to orthophoto/DEM in an hour or two on CPU, but Micmac now supports CUDA/OpenCL so I'm going to see what happens when I bring my GPU into play. Planning to build a new computer soon for Micmac and XPlane, probably VR ready for when XPlane 12 comes out, which will give me plenty of power if it runs well on GPU.

 

[clolsonus]

Now that's a nice takeoff, @clolsonus! You are around the same weight and cruise speed as me, so what motor/prop/ESC are you running? Your toss is much gentler than my javelin pitch. Do you pull any flap on launch? It doesn't look like it.

I bought the Maja on consignment and have always ran what it came with, but when I did my research I found out that Bormatec doesn't supply the power package... the previous owner wimped out obviously.

Off topic: Cool to see another guy doing mapping here! I'm still using an old Canon point and shoot, only 4mp. Mostly because it fits in the Maja bay very well and has very little shutter lag with CHDK, compared to any camera I've considered replacing it with that would be economical enough to wreck up in one of my many takeoff-related crashes. Also, at legal altitudes I am getting plenty of ground resolution with it for my purposes, so more megapixels is just more work for my computer. Of course, higher altitudes and more megapixels would let us cover a LOT more ground in the same time...

You must have a pretty good computer setup to stitch 24mp x 2450! I'm running Micmac for stitching, 12GB ram on an i5-3570. Machine is about 5 years old. I can do a quarter section to orthophoto/DEM in an hour or two on CPU, but Micmac now supports CUDA/OpenCL so I'm going to see what happens when I bring my GPU into play. Planning to build a new computer soon for Micmac and XPlane, probably VR ready for when XPlane 12 comes out, which will give me plenty of power if it runs well on GPU.

Hi Evranch, we are flying the talon with an E-flite Power 25 (you'd have to look up the specs on that motor) with a 12 x 5 folding prop (by aeronaut?) and a 50amp castle creations speed controller.

No flaps on launch. The autopilot is an in-house developed design at the U of MN UAV lab. (not any of the popular diy boards or software stacks.) For auto-launch it starts ramping up the motor and when it feel like it really wants to go, I give it a light toss and that's all there is to it.

In my experience for mapping, there are tons of tradeoffs for lens, megapixels, altitude, speed, etc., and then there are tons of different use cases for the maps. For us we are looking for details (trying to spot the colored berries on an invasive vine) so we need all the detail we can get. The Sony A6000 has a really nice sensor, so beyond all the other specs and flight choices, when you zoom way in you get a little bit more useful data per pixel. So for our specific project I have a hard time imagining a system that could do much better, but for any one else's project the balance might tip a totally different direction.

I've never heard of micmac, but like everything else, if you've found a way to make it fit what you need then life is pretty good! Because we are using the images to hunt for a needle in a haystack, I actually came up with some of our own software to fit the images together seamlessly and a visualizer that can draw them stretched/warped ... again for the purpose of finding needles in haystacks. If you are looking for DEM or crop health or construction site monitoring, that's a totally different thing and the commercial tools can really work great for that. Here's a long rambling video to go along with my long rambling reply!

 

[evranch]

There are a couple different kV versions of that motor available, so if you wouldn't mind taking a look at the nameplate some time I would really appreciate it. Either way, the 870kV version can deliver 600W to my 300W, so you are capable of running double the power. I would suspect at 12x5 you are higher kV than me, since my motor has no trouble swinging 12x8. Or you are running more cells, are you 3s or 4s?

Neat project and one that is interesting to me as my operation is sheep on native prairie. I'm conservation minded myself and controlling invasives is a constant struggle. One of the things I do with my UAV is monitor the growth and control of patches of Canada thistle, the main weed that sheep don't like to eat. Thistle is a lot easier to spot though, it's a big green patch in late summer.

Very different usage case from me as my primary goal is DEM. I live in the hills. I'm looking for routes to run fence that follow easy contours, building sites that require minimal effort (cut into this hilltop, push the dirt there, flat site for new barn. How many yards would I have to move?) and overland flow information that can help me improve drainage and infiltration of the spring runoff. I'm also looking at slope data to see where I could cut in small swales or terraces to improve infiltration.

My latest project was some moisture farming where I set up a couple trial sections of snow fence over the winter and mapped size and location of the created drifts right before the melt. I want to try to catch the snow further up on the dry hilltops to improve moisture distribution and grass yields.

Micmac is a FOSS package developed by a couple French institutions. Challenging learning curve but very powerful software, it produces good orthophotos, DEMs and point clouds. It has a couple different pipelines and a million ways to tweak them to suit your dataset. Definitely worth a look if you have never seen it. http://micmac.ensg.eu

 

[Merv]

My latest project was some moisture farming where I set up a couple trial sections of snow fence over the winter and mapped size and location of the created drifts right before the melt. I want to try to catch the snow further up on the dry hilltops to improve moisture distribution and grass yields.[/URL]

Interesting project. Are you able to measure the volume of the snow drift?

 

[clolsonus]

Hi evranch, I'm out of the office today, but just got confirmation back that we are flying the 870kv version of the power 25 with our Talon. Combined with the 12x5 prop seems to be a nice combination for the airplane. Plenty of climb, and we can push up the speed quite a ways above our lazy 25 kts if we don't mind using our battery a lot faster. I've done that on some super windy days when the plane can barely make headway at 25 kts. The folding prop is nice on touch down to minimize the chance of breaking a blade or bending a prop shaft (which we've done with a non-folding prop.) It's easy to catch a blade in long grass even if you are protected from striking the hard ground.

 

[evranch]

Thanks for your help! I have always been bad at motor/prop combinations for some reason. As such I like to install something that has been proven to work.

It's interesting that your much more powerful motor is swinging less pitch, and less kV, yet you obviously have much more thrust as your airplane leaps into the air. It makes me think my motor may be overloaded and not delivering its full potential - yet it is drawing its rated 30A. Either that or you are running a 4s battery.

Maybe I need to build a test stand or balance the plane on a scale and start testing static thrust with some different props.

@Merv and anyone else interested, check out these images with 5cm contours:

https://imgur.com/PtCMiaX

The snow fence on the left was on a high hilltop and was pretty much destroyed by the wind. It managed to catch < 5cm downwind and "a skiff" upwind. Yet as you see, the rest of the hilltop is completely bare. Any snow cover at all protects the soil from drying over the winter and gives a much needed boost in the spring. I think this is worth trying again with wooden pallets rather than flimsy plastic.

The snow fence on the right remained intact all winter and caught a pair of drifts according to the two prevailing winds that blew over it. As it's on a hillside, the next step is to capture the hillside in the spring before any grass grows, subtract the two elevation models, and calculate the actual depth and volume of the drift. It appears to be about 30-50cm deep at a glance.

 

[clolsonus]

Hi evranch, just confirming we are running a 4S (8000 mah) battery on our talon. I've been stitching some snowy data sets lately and it's interesting to follow deer trails (or snow machine trails) and figure out where the deep snow is when the deer start bounding versus walking. I'm working on one data set now where you can see dozens of spots where the deer bedded down for the night, but so far I haven't spotted any actual deer in the pictures.

 

[evranch]

Oh. Uh oh. I have a good collection of large 3s batteries here that I'm not prepared to replace... now to try to figure out how to calculate a motor that will deliver the equivalent thrust on 3s. Obviously a ~600W motor with significantly higher kV and probably an 80A ESC.

On the plus side, that means my cruise efficiency is actually pretty good if I am nearly matching your cruise current on 3s.

Or... I can go back to the twin concept. I found a video of someone's Maja twin conversion:

Video is nice and smooth, doesn't look like any vibration is coming in from the wing motors.

Sheep don't bound, they cut nice straight trails through the snow:

https://imgur.com/FghuzCW

Image resolution is cut down significantly so that it's not a huge file, btw. Gaps were caused by poor overlap due to heavy crosswind, but I needed this dataset before the snow started to melt!

 

[evranch]

Did some more research and Bormatec also made a twin: The Explorer. It looks like it's just the Maja with an H-tail and twin wing motors. Look at that climb!

 

[Merv]

I'm seriously considering ditching the pusher for a pair of power pods on the wings, swinging the same 12x8E or the more efficient 11x5.5E.
Other options?.

You might consider a prop with less pitch, a 12x8 prop is built more for speed than torque. It’s likely you could use a 13 inch prop. Something like a 13x4 may improve take off’s. Lower pitch equals more power at lower airspeed, high pitch equals more power at higher airspeed, the plane will go faster.


Great pictures, 5 cm contours is very impressive. I’d like to know how you did this. I’ve wanted to do the same thing only with crop height. Fly the bare field to get ground contour, then again as the crop grow to measure height.

 

[evranch]

I've been kicking around numbers in eCalc trying to make some sense of it. Unfortunately, I need speed, thrust and efficiency, and everyone knows you can't have all of them. Mapping airframes like to fly fast and have high stall speeds. I found the document with recommended motor and prop, and they are saying 12x6 and 400-500W, 50A ESC.

eCalc's estimation of my stall speed is pretty close to reality (30kph) and it warns that I should have a pitch speed at least double that. 13x4 only has a pitch speed of 46kph, which is not even Vr for this airframe! 12x6 is 67kph and 12x8 is 83kph. Yes, maybe a bit excessive. My 11x5.5 was good for 73kph.

However, with the motor I have, the 12x6 does not deliver much thrust! Only 0.39:1 TWR.

In fact, eCalc throws TWR warnings for just about any prop. The 12x8 puts out 0.44:1 which is acceptable, while the 11x5.5 and the 13x4 both only make 0.33:1.

So I tried going twin engine. I grabbed a random motor from the database that felt around the right size. 2x Turnigy D3530/14 1100kV swinging 10x5E. This prop pretty much maxes out the motor, 21.8 of 22A max. Brings the TWR up to 0.6:1, pitch speed of 72kph.

9 minutes at full throttle! My current setup though, is rated at 12 minutes at full throttle. And after all, we're delivering 2334g of thrust instead of 1690g. That's actually more efficient - at takeoff. But at cruise, you have two motors throttled back, losing efficiency.

Also I'm running in and out to check on my new lambs while figuring this out. Thought this would be a good project to plan out in between, but it's harder than I thought.

 

[evranch]

Maybe I'll do a little writeup on what is involved in getting a basic mapping setup off the ground sometime. Here's the basic requirements, though. Honestly, it's not pick up and play but it should be well within the abilities of most of the people here.

You need at the least, an autopilot capable of full auto flight (Mine is Pixhawk running Arduplane), a real camera (With a shutter and a decent lens, not an action cam or phone), a good RX/TX combination with enough range to cover your mapping area and further enough that you can recover an attempted flyaway, and a heavy, sleek airframe that cuts through the wind and holds a solid heading. If your vehicle is buffeted around by turbulence, your pictures will be useless.

You also need good piloting skills. Definitely only for the RC pilot with plenty of experience. You are the autopilot's co-pilot and instructor. It only can fly as well as you can set it up and train it. At some point, you WILL have to slap the mode switch out of auto and pull your airplane out of a power-on stall, unexpected knife edge or spiral dive. Just last week mine cut throttle and started heading straight for the ground due to a glitch. You have to realize what is going on and correct it before you are fixing it with glue... and buying a new camera.

You also need a decent computer to assemble the photos. I would say at least a quad core i5 and 4GB RAM to try it out, to do up to maybe 10 acres. Ideally running Linux. Requirements go up significantly as acreage grows, and if you run out of RAM everything will slow to a crawl. You can use Windows, but you will need more RAM due to its higher overhead and worse swapping behaviour. Some paid mapping software will only run under Windows and will require a beefier computer. Fortunately, this sort of software costs so much that you won't have to worry about buying or running it.

There are also some options where you can upload your photoset and get it stitched in the cloud. Never tried them myself.

Micmac is the software I use. The learning curve is a steep cliff you have to scale. You know how some software is described as a toolkit? Micmac is like a sack of old tools you got at an auction sale. You reach in the sack, and you pull out a tool you've never seen before, and who knows what you are supposed to do with it. Oh yeah, and you have to download the source and compile the software yourself, first! However, it's free, and it yields good results if you can learn to use those tools. There are tutorials online with sample datasets.

 

[DamoRC]

I've been kicking around numbers in eCalc trying to make some sense of it. Unfortunately, I need speed, thrust and efficiency, and everyone knows you can't have all of them. Mapping airframes like to fly fast and have high stall speeds. I found the document with recommended motor and prop, and they are saying 12x6 and 400-500W, 50A ESC.

eCalc's estimation of my stall speed is pretty close to reality (30kph) and it warns that I should have a pitch speed at least double that. 13x4 only has a pitch speed of 46kph, which is not even Vr for this airframe! 12x6 is 67kph and 12x8 is 83kph. Yes, maybe a bit excessive. My 11x5.5 was good for 73kph.

However, with the motor I have, the 12x6 does not deliver much thrust! Only 0.39:1 TWR.

In fact, eCalc throws TWR warnings for just about any prop. The 12x8 puts out 0.44:1 which is acceptable, while the 11x5.5 and the 13x4 both only make 0.33:1.

So I tried going twin engine. I grabbed a random motor from the database that felt around the right size. 2x Turnigy D3530/14 1100kV swinging 10x5E. This prop pretty much maxes out the motor, 21.8 of 22A max. Brings the TWR up to 0.6:1, pitch speed of 72kph.

9 minutes at full throttle! My current setup though, is rated at 12 minutes at full throttle. And after all, we're delivering 2334g of thrust instead of 1690g. That's actually more efficient - at takeoff. But at cruise, you have two motors throttled back, losing efficiency.

Also I'm running in and out to check on my new lambs while figuring this out. Thought this would be a good project to plan out in between, but it's harder than I thought.

That way madness lies!

Kidding, but plugging numbers into eCalcs could eventually drive you insane in an effort to find the absolute optimal power setup for your craft.

A a few things to realize. First, your current setup is significantly underpowered (like you didn't already know that) and over propped. From what I can find out about that Pichler motor online, max recommended prop on 3S is a 10 x 5 so its just not gonna get you the thrust you need with a 12 inch prop, regardless of the pitch. Yes, once in the air you can throttle back and it can cruise, but when the plane tries to pump 100% throttle to take off, you are probably on the worst part of the motors power curve, burning amps but not getting the RPM or thrust. Sure it will get there eventually, but its probably killing itself trying to get there. What is your current draw during takeoffs?

Given that the twin setup is likely to be more inefficient and the fact that at the moment you have a nice flying plane so why mess with it by significantly altering its configuration to a twin, then the simplest option is to pick a better motor that will give you what you need on 3S (like you didn't know that already also).

Given that you already have a ~1000Kv motor and the props that work with it, I don't see any need to significantly increase the Kv. Remember, Kv isn't your problem here, the fact that your motor is too small to drive the 12 inch prop at launch speeds is. So, a bigger, higher wattage motor at 1000-1100kV is what your are looking for. From a bit of googling and faffing around, motors like the NTM PropDrive 3542 (they have 1000Kv and 1250Kv versions), for a few bucks more the Turnigy Aerodrive SK3 motors in the same size are good for 600-800 watts. If tyou want higher end, better quality motors then one of the Eflite Power Series (the 25 probably) should see you good.

 

[evranch]

The reason I kept at it with that prop was that it draws 30A at full throttle in flight, which is the maximum current rating for the motor (and the ESC, a little risky, but the 30A is only sustained for less than 10 seconds at takeoff). The initial spike is to 35A static, but it drops off to 30A within a second or two as the airplane picks up speed on the takeoff roll. After about 10 seconds at 30A I pull back the throttle and climb out at 25A.

Can a brushless motor be overburdened, and not coming up to full speed when it is still drawing rated current? ESCs do not enter a protection mode as far as I know, the current is dropping off due to unloading and not due to current limiting... right? I'll feel pretty stupid if that's what's going on.

Considering motors are a large part of my business, brushless hobby motors and ESCs have always confused me a bit. Probably because they don't seem to have set frequencies and speeds like a 3 phase VFD system would, but just seem to run away up to their kV much like a traditional brushed DC motor. I find it really odd that you can buy a 4 pole 1000kV motor and also a 10 pole 1000kV motor.

I noticed that the way the Turnigy products are rated at least, they list the wattage at their maximum voltage input I.e. the Aerodrive SK3 3542-1000 is rated at 670W, but only 45A. If driven with 3s at 11.1V and 45A, you are only getting 500W out of it. You need to use a 4s to get the 670W.

Still, 500W would be a lot more than I'm getting now. That motor actually looks like a pretty good choice. Edit: Propdrive 3542-1000 is in eCalc, will swing a 12x6 at 49A. Gets me to 0.6:1 TWR without having to make any major mods. Good call. With the same 12x8 it will draw 60A? It still confuses me how one 1000kV motor can draw twice as much current as another with the same prop.

Would it be better maybe to go a little higher on the kV just in case? i.e. the 3542-1185. Rated for a couple more amps, too. I can always go to a smaller prop or lower pitch, but it's hard to go much larger for fear of prop strikes, either on my hand or on the ground during rotation. I do have a couple 11s and 10s kicking around here for testing.

Where should a guy draw the line? These motors are so economical it's easy to go buck wild. The 3548-1050 is a beast, rated at 820W on 4s. At some point, these big motors will be making a tail-heavy plane even more tail-heavy, I guess.

 

[clolsonus]

Maybe I'll do a little writeup on what is involved in getting a basic mapping setup off the ground sometime. Here's the basic requirements, though. Honestly, it's not pick up and play but it should be well within the abilities of most of the people here.

You need at the least, an autopilot capable of full auto flight (Mine is Pixhawk running Arduplane), a real camera (With a shutter and a decent lens, not an action cam or phone), a good RX/TX combination with enough range to cover your mapping area and further enough that you can recover an attempted flyaway, and a heavy, sleek airframe that cuts through the wind and holds a solid heading. If your vehicle is buffeted around by turbulence, your pictures will be useless.

You also need good piloting skills. Definitely only for the RC pilot with plenty of experience. You are the autopilot's co-pilot and instructor. It only can fly as well as you can set it up and train it. At some point, you WILL have to slap the mode switch out of auto and pull your airplane out of a power-on stall, unexpected knife edge or spiral dive. Just last week mine cut throttle and started heading straight for the ground due to a glitch. You have to realize what is going on and correct it before you are fixing it with glue... and buying a new camera.

You also need a decent computer to assemble the photos. I would say at least a quad core i5 and 4GB RAM to try it out, to do up to maybe 10 acres. Ideally running Linux. Requirements go up significantly as acreage grows, and if you run out of RAM everything will slow to a crawl. You can use Windows, but you will need more RAM due to its higher overhead and worse swapping behaviour. Some paid mapping software will only run under Windows and will require a beefier computer. Fortunately, this sort of software costs so much that you won't have to worry about buying or running it.

There are also some options where you can upload your photoset and get it stitched in the cloud. Never tried them myself.

Micmac is the software I use. The learning curve is a steep cliff you have to scale. You know how some software is described as a toolkit? Micmac is like a sack of old tools you got at an auction sale. You reach in the sack, and you pull out a tool you've never seen before, and who knows what you are supposed to do with it. Oh yeah, and you have to download the source and compile the software yourself, first! However, it's free, and it yields good results if you can learn to use those tools. There are tutorials online with sample datasets.

Can I push back a tiny bit on a couple things? (I realize everyone has different experiences and different purposes and needs and goals which is why we end up with different solutions, none necessarily being 100% better than all the others for everything ... so all this is based on my own experience and thinking, that's all.)

First I do agree that fixed wing mapping generally is more complex ... especially with systems we part together ourselves. But my self built fixed wing mapper can collect 8-10x the amount of data per flight compared to a brand new dji phantom 4 pro v2.0 ... and the better camera quality of my fixed wing system shows in the final result.

I don't believe you need a heavy sleek airframe. I believe many mapping platforms end up this way because the design evolved into being a lot heavier than originally intended. I've seen the design process go like this (personally!) 1. Let's design an awesome flying plane ... high fives! 2. Oh we want it to fly 90 minutes so lets triple the battery weight. 3. It needs to do something so lets put in a heavy camera. 4. Now the airplane is almost impossible to launch safely. :-( But it doesn't have to be that way. (unless you have a design constraint like your airplane needs to fit into carry on luggage so you can travel with it.)

I believe you can still get pretty good pictures on windy days. I never fly with a shutter speed less than 1/800 and I usually try to fly with 1/1250 or even 1/2000. This really helps minimize any motion blurring and smearing. It's one of the reasons we picked a sony a6000 with the ability to control all these settings manually.

Let me also make a comment about needing good pilot skill to save the airplane from the autopilot doing weird or dumb stuff. I fly an in-house developed system myself (a friend/coworker designed the hardware and I wrote all the code.) So I understand a bit about bugs, and I understand that px4 and ardupilot have grown to become ginormous complex systems. But I don't think we should tolerate bugs that could put the aircraft at risk or make it do something scary. Those just shouldn't exist and we shouldn't have to live with them. I could get my soap box out of the closet right now and go off on a 45 minute bug and development philosophy rant, but I'll just stick with my abstract here.

After all this I'm reminded of an old college professor that didn't care what our code looked like as long as it ran and produced the desired result. I struggle with that to this day because I fancy myself a code artist. But it's good to remind myself that at the end of the day, if a system is getting the job done, then that is the most important thing by far. So I'm not trying to be critical here, just sprinkle on some of my own experience and thinking. All in good fun!