More crazy homebrew FC experiments

[Cyberdactyl]

Yea, going with HDPE or overly thick carbon fiber or G10 is great for frame strength, but it just passes those hard accelerations on to something else. Hopefully to a zip tie or velcro.

I don't have any pics, but I had a mishap with my cyberhex over the holidays at the parents house. I had the KK2 set for a 3S and forgot to reset it, and was flying a 4S. At ...just... about the time I started wondering why I was getting so much time from the battery, the KK2 went from silent to a constant chirp. I was about 150 feet up and it died about 3 seconds later at around 75 feet, as I was desperately trying to get it down. I was flying at a large horse pasture and it fell into soft earth with high grass and I only broke one of my plywood landing legs. . .or so I thought.

Came home and went out to Dix flying field this past weekend and was flying around at about 200 feet in hard wind and one of the F20A ESC cap wires broke loose. I suspect it had jammed against the frame from the first crash, by the way I have them mounted, and I didn't notice. The crash was odd. It looked as if the wire was sporadically making contact. It would swoop and partially recover, then dive hard and repeat. It did it about three of four times before it lost the battle. This time I broke two props and all the legs and ripped the male XT60 wire free from the harness side.

It makes me wonder how much trust I would put into a 'heavyish' multirotor if it had been crashed, especially if it was tasked to carry something expensive like a multi-thousand dollar DSLR on a nice gimbal.

 

[RAM]

You mean plastic?

I've mentioned these before, but I've been flying with HDPE booms, and other than being about 50% heavier than the typical 1/2" square spruce/pine , it's practically indistructable:

Similar idea. Just make a boom that has a stress relief point in it so it's stiff for normal flight but flexes a little on a hard impact. I think you could do it with a small plastic tubing joint in the booms.

 

[Craftydan]

Yea, going with HDPE or overly thick carbon fiber or G10 is great for frame strength, but it just passes those hard accelerations on to something else. Hopefully to a zip tie or velcro.

For CF I'd agree, but CF is prized for it's rigid strenght, and comes with the cost of either brittleness if it's too weak or stiffness (passing along the impact force) if it's too strong. HDPE is, well, "plastic" in it's deformation properties.

It will bend under impact absorbing much of the force and act as a shock absorber. Unlike a wood boom, it will also bend back so it can be used again.

Now motor's and landing skids are the first to touch the ground, with nothing between them and the plastic but a ziptie, so the weak little ziptie inveribly breaks first (instead of transmitting the change in momentum of the airframe into the motor). Props too, for the same reason. Never had the frame break at the mercy of a crash. Never had a motor break on these booms. I've had landing skids break, but sliced PVC isn't very impact resistant -- a little better than the ziptie holding it on, but not by much. It's cheap -- I don't expect them to always survive.

 

[jhitesma]

Hi jhitesma,

Hey, been so busy with the family and work the last few weeks that I have been ignoring FT. Just no time to play, going to sleep now after working all day But I read the email notification about your post, and just wanted to chime in. I have my TL setup with 6 flight modes on a 9X controller. I have the 3 pos switch and Ail switch setup to work together, so that I have 3 flight modes with Ail up, and 3 more with Ail down. There is a video on how to do it on stock firmware:

Yeah, I'm running er9x (want to switch over to open9x but don't feel like recreating all my profiles) and have thought about setting up something like that...but I have a hard enough time with one switch. Trying to fumble around with two switches and remember which "bank" of settings I'm on is more than I want to deal with while flying.

I'm more tempted to remove one of my pots and wire in a 6 button setup with indicator lights like I've seen some people do. But until I pick up a second TX I'm not likely to do that either.

As for the broken booms. I've never been one to keep my toys pristine - I build them to use them and push them to their limits Plus I enjoy building just as much as flying so repairs don't bother me too much.

I've been thinking about HDPE booms for a long time - in fact I started thinking about it before Crafty's friend initially posted about the ones he was selling. I even contacted him when he posted but the 10" length limit was a deal killer for me since I use longer booms.

I've thought about just buying a thick cutting board and making my own booms from that - they wouldn't be square but should be thick enough to be stiff enough...just one of those things on my project list

I've also been wanting to try replacing the metal screws in the knuckles with nylon screws to see if the screws could act like a fuse in a crash since it would be a lot easier to replace a nylon screw in the field than to replace a whole boom. Might be more expensive in the long run though and I'm not sure if they'd give enough clamping strength to work well.

I have noticed though that my lexan knuckles clamp a lot better than my old plywood ones did. With the ply it was always easy to replace booms and I almost never had to loosen the screws holding the knuckles to the center booms. But with the lexan it's hard to get the booms in there and I have to loosen the inner screws to get new front/rear booms in. So maybe with the lexan providing clamping like that nylon "fuse" screws would work out. Again, just something else on my list of ideas to try!

 

[x0054]

Have you considered some CF booms? A cheaper option would be aluminum booms. I really want to build a knuckle quad with aluminum booms, and weld or solder the booms together, rather then screw, so the entire frame could be one piece. Aluminum does bend easily though, so that could be a down side.

 

[cranialrectosis]

I like what Metal Head has done with aluminum for copters. I note they now make 'upgraded' booms from G10, likely due to the bending issue.
http://quadcopter.us/store/index.php?id_category=14&controller=category

 

[jhitesma]

Aluminum tends to actually be heavier than wood. And you're back to the "just transferring the forces somewhere else" issue when you start using stronger booms like that.

Broken booms don't bother me, they're cheap and easy to replace. If I didn't want to break so many I'd fly gentler...but I get bored flying gentle

That's what I love about my knuckle. It's super cheap to build and repair. Crashes look spectacular and freak people out because bits go flying everywhere. But 30-40 minutes later a handfull of zip ties and about $.50 worth of wood with a few holes in it and I'm back in the air.

I thought about doing carbon booms (I love David's new cf tubes) to make it lighter and more rigid then flying it more "responsibly". But that just sounds expensive and less fun

 

[jhitesma]

Finally got out to a larger field to give this a better try today. Only had Acro/Horizon/vbar setup. Acro/Horizon running with autotuned PID's on the default tuning settings, vbar on it's default settings. I played a bit of pass the TX with Alex (my friend with the warp) and we caught it all on hat cam:

I'm still not entirely happy with how horizon performs on Tau. It does this thing where it holds attitude instead of self-leveling if you go directly into a yaw spin. On MW/Baseflight in Horizon I've never seen that happen before and it self-levels more or less as soon as I enter the yaw. It seems to happen worse when I transition into a yaw spin opposite of the direction of travel. So if I'm going right and enter a right spin it doesn't do it, but if I go right and enter a left spin it does.

Acro feels great. I have it setup really twitchy right now (when Alex tries it you can see how hard of a time he has controlling it because of that.) and this 380mm quad can toss around almost as well as my 250mm. If I had the park to myself and wasn't trying to keep from crashing so Alex could have a turn I'd fly it just as aggressively as the 250 and have no doubt it would keep up. It feels more locked in than MW ever felt for me. I need to upgrade my firmware and try the MWRate mode that's been added...but I'm really not in a big rush because I have zero complaints about Tau's native acro mode. Not sure if that's just a limitation of my flying skills or if it's confirmation bias on the behalf of the people who say Tau doesn't feel as locked in to them

vBar is still....well...interesting. I can feel some potential there and it's kind of neat to fly with - it really feels more like a heli than a quad. But it needs some tuning. Peabody has said a couple of times it should allow for very aggressive flying and flips but right now it's still more sluggish than I have the traditional modes setup for. Will take a look at the vbar tuning soon and see if I can't get it to to where I can flip in that mode. It's almost like a less robotic acro mode - the quad still responds well and doesn't really auto-level...but it doesn't "snap" like it does in acro. I guess you could kind of call it a damped acro mode. But I'm just still having a hard time explaining it and it's something you kind of have to experience for yourself.

Didn't have time to try out any GPS or baro modes today. Still want to experiment with them more. With a "new" refubished 2012 Nexus 7 I had bought for my daughter for Christmas but am claiming for myself since I ended up getting a different (newer/cheaper/has a camera) tablet for her instead I should be able to try out some of the tablet control modes soon. Just still have the tablet in hiding until after Thursday

 

[Mattmax]

Nice build up.
I'm planing to build up an quadcopter with this board too, because I think it's more fun to build up all components than to buy a FC.
So, I have some questions to you and hope that you can answer me some of them. I also write in different forums.
can I use all supplied UART / USART of this board or is this not possible?
I don't have taulabs used till now, becaus ei have now working board with taulabs. So can I configure there everthing? I mean is it egal where I put my GPS, TELEMTRY; BLUETOOTH, Battey-controll or are there some pins where I have to connect them with?
I also found the pinout on github, the same file you are using. But there are only two USART supported and not the rest.
So, finally I hope someone can help me,

 

[jhitesma]

Thanks, I'm not sure I can be of too much assistance though, I've only setup and used UART 2 and 3 so far. It would probably take some digging in the code to figure out just what the situation is with the others:

USART1_RXI (PB7)
SBUS_ENABLE (PD9)

USART2_RXI (PA3)
USART2_TXO (PA2)

USART3_RXI (PB11)
USART3_TXO (PD8)

unused
USART4_RXI (PC11)

unused
USART5_RXI (PD2)

From the pinouts.txt - it looks like there are potentially 5 UARTS - but #1 has "SBUS_ENABLE" listed instead of a TX pin so I'm not sure what's going on there and haven't tried doing anything with it yet as I currently don't have a need for more than 2 UARTS on this build. It could just be that's the official pin designation and they're using it as TX in the code instead. Also 4 and 5 appear to have no TX pin defined - I assuming because whoever originally did the flyingF4 setup didn't have a need for them on their built.

Once you have the UARTS physically wired you can assign them logically to any function you want in the GCS. I've put my GPS on both 2 and 3 as well as my bluetooth. I haven't done anything else with the UARTS on this board yet - I probably will in the future...but have had a lot of projects on my table lately

 

[Balu]

Do you have a recommendation for a bluetooth module? Do I have to look for something special?

 

[jhitesma]

Do you have a recommendation for a bluetooth module? Do I have to look for something special?

I just have the one HK sells for MW - it's worked great for me. A little flaky now after having been crashed a few times...but otherwise "just works" for me (once I remember to set the baud rate on the FC correctly at least - the default rate on Tau was lower than I had my BT set for.)

 

[tuxit09]

jhitesma,

I just wanted to say thanks for spending all that time and hard work documenting this build. I plan to reproduce your work soon.

How do you feel about the position hold accuracy now and did you ever achieve waypoint flight?

Please don't give up the fight with this build. More STM32F4 projects are in our future and we can stand to use more development like yours!

As I said I want to reproduce your work and If I get up to speed here then I would love to help test and contribute any way I can. All components are ordered and some already in hand.

 

[jhitesma]

Sadly I haven't had time to really put pos hold to a good test yet Just been a crazy couple of months for me and not many chances to get out flying. The few times I have got out I've kept it in acro/horizon/vbar and just had a freaking blast flying the snot out of it like it was a 250

I'm just so in love with the way it flies manually I keep running out of batteries before I get around to testing autonomous modes

That and I'm still working on figuring out why USB on my notebook isn't playing nice or trying to get one of the two other notebooks I inherited recently setup for it. So no testing of waypoints yet...I really want to add the flash chip I have to it before I try that anyway...and if I do that I'm going to want to play with logging....

Basically summer is when I do a lot of hacking...winter is when I have too much time flying to make progress on testing new stuff - goes with living in the desert I guess

Always glad to hear someone enjoyed/enjoys one of my builds and look forward to seeing more people doing similar things. I will add that I need to reinforce the mounting of the board. The CA I used to mount the sockets to the perfboard is failing. I need to add some epoxy to it to help make sure it doesn't decide to go flying on it's own mid flip

I really am itching to try pos hold in a bigger field where I'll be more confident letting it go longer by itself, and then moving on to waypoints. But I really need to get this hex built and this balsa plane too...just way too many projects on my table and way too little time!

 

[tuxit09]

I hear ya. I'll be building this one with you soon enough. I had a couple thoughts posted on the f4 thread over at taulabs. http://http://forum.taulabs.org/viewtopic.php?f=21&t=114&p=4013#p4013

http://www.virtualrobotix.com/profiles/blogs/how-to-connect-sd-card-on-vrb-4-5-for-full-nutt-x-operations (one option)

and if u look at the sch for the px4 they used the sdio gpio directly. I know the flash ic is cheap but everyone has sd cards laying around.

 

[jhitesma]

The big problem with SD is speed from what I understand. There was talk on the MW forums about setting up logging to an SD card but the speed of the SD cards just wasn't going to be fast enough to keep up with the data. That could have been due to speed limitations from the 8-bit processor...but I seem to remember it was actually the speed of the SD cards themselves that was the limit. Probably still worth at least looking into.

 

[jhitesma]

So I really wanted to fly this weekend...but my wife had to work Saturday and I had to take my daughter to a classmates birthday party which turned out to be a much bigger affair than I expected (it started at 2, we left at 6:15 and they had just gotten to cutting the cake and were still going strong!) And Sunday I had to spend the whole day elbow deep in engine compartments - first replacing the bulk of the A/C system in our car after the clutch on the compressor seized on Monday leaving us down to one vehicle all week. And then helping a friend install a new tensioner and belt in his SUV. Crazy coincidence...his tensioner pulley seized Sunday before the super bowl - it had been making noise for weeks and I kept telling him to let me take a look at it and he kept insisting that would just cost him money and he'd rather wait until something breaks. I warned him that would cost more but he was stubborn. So instead of just replacing a $15 pulley he ended up having to replace a $35 belt (that no one in town had in stock so it had to be ordered) and a $150 tensioner which broke when the belt snapped because the bearing on the pulley seized. Then the day after the pulley on my A/C seized. Which really sucked because I'd known my pulley was going and had bought a replacement two weeks earlier - only to discover it required a special tool to remove and replace. I had scheduled time with a friend who had the special tool but he wasn't available until Yesterday so I had hoped the car would make it that long. Of course the day after I set that up with him it seized leaving us without a car for a week and resulting in $400 worth of parts needing to be replaced instead of $70. Grrr.

Long story short - I didn't really get to fly this weekend. I did get to do one short test flight because I did make some progress on continuing these experiments.

Saturday night after the party and after my daughter went to bed I finally booted up my notebook and updated my git repo with all the changes that have happened in Tau since mid October when I last updated. Ran into a little snag because the changed versions of QT on me so I had to reinstall QT then log out and back in before things would compile. But eventually got it going. And suddenly GCS on linux started playing nice with USB again! So that's nice as now I can use my notebook out in the field and start experimenting with waypoint flying - fun! And I now have MWrate mode on my knuckle to experiment with which will also be fun.

After I got all that taken care of I did a short test flight on Sunday after finishing with my mechanic duties which went well confirming everything was still ok. So Sunday night I dug in on the upgrade I'm really itching to try....oneshot.

First I had to figure out how to merge in a branch from a different Tau fork - it's not that much different than merging in changes from the official branch but did require me to setup a new remote in my local repo. Once I got that figured out I pulled the oneshot dev branch from ernieift and was able to get it to build. Flashed it to my knuckletau and I now have "oneshot" as an option for my outputs! (it's only available through the uavobjects browser right now - not in the GCS GUI yet.) So my knuckletau is all setup for oneshot as far as I can tell.

Except the ESC's on there aren't ready yet. Tonight I need to fix that and it should only take a few minutes. Just need to upgrade from blheli 12.2 to 13 and they apparently should autodetect the oneshot signal. I'm really curious to see how the combination of autotune and oneshot does. I'll re-run autotune before I do the switchover to oneshot just to make sure I have a good baseline on the same firmware - then I'll update the ESC's and see how things go.

I don't expect a huge difference. This is a bigger slower quad and I suspect the difference oneshot makes will be minimal as a result. But I'm curious to see how well autotune does with it and if I can notice any difference overall.

So stay tuned...more tests underway!

 

[cranialrectosis]

Shoot some video of auto tune please.

 

[jhitesma]

Again? I doubt it will look very different than the last time

But I will try to remember to video again just in case things get "exciting"

 

[cranialrectosis]

I must have missed an episode.

Are you doing anything to fly during autotune?