• This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn more.

Flitetest DIY Gremlins

#81
Hello everyone!

I was lucky enough to get my hands on one of these power pack combos about a week or so after they showed up on the Vlog. I finally got the rest of my parts last week and gave my hand at designing a frame. Here's the results:

Frame.jpg

I've never printed in anything but PLA. I actually decided to try out some Ninjaflex Armadillo. It came out a bit more bendy than I planned, but still pretty stiff and very durable.

readytogo.jpg

Here we're all wired up with Cam, vTx, OSD, and XM+ receiver. So tiny!


And here's the finished gremlin:
alldone.jpg

89.9g AUW

Since taking that pic I've had to go back in and move the signal wire for motor 4 and remap in betaflight, since it conflicts with LED resource when using dshot. Figured that I needed to do that last night, so I've only had a little flight time with it. The punch out is insane!!!!

This is technically my first drone. Well second if you count the monster with 12" props that was driven by an unholy marriage of an arduino and a wii sensor that I was not so gentle with. But that one didn't fly so I'm counting this as numero uno.

I've been researching mini and micro drones since December, itching to get in the air. Definitely well worth the wait!
 

jhitesma

Some guy in the desert
Mentor
#82
Well it flies.

20170407_081411.jpg

Kind of hard to get a photo of by myself. Took these while running dRonin's autotune:

20170407_081416.jpg

Gotta love being able to snap photos while tuning :) But this was the 3rd tune on the pack and it was getting low so not touching throttle was risky. Almost hit the ground at one point and had to quickly get back on the stick.

Tune results weren't bad:
http://dronin-autotown.appspot.com/...hdXRvdG93bnIYCxILVHVuZVJlc3VsdHMYgICAkJa2xAoM

First tune I got a 25ms tau - fairly respectable for something this small. Last go it sagged to 34ms but I wasn't surprised since my hover throttle went from just under 50% to a bit over 50% due to the pack sagging, so a loss of performance was expected.


It's by no means done yet though. I made a few little tweaks to the frame and re-printed it. I shrank the FC mounting posts by .5mm and raised them .5mm taller to clear the DSM port. Which I may go ahead and just remove anyway since all my DSM sats have the connectors removed so I just soldered to the back of the port anyway :p

Additionally I added a small gap to the motor pods for motor wires to pass through:
Screen Shot 2017-04-07 at 9.33.01 AM.png

And finally I lengthened the arms 5mm each so it's now 100mm motor to motor :( This was to give room for the ESC's to fit on top of the arms...but...I ended up mounting them underneath anyway :p Mainly because I decided I really need to switch to a 4 in 1 esc for this guy and didn't want to cut my motor wires. So...the way my ESC's are laid out is pretty goofy. Basically the ESC for each motor is on it's opposing arm. This caused me a little bit of confusion when I wired up the motors and they were responding in the wrong location - but thankfully dRonin makes it super easy to remap those.

The build is super ugly right now, I was really rushed because I desperately wanted to "just get it in the air" last night but we did a family trip to the county fair after work, then I had to wait for the new frame to print so it wasn't until 10PM that I even got to plug in my soldering iron.

20170407_000311.jpg

Doesn't look THAT bad from the top...other than not having the FPV cam on yet. I got a mount printed for my TX02 cam but need some spacers before I can install it.

From the bottom though:

20170407_081931.jpg

The ugliness really shows. The RX is just shoved in there with wires that are really too short. If I remove the connectors from the RX and the FC I can probably make the RX fit under the FC for a final build.

Also, all that extra wire. I need to shorten the signal wires on the ESC's, shorten the power wires on the ESC's, and remove the signal ground wires (they're not connected right now and the pico doesn't have an obvious good way to connect them.) That will probably shed a few grams.

And I may as well go back to my original 90mm frame if I'm going to assemble it like this.

But...I want to add some tabs on the side of the frame for the battery rubber band so going to print one more frame anyway. And if I'm going to go with the ESC's underslung like this I may just shink down to a 80mm frame while I'm at it...depends on what mood I'm in.

Even as a 100mm it's a little sketchy with high rates. Feels nice and locked in, but easy to get into trouble fast LOS on something this small!


I'm also already using it as a test bed. As already mentioned I'm running dRonin on it - and pico blx support isn't yet official on dRonin so this is an experimental build right off the start. I started with a recent nightly but pico blx doesn't have the new autotune code enabled due to a memory constraint one of the early testers ran into. Since that test though some tweaks were made to the new autotune code to make it less memory intensive due to Brain RE1 having the same issue. The lead developer suggested that I try enabling that and seeing how it does. The change was super simple, just two lines to adjust in a header file: https://github.com/d-ronin/dRonin/pull/1701

This morning before work I quickly gave it a go...and got back results.

Screen Shot 2017-04-07 at 10.11.19 AM.png

Screen Shot 2017-04-07 at 10.11.06 AM.png

Those are the most interesting bits of the new autotune code results. Y is pitch X is roll. The green line is what autotune commands the quad to do - the blue line is what the gyros read out in response. The better the two lines match up the more accurate the tune is. And the smoother the blue line - the less noisy the quad is. This is by far the smoothest result I've got.

Which honestly surprised me because I'm not overly impressed with these 1103 motors. When I setup my neutral points on them the first one spun up nice and smooth and quiet at a lower input than I expected - I was hyped. Then the second one...needed more input than I expected and I could see it wasn't perfect - the shaft is apparently very slightly bent as I can see a tiny wobble in it :( I did drop one of the motors once or twice so that's probably my fault. But...the next two motors fresh from the packages and un-dropped also took more power and were louder and rougher sounding than the first two. So...not overly consistent. But can't argue with the end results, I was doing flips in the office this morning - which wasn't a great idea since I can barely tell the from from the back flying LOS on this thing!

Oh yeah, AUW with battery but no FPV yet ready to fly came in at 79g. Bit heavier than I was hoping, but still a lot of weight I can ditch.

And also - managed to have a few decent crashes already. Into the bricks in my yard, into the wall of our office, off the floor of our office...so far the PLA frame has held up better than I expected with no damage!

So tonight hopefully I'll have time to move this to a slightly smaller frame, clean up the wires and RX, then get some packs charged, add the FPV gear and rip up my backyard if all goes well! Guess I should try and get my DVR working too now....
 
Last edited:

LitterBug

Troll Spammer
#83
Great stuff @Darksabre_X and @Jhitesma! How are you getting those dRonin autotune graphs @Jhitesma? I need to do some reflow work on my 4in1 ESC before mine goes up again. Knocked a cap off that was really close to one of the mount holes and it is totally messing up my FPV signal. Picked up a cheap hot-air rework tool yesterday and need to play on some old scrap boards before working on the real thing.

Cheers!
LitterBugs
 
Last edited:

jhitesma

Some guy in the desert
Mentor
#84
Great stuff @Darksabre_X and @Jhitesma! How are you getting those dRonin autotune graphs @Jhitesma?
I talked icee into sharing his jupyter notebook that does the new autotune so I wouldn't have to keep bugging him to do new tunes for me :D Took me a bit to get the right version of python and find all the prereq's to get it running but mostly got the hang of it now. Though there are still little things that trip me up. Like a change that got merged into next which changes when some filters run so there's a change in how the notebook has to be configured now. Plus I'm still learning what all the results mean. The full results are here: https://nbviewer.jupyter.org/gist/jhitesma/4f0e836044f0c24dd618b9f58c036171

Today I learned that the last six numbers are axis correlations -> r:r r:p r:y p:p p:y y:y so now I know how he can tell when there's bad cross-axis coupling in a tune :D


I need to do some reflow work on my 4in1 ESC before mine goes up again. Knocked a cap off that was really close to one of the mount holes and it is totally messing up my FPV signal. Picked up a cheap hot-air rework tool yesterday and need to play on some old scrap boards before working on the real thing.
I really want to get a 4in1 on this...but kind of want to find one that can handle 3s now that I've tried this. It's not underpowered...but I feel like 3s would be a lot more fun ;)

This is also still on those cheap 2030 banggood props not cut down so a full 55mm. I may try some cut down to see if it does any better. And I ordered some of the HQ T3x3's and T3x2's from gothelirc since they're supposed to be the best on these motors. Though apparently HQ isn't going to make any more of the 3x3's and gotheli's stock is pretty picked over :(
 

LitterBug

Troll Spammer
#85
I'm running a 10A 4in1 that does 2-3S. Got mine from FlexRC with the frame and motors. I've seen them elsewhere, but you have to look around for them.

Cheers!
LitterBug
 

jhitesma

Some guy in the desert
Mentor
#86
I'm running a 10A 4in1 that does 2-3S. Got mine from FlexRC with the frame and motors. I've seen them elsewhere, but you have to look around for them.
Good to know, I haven't really started shopping around yet since the budget is tight with the fair in town this week ;)

I did a few more flights with the last of the power in the pack here in the office...and have to say 2s is honestly more than enough for inside. This is a LOT more power than a whoop style setup as is.

LOS with single color props and acrodyne with my rates turned up is proving to be a great way to test the durability of the frame :D :applause: So far it's holding up great, just a few scuffs. But a few of the office plants have a few less branches on them :black_eyed:

Good thing I only had time to charge one pack. Wish I'd had time to get the FPV cam on and charge the rest of the packs!
 

jhitesma

Some guy in the desert
Mentor
#88
Got the FPV gear added today:

20170408_174501.jpg

Current ready to fly weight with 450 2s pack is 80g on the nose. And there's still quite a bit of extra wire I can trim out of there. Not as light as a hoverbot...but about the same as the CF frame Verris is building up.

Having a blast flying it. It's a lot faster than I expected....first FPV flight in the backyard I was caught off guard and crashed into the side of the house at high speed - no damage. In fact, despite crashing it quite a bit so far the most damage I've done is knock the FPV cam off. It's proving very sturdy. I did knock a prop off a few times, but not lately.

I may print a version in ABS...it should be a little stronger and ABS is about 20% lighter than PLA....could help shed a few more g's.

Speaking of props. I love living in the same state as gothelirc.com ordered some HQ T3x3 and T3x2 props on Thursday, they were in my mailbox (with a watermelon jolly rancher) on Friday:


20170407_203442.jpg

Going to give them a try later...but want to get a bit of DVR footage with these props first since they're working. The HQ's look a lot bigger so I'm not sure how well they'll work. I've heard good things about them but they sure seem like they'd result in lower tau since they've got to weigh more.

Have more construction photos and am almost ready to share a STL - but may start a separate thread for that...undecided so far.

Would still like to get some confirmed measurements off the emax motors so I could do a version of this for that setup.
 

jhitesma

Some guy in the desert
Mentor
#89
Oh...and...was watching a makersmuse video on Fusion360 the other day and he commented on how it usually takes at least 3 tries for a design to go from first print to usable print.

I wish I was that good:

20170408_183312.jpg
 

JimCR120

Got Lobstah?
Site Moderator
#90
My Gremlin - In design phase

Well I've been a bit stagnant on actually building my projects due to learning Inkscape so I can design and skin them. My first project is the FT Gremlin. No fancy 3D printer for me yet so it's just wood and foam for now. We'll see how it gets tweaked.

I already know I want to space the motors apart a bit more to accomodate the prop guards/ducts. I also need to incorporate some standoffs so I can put some of the components in layers.

This was done in Inkscape using the layers so I can assign and hide parts of the quad. I have an Airframe layer and an Electronics layer. When I get this how I want it I can hide the electronics and print the scale airframe that I can transfer to some hobby plywood and cut.

I'll also plan for a battery holder and camera mount.

Screen Shot 2017-04-10 at 12.26.43 AM.png
 

jhitesma

Some guy in the desert
Mentor
#91
Well, crashed mine hard enough and repeatedly enough yesterday to feel fairly confident in sharing it finally - so I put it up on thingiverse:
http://www.thingiverse.com/thing:2240071

(Also, had to get permission from the guy who's frame inspired mine since he had originally shared it with a no-derivatives license. After I contacted him for permission he removed that restriction :applause: )

So far the PLA has held up great. Even after a couple of high speed crashes into pavement and my house no damage. And the gyro plots are smoother than Verris's CF build which isn't really any lighter. It may be even better printed in ABS which should be a bit tougher and lighter so I may give that a go...I just hate printing ABS.

This thing is an absolute blast. My only complaint is that I'm not nearly a good enough pilot to fly it in my backyard :( But, hopefully being this small I'll get to fly it in more places and build those skills finally ;)
 
#92
Welp, I'm done for now. No more of the stock props left. :(

Well worth the wait and the crashes. Now all I want to do is get back in the air!

Anybody got any good recommendations on alternate props for these power pack motors? I think I read somewhere that the prop holes on top are spaced wider than other 1104s on the market, but I have no way to check if that's true. If they are, I'm grounded until Emax starts selling these 2345 props separately. Or until I buy a different motor/prop combo.
 
#93
I've been curious about building an FT Gremlin myself. I've never built a drone before, but this may change my mind! ;) Will be definitely be keeping my eyes out on this one...
 
#94
This thread is making me so excited!

Also Josh B's comments in the last podcast. GREMLINS SOON!!!



Still waiting on shipment from Emax on an 1104 power pack.


I did get a set of just the motors today in the mail which was a separate order so some of their stuff is shipping.

For these motors I'm waiting on the new RaceFlight Milivolt FC and Spark 4-in-1 ESC. I'm hoping these wont be too big for this application (20mm mounting holes).
 

jhitesma

Some guy in the desert
Mentor
#95
Well my little PLA frame is holding up great.

I reprinted the camera mount in TPU though since it broke too easily in crashes. TPU version has held up great.

20170416_211022.jpg

I have had a few little issues:

1) When I crash (or just land) the battery lead gets pushed up into the prop and is now in danger of shorting out due to the insulation being damaged. It's really an issue with my battery leads length than anything else. But I may replace the lead on the FC with a slightly longer one so I can have a bit more flexibility in positioning it.

20170416_211007.jpg

2) I utterly destroyed the 5.8 cloverleaf on my TX02 aio cam/vtx. It took quite a few good hits but Friday I hit the ground upside down at full throttle (taking my daughter for a "ride" and I only have one pair of goggles so I had to fly LOS and orientation is REALLY hard on something this small!) and the antenna was completely flattened. I might be able to repair it but the shield on the coax tore at the top and bottom and the center conductor tore out...so...easier to just start over. I ordered a 3 pack of replacement antennas but for now did a "mullet mod" with a dipole that came with one of my old VTX's. See, there's a reason I didn't just toss them :) Just pop off the plastic casing, cut the coax in the hinge area...and you've got a "free" dipole. I'm not a fan of this solution since a linearly polarized TX antenna and circularized RX antenna gives a 3db loss which is pretty significant with a 25mw vtx...of course my TX02 is 200mw so not as bad but I still don't like it. And multipathing at 5.8 can get nasty which is why we run CP antennas in the first place. But point is I'm up and flying and the "Mullet" style dipole IS much sturdier. I gave this little guy far more crashes with this than with the original antenna and it's holding up great.

20170416_211017.jpg

3) I bent a motor shaft :( Even with the protection around the motors on this frame I managed to crash into pavement hard enough to bend a shaft :( It still flies...but not nearly as smooth anymore. Ordering some new motors (I'll have 3 good ones left...maybe I should get crazy and try a tri again...) Of course on the upside it's pretty impressive that my little PLA frame survived a crash hard enough to bent a metal shaft!

20170416_211034.jpg

4) The accels on this piko blx board leave a lot to be desired :( Leveling mode is not great. They gyros are fine, in acro mode it's super locked in and flies great. But in leveling...it drifts to the left no matter what I do. I won't go into the whole list of things I've tried but yeah...sometimes it just drifts a little and I can kind of live with it...but a lot of times it drifts bad and I can barely even keep it under control. Apparently this is fairly common on these piko blx boards too. That was one of the reasons why I chose to get a clone board - to see if it would be any better in this respect as one theory about what causes is that it could have to do with how the soldering is performed in manufacturing. Since my clone that's mostly likely not made in the same facility has the same tendencies it's probably more a matter of design and physical stresses on the board. Since I don't really fly in leveling mode this isn't a huge deal to me...but leveling is "needed" for autotune on dRonin so it's nice to have. I put needed in quotes though as there is an undocumented way to autotune in acro. You can setup a custom stabilization bank and select "SystemIdentRate" as the mode for each axis. That runs autotune in rate/acro mode instead of leveling - but it's riskier and not as well tested. A few of us have tried it and had good results but it's a bit more of a handfull while doing the autotune wiggles.

There's also a very good chance the the leveling issue IS still configuration related. It's very hard to zero the accels on this build because the bottom of the quad isn't flat. So it could just be the ways I've been devising to make it sit "flat" while doing that not being very accurate. So not tossing anything under the bus entirely just yet ;)


Overall though I'm loving this little guy. I still haven't tried the nicer props I got. I've actually been on my first set of these cheap banggood 2030's until today when I finally switched to a fresh set to see if the somewhat ragged ones were the cause of my leveling issues. Then had the shaft bending crash on the first flight with fresh props and made my fresh props worse than the ones I just replaced...because of course I did :p

I also haven't flown a full pack FPV yet because I really need to design and print some prop guards so I can bounce off of things instead of crash so quickly when the speed catches me off guard :D


I also still really want to adapt this design for the emax parts...but still haven't seen confirmed measurements on the motors. I have seen two different dimensions posted in various places so I'm hesitant to trust either. I know in the last podcast they guys said they'd be sharing them soon...and I know TJ said so back on like page 2 or 3 of this thread. So I know they're coming...but figured I'd just mention it again as a little reminder ;)
 

LitterBug

Troll Spammer
#96
Mmmmmm.... Me thinks vibrations from that bent motor could be contributing to that accell issue. This quad frame looks amazing! I'm all for you doing a Tri. Will save me the work later this year when all my current stuff is done. (way too much on the bench already). I have plenty of motors, just need to come up with a frame and FC. May have to do 1306 and 1104 tris.

Cheers!
LitterBug
 
#97
Does anyone have a 3D model of the camera used in this setup? I'm trying to design a more robust camera holder for my micro quadcopter, and I need better measurements for the modifications. Thanks in advance!!
 

cranialrectosis

Faster than a speeding faceplant!
Mentor
#98
I'm all for you doing a Tri. Will save me the work later this year when all my current stuff is done. (way too much on the bench already). I have plenty of motors, just need to come up with a frame and FC. May have to do 1306 and 1104 tris.

Cheers!
LitterBug
I would love to go here but wonder where to get small enough MG servos to handle the tilt.
 

cranialrectosis

Faster than a speeding faceplant!
Mentor
#99
There's also a very good chance the the leveling issue IS still configuration related. It's very hard to zero the accels on this build because the bottom of the quad isn't flat. So it could just be the ways I've been devising to make it sit "flat" while doing that not being very accurate. So not tossing anything under the bus entirely just yet ;)
Can you print a stand?