Have we come too far too fast?

[jes1111]

Re: Al's post
I entirely agree that customer expectations are absurdly high. And, yes - this stuff is really, really, REALLY complicated! Just look at the technical discussions that go in the developer forums. Here's a current example from OpenPilot:

http://forums.openpilot.org/topic/12238-multirotor-3d-position-hold-d/

I have no idea what they are talking about - way, way, WAY over my pay grade! "It's English, Jim, but not as we know it."

 

[R_Lefebvre]

Stacky, you are right. But wouldn't it be possible for the manufacturers to extensively test their product just with a few selected components and then give a recommonation ?...especially Flight Controls. I think the right way for manufacturers should rather be specialization than hunting for every fish in the pond. At least that's how I do it in my business, and this works very well.

Chris

Well, isn't this what DJI tried to do with the S800? And what happened to it?

I think a big part of the problem is as Al said: The airframe and assembly.

I'm in the market for a higher end octocopter, and I'm frankly just really disappointed at the prices I'm seeing for what I consider to be... pretty flimsy stuff. They all look like carbon fiber Mechano sets. And I think that is leading to a lot of the problems. So many people using round booms without even a helical strake to reduce vortex shedding, nor do they include any vibration damping on the motor mounts. All this leads to vibration which confuses the IMU. And don't even get me started on the DJI plastic arm system. Soon as I saw that, I wondered what they were smoking over there. Any 3rd year mech eng student could have instantly told you that is a terrible design for the job with very low torsional vibration. And now what are we seeing?

At the lower end of the market you've got guys trying to fly wooden frames, PCB frames, etc...

Then on the electrical side, most people are using independent ESC still, which leads to a rats nest of wiring, which in turn is prone to failure, and also causes crazy EMF problems affecting the compass calibration and also the GPS reception. Here DJI is one of the few who are doing it right, and mounting the GPS and Mag on a stick to get it as far away from the noise as possible. I don't know why it has taken the market so long to come up with integrated ESC's, I just discovered the HobbyWing Skywalker Quattro yesterday, and the Droidworks Aerodrive system this morning before I wrote this. These will drive up reliability, but still most poeple are using discrete ESC's.

Yet when it all goes wrong, the flight controller gets blamed.

Al is right on the money. These controllers are HUGELY complex. The PX4 board just announced yesterday is using an STM32F4 processor running 210MIPS, which is more powerful than a Pentium processor from 1994. That's incredible! The science of these things has gone from just barely flying a few years ago, to plug and play waypoint flying today. And all on a hobby budget!

So have we come too far too fast? I think the answer clearly is yes. But that's not a bad thing. Sure, things aren't perfect, and some people may experience crashes. But you also have access to an incredible amount of technology at an incredibly low price! Your other option is to purchase something from the low-end military-industrial-complex for something approaching $100,000, and then that craft would barely be able to lift an large P&S camera!

We're all on the bleeding edge here, it's true. But I see this as a really exciting time, and I think being on the bleeding edge of something as revolutionary as personal autonomous flying robots... is not a bad place to be. I've seen a number of people compare this to the personal computer industry in the late 70's, and I think that's right on the money. Back then, people asked "So... why are you putting so much effort into this? What are you going to do with a personal computer?!"

I can only imagine what it must be like now, to be one of the military industry guys watching what's happened in the past few years, after themselves having invested MILLIONS of dollars and trying to sell really small quads for $60k to recoup their costs...

 

[ChrisViperM]

Agreed...the last two lines in your post went a few times trough my head already...and the interesting thing is that the development of military UAV's and zivil UAV's went more or less parallel, no one can say someone stole military development knowlege and tweaked it to zivil use....just wondering what all these highly spezialized military supply companies are trying to sell after having this bright Generals pulled over the table for so many years....


Just to have a smile...maybe that's the future of aerial filming :

The comment of the pilot was... "This is the most stable and smoothest vibration free helicopter I have flown"

 

[R_Lefebvre]

I know in our case it would be pretty impossible to accuse us of stealing anything, as the development is very public, lots of records of what we've done to get here, and failures along the way!

I think the apparent parallelism is simply a case of "shark and dolphin evolution". Two animals which took very different evolutionary paths, but ended up very similar in form because they are both adapted to fill pretty much the same role. That type of thing can't be avoided when you are developing something which does the same job.

Still, I hope we never get into epic patent battles like Apple/Google are in. That whole situation is SO stupid, and a complete waste of resources.

 

[jes1111]

round booms without even a helical strake to reduce vortex shedding

I've never seen helical strakes used on a multirotor arm, but I can't see that they'd be effective, or even appropriate. Low frequency resonance (which the strake is designed to disrupt) is not an issue (that I've heard of) on MRs. A "streamline" arm profile should be effective in preventing vortices in a coaxial setup. For an example of the shallowness of engineering applied to MR frames (with which I completely agree), take a look at Droidworx's after-thought "Boom Nacelles".

vibration damping on the motor mounts

The jury is definitely still out on this idea. Personally I believe this is a bad idea - any flexibility in the mounting of the motor will result in a change of angle of the rotor disk relative to the arm/fuselage as the loading changes. But I, too, would like to see the FC originators being more proactive in assessing (and therefore making recommendations about) these kinds of variables/options.

 

[R_Lefebvre]

How low frequency is the vortex shedding? I believe it is high enough frequency to cause problems (10-100hz) but that is just a WAG. We aren't flying Tacoma narrows bridges here. These are small diameter booms, with very high windspeeds. I assume we've all seen the video of the Droidworx shot externally by a Red in high-speed. When they slow down the video, you can see the arms vibrating significantly. I assume they balanced the props, so if vortex shedding isn't the cause, then what else could it be?

Yes, I've seen the boom nacelles, Steadidrone is also doing something similar. I think it's "better than nothing", but I'm surprised nobody is yet making aerodynamically profiled tubes. I know you can get these in aluminum.

Of the frames I've looked at, I prefer the CarbonCore arms. They make much of the construction simpler (no tube holders required), but I do not like the way the rest of his frame looks with "Tab A in Slot B" type construction all held together with hex spacers.

I want to experiment with motor vibration damping after seeing some videos of it in flight. The quad becomes so quiet, it has to be doing something good. I think there's a feedback, and much of the buzzing sound from the props is an imbalance in one prop exciting the other props to vibrate, which all cause eachother to vibrate more...

From the perspective of the FC makers, I can say that vibration caused quads to be unflyable with Arducopter. We did make improvements to it and I think only the worst vibration causes problems anymore, but I think it still affects them to some extent. And for SRH, it's still a massive problem. The heli has to be quite smooth to work well.

 

[R_Lefebvre]

Oh, and another thing that bugs me... motors that use a collet prop adaptor that clamps on to the 3mm motor shaft, 10mm away from the face of the motor. Seriously? I'm trying to do a budget build, and 99% of the motors I'm looking at are like this. It's just silly. The prop adaptor has to be a solidly machine part that clamps directly to the face of the motor. T-motor seems to use this, but more than I want to spend.

At least if the collets had a through-hole so they could be slid right down to the base of the shaft. Geez.

 

[jes1111]

"things that bug me" - yep, I've got lots of those too, but that's a whole new thread.

Not impressed by the CarbonCore arms - the profile is just a rectangle with filleted corners - that seems like a lot of trouble/expense for a non-ideal solution, aerodynamically speaking.

The only aluminium streamline-profile tubing I've been able to find is far too thin-walled to be effective on anything but the smallest micro-quad. Almost zero torsional rigidity. There is a streamline-profile CF tube available from (I believe) K&S, but it's unidirectional pultruded so, again, low torsional rigidity and poor crushability.

I drew up this mount (attached) more than a year ago whilst discussing anti-vibe motor mounts on another forum. It's based on the idea that the rubber (black parts) would be less willing to compress than shear, thereby limiting torque-induced rotation of the motor body whilst still providing a damping interface between motor and mount/arm. Never tested it because I subsequently concluded that vibration damping was more effectively (and correctly) placed elsewhere.

 

[jes1111]

STM32F4? Yep - OP Revolution has one of those, too

 

[DucktileMedia]

I would like to give +1 on this post. You guys explained and tuched every part off multirotors questiones and problems.
I am aware of price dumping with kopters that were baught with mommy money but they will rethink the whole thing when they have a first crash.

Let me do one story:
We lost one job due to that we were to expensive (for 3 day shoot the difference was maybe 2 3k euro) and the guy that got the job, did not finish it. First kopter fell down and broke the red camera. Second copter could not even lift it so he went home with a loss of all expences and a total damage on his kopter and the producer was left without any footage from air and a broken RED. This is why the ones that hire or should always check if the client can pull it off. And i must say that the pilot is the boss on the shoot. If he feels that it is not safe to do it he should abort or try to find another solution. There is a difference in flying 5kg payload in mexico, dubai, jakarta, germany, sweden ...
Temperatures, moisture, preshure, winds are all factor in this. Termperature on controlers (esc) rises expotencialy with outside temp. Depends on payload and setup but just to show an example on how this goes.
That means on a sunny day with temperatures from :
20 deg outside: 30 deg motors and 40 esc
25 deg outside: 45 deg motors and 55 esc
30 deg outside: 65 deg motors and 80 esc (close to limit for most models)
35 deg outside: 80 deg motors and 100 esc (no safe flying for most models)

Dji in manual is like mikrokopter. THis is manual, no help from presure sensor, smart electronics so its like clasic hely.... So u need to be skiled on quad to make it fly, hexa and okto are easyer to fly in manual but still hard for someone that is used to atti mode.

I take this comment as a complete insult! Maybe others that fly in manual mode have it setup to work that way. This particular Wookong simply would not fly in manual mode. It had erratic control that instantly flipped it upside down. this was NOT pilot error other than the fact the switch was somehow flipped before take off. We can both fly in manual mode perfectly fine and truthfully I am starting to prefer it now that I am using Hoverfly. But please don't pat yourself on your royal back so much that you look down on everyone else.

 

[jes1111]

Then on the electrical side, most people are using independent ESC still, which leads to a rats nest of wiring, which in turn is prone to failure, and also causes crazy EMF problems affecting the compass calibration and also the GPS reception. Here DJI is one of the few who are doing it right, and mounting the GPS and Mag on a stick to get it as far away from the noise as possible. I don't know why it has taken the market so long to come up with integrated ESC's, I just discovered the HobbyWing Skywalker Quattro yesterday, and the Droidworks Aerodrive system this morning before I wrote this. These will drive up reliability, but still most poeple are using discrete ESC's.

I knew I'd missed out something for your earlier post

I can't agree on these 4-in-1 and 8-in-1 thingees. They turn multiple low(ish) cost parts into a single, high-cost part (and, potentially, a single point of failure that could bring the craft down). Assuming they use discrete processors for each motor, what's the benefit of combining them onto a single board other than the "ooh" factor? All I can see is disadvantages - financial, thermal, etc.

Now - a so-called "power distribution board" that acted as a backplane for pluggable ESCs - that would make more sense to me.

 

[R_Lefebvre]

I knew I'd missed out something for your earlier post

I can't agree on these 4-in-1 and 8-in-1 thingees. They turn multiple low(ish) cost parts into a single, high-cost part (and, potentially, a single point of failure that could bring the craft down). Assuming they use discrete processors for each motor, what's the benefit of combining them onto a single board other than the "ooh" factor? All I can see is disadvantages - financial, thermal, etc.

Now - a so-called "power distribution board" that acted as a backplane for pluggable ESCs - that would make more sense to me.

High cost is not a given. The Quattro can be bought for $30, and a 20A ESC is at least $10. So it's actually cheaper. And to me, if the whole thing went, $30 is not a big deal. I am struggling with the idea of using 2 of these on a Octo, however. If something common to one of them failed, I'd lose 4 motors instead of just 1. Of course, that also means there are less failure points in the first place...

I agree, your suggestion would be the best. Modular ESC's built right onto a PDB.

I just look at the mess of wiring on my quad and agian, I shake my head. There's got to be a better way. And that's with a PDB that attempts to simplify the power and signal wiring.

 

[jes1111]

The Quattro can be bought for $30, and a 20A ESC is at least $10. So it's actually cheaper. And to me, if the whole thing went, $30 is not a big deal. I am struggling with the idea of using 2 of these on a Octo, however. If something common to one of them failed, I'd lose 4 motors instead of just 1. Of course, that also means there are less failure points in the first place...

Four ESCs for $30 - even scarier!

I just look at the mess of wiring on my quad and agian, I shake my head. There's got to be a better way. And that's with a PDB that attempts to simplify the power and signal wiring.

IME the smart way to do it is the "old-fashioned way" - a wooden board - draw out the fuselage/arm layout on it - hammer nails at every corner and start/end point - build and bind the "wiring harness" as a complete component - install on the craft.

 

[R_Lefebvre]

Four ESCs for $30 - even scarier!

IME the smart way to do it is the "old-fashioned way" - a wooden board - draw out the fuselage/arm layout on it - hammer nails at every corner and start/end point - build and bind the "wiring harness" as a complete component - install on the craft.

Well, lots of people using $10 ESC's, and if you figure the savings due to the integration... I don't think it's unreasonable.

Structured wiring harness, another pretty good idea. Still, even the best laid wires are causing EMF. The idea is to reduce the wiring as much as possible.

 

[jes1111]

Structured wiring harness, another pretty good idea. Still, even the best laid wires are causing EMF. The idea is to reduce the wiring as much as possible.

Forgot to mention - twisted wires - lots of twisting - twist - twist - twist - then twist again! (P.S. Use hot glue to stop 'em untwisting!)

 

[Lanzar]

I take this comment as a complete insult! Maybe others that fly in manual mode have it setup to work that way. This particular Wookong simply would not fly in manual mode. It had erratic control that instantly flipped it upside down. this was NOT pilot error other than the fact the switch was somehow flipped before take off. We can both fly in manual mode perfectly fine and truthfully I am starting to prefer it now that I am using Hoverfly. But please don't pat yourself on your royal back so much that you look down on everyone else.

i did not want to offend you but not holding the transmiter in hands and fliping it in manual mode' what did u expect. And from personal experience wookong on small quad in manual is a nightmare to control. But then again same setup with another wookong and same settings reacts totaly diffrent.

 

[DucktileMedia]

Your post implies that one needs to learn to fly a quad first, implying we dont know how to fly our $12k 25lb hex. Accidents happen to everyone. We were lucky it was in an open field and not over one of the models laying at the resort. My point was that manual mode on the DJI sucks! If you compare it to flying a srh or Hoverfly or even Xaircraft for that matter, it is absolutely horrible. but what happened was not because it was "hard" to control, it was impossible to control. It doesnt just flip for no reason its own. I have video of it but it doesnt help much as it justs takes off wobbles and flips.

 

[mailman35]

how is the quattro? i just saw it on himodel.com

High cost is not a given. The Quattro can be bought for $30, and a 20A ESC is at least $10. So it's actually cheaper. And to me, if the whole thing went, $30 is not a big deal. I am struggling with the idea of using 2 of these on a Octo, however. If something common to one of them failed, I'd lose 4 motors instead of just 1. Of course, that also means there are less failure points in the first place...

I agree, your suggestion would be the best. Modular ESC's built right onto a PDB.

I just look at the mess of wiring on my quad and agian, I shake my head. There's got to be a better way. And that's with a PDB that attempts to simplify the power and signal wiring.

 

[R_Lefebvre]

how is the quattro? i just saw it on himodel.com

I actually never got around to ordering it unfortunately. I decided that the octo I'm building really needs 40A ESC's, so it wasn't suitable. Still, I probably should pick one up to use on my little quad. It would sure clean up the mess of wiring!

 

[mailman35]

thats what im thinking, it would be nice if it did come in a 30a model.
just curious as to how it works with some of the FC's, naza, mwc,etc.