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

Modular/Swappable glider platform

#1
Over the past few months, I've been messing with gliders. It all started when I discovered that I could thermal with my Old Fogey swappable.

I found the Mini-ELF DLG and figured something like that (2-channels, lots of dihedral, small) would be good for me to start with. Like anything I now fly, I had to build it (I get so much more enjoyment when I build them myself). So, from the drawings I found, I scaled them and built my own Mini-Elf...

Mini-Elf.jpg
Wingspan: 800mm

I flew it a few times, then one launch it took a dirt nap. The pod was toast, but the rest of it was fine. So, I redesigned the pod, gave it a V-tail and tried again...

Mini-Elf-v2.jpg
The tail separates from the rest of the pod and there's also a removable battery cover not visible because of the LiPo sandwich.

The v2 flew fine, but I kind of wanted to experience motor gliding so, I built a power pod for it. Then I had a midnight revelation and redesigned it yet again. This time, I made the pod easier to swap out (in-field).

Then, I saw the recent Air to Air episode where they talked about a swappable glider. I thought, "I'm almost there already."

I took it a step further today by making a cruciform (cross) tail for it, so here it is...

Mini-ELF-v3-overview.jpg
From top: 800mm KFm2 with DLG grip; 1m KFm3 with DLG grip (since this is v3, I named it 3LF--lame, I know); V-Tail attachment (left); Cruciform (cross) attachment (right); glider pod (left side); motor pod (left side); electronics pod (right side).

Mini-ELF-v3-pods.jpg
A closeup of the motor and glider pods.

Mini-ELF-v3-socket.jpg
A closeup of the electronics pod. Also shows the tail mounts.

Mini-ELF-v3-1mCruciform.jpg
An example configuration:
1m KFm3 Cruciform (cross) tail DLG with 2, 1S 400mAh Zippy Flightmax LiPo. AUW: 98.8g

I haven't bothered covering the pods yet for a couple reasons--mostly because I want to get it right before I make it pretty.

This may not be what Josh B and David had in mind by "swappable glider", but I've had a great time building it.

Cheers!
Randall
 
Last edited:
#2
I took the 1m cruciform config out for a few tosses this morning. Got a nice 50 second flight on the third discus launch.

I got the bright idea of adding a bit of down elevator when the GEAR switch is thrown (I've seen DLG clinics that mentioned doing it). I added too much and lawn darted it about 15 feet in front of me. Heart stopping!

Fortunately, the only damage (aside from a dirty nose) was that the rear wing mount was partially separated from the wing. A bit of CA and a piece of 1mm depron for reinforcement solved that.

Now for the purpose of this reply... maybe it needs to be it's own topic?

I was thinking of building another 1m wing with flaps and independent ailerons (for crow). The difficulty is that I'd still like to be able to use the motor pod.

Does anyone know of a device or a way to split a channel into upper and lower halves? What I have in mind is using the top half of throttle to control the motor pod and the lower half to control the flaps. I have a bunch of 12F675 PICs if anyone can point me to some code.

Thanks in advance for any help.

Cheers!
Randall
 

Craftydan

Hostage Taker of Quads
Moderator
Mentor
#3
Does anyone know of a device or a way to split a channel into upper and lower halves? What I have in mind is using the top half of throttle to control the motor pod and the lower half to control the flaps. I have a bunch of 12F675 PICs if anyone can point me to some code.
Randall,

Depending on your Tx, you might get away with a few mixes: first bump up your offset on your THR control so it's at 0% at the middle and 100% at the top, then mix your flaps channel to the trottle control. Under power you'll probably want the flaps all up, so that at 50% postion set the flaps to be all up, and at 0% postion all down.

the pic as a "servo translator" might work to reverse this (sawtooth: at bottom -> 0, at middle -> 100, above middle -> 0) but if you like having the flaps drop as the throttle stick does, and you can't overdrive your servo/linkages, KISS applies.
 
#4
Thanks for your response Crafty Dan.

The issue I'm having is that the throttle and flaps are both on the same channel--which makes using a mix difficult. I spent a bit of time messing with er9x last night too, trying to see if there was a way to do it with a complex mix.

During a midnight inspiration, I found a couple snippets of assembly language for PICs: one that takes servo input >1700ms and turns an LED on--and turns if off <1300ms; and another that is a PIC based servo tester (generates PWM). I'm muddling my way through merging the two into what I want to do with it: top half of throttle controls ESC, bottom half lowers flaps.

I knew I should have learned assembly language earlier!

I'm still open to ideas. I don't know when I'll be able to figure out the code.

Cheers!
Randall
 

Craftydan

Hostage Taker of Quads
Moderator
Mentor
#5
PHP:
Randall,

Ah! Mixing a multiplexed channel - now that is interesting.

I've played quite a bit with the 12x675's before, and an rx-side channel splitter sounds like an interesting problem. If you'd like a hand with it, PM me code and we can talk specifics.
 
#6
A fellow over at RCGroups actually sent me a link to a servo reverser that will almost do what I need: http://www.rcgroups.com/forums/showthread.php?t=1895138.

I just need to figure out how to expand each half of the pulse width to a full pulse width on each output--and find the time to do it.

One more question if I may:
I read somewhere online that PWM for servos is 700ms to 1700ms (midpoint at 1200ms) with 300ms delay to separate channels (1000ms - 2000ms total), but my Turnigy 7-in-1 tester goes from 700ms to 2300ms (midpoint 1500ms). Should I be sticking with the spec, or go to 2300ms?

Thank you Crafty Dan for your help.

Cheers!
Randall
 

Craftydan

Hostage Taker of Quads
Moderator
Mentor
#8
Randall,

That's a new spec for me, but it could be. Most of the servos I've used had hi-tech's (and possibly other's) pulse pattern of 1-2 ms pulse every 20 ms, centered at 1.5. The frequency doesn't have to be as low as 3 Hz for channel separation -- there's only one channel per signal line (the Rx/Tx, however may not be up to the task of sending n-channels at 50 Hz).

No surprise, but center/edges may be manufacturer specific -- one more reason most Txs have throw settings. If you've seen someone spec .7-1.7 for your brand servo, I'd ignore what the tester can do - it's built to be flexible w/ everybody's specs.

Hmmm. looking at the code/specs, I'd forgotten the 675 only has one 16b and one 8b clock. Seeing as you have one synced, then three independent events (input transition to low, throttle transition to low, and flap transition to low), might be dicey. If you can step up to a 615 or 617, you'll pick up another 8 bit timer. (I'd avoid stepping up to the enhanced midrange processors -- they've got *fantastic* specs, but far more h/w to setup -- I've had little success getting everything right).

The algorithm should be simple: In ISR, Set both "outs" high when your "in" transitions high, setup T1 (16bit) as your read, and T0&2 as single shots to turn off your outs at the pre-calculated time. What pre-calculated time you ask? Use your non-ISR program loop to convert the pulse width you read into counts for each of your outs (and error checking, etc., then idle until interrupt on next in pulse). By the time the next pulse comes in, you'll be ready to send the match pair for the last pulse. You'll always lag by one cycle, but your signal will be stable.

for the 675 you could calc the stop time for out1 and out2, load the lowest into T0 at "on", then load (high-low-lag) when it triggers, all the while being careful with the out1 is close to out2 -- which is just messy.


Hmmm. On second thought -- Bruce's code should work just fine. I'm so used to interrupts that I didn't consider just running a delay loop.

Noticed also that the code implements a trimpot and LED -- nice features, but also consider using grounded pins for "servo reverse". you'll have three free to use as you see fit -- use them wisely.
 
Last edited:
#9
I've written the Process tag for the 700ms - 2300ms pulse width, but it can easily be changed.

I have no need of a trim pot or LED and will likely remove them--and also to keep the weight down. By grounded pins, do you mean putting a jumper so that I can reverse one or both servos?

Anything else I can remove from the circuit to keep the weight minimal? Can I strip it right down to servo connectors and the PIC?

Thanks for your help Crafty Dan.

Cheers!
Randall
 

Craftydan

Hostage Taker of Quads
Moderator
Mentor
#10
Haven't seen your circuit, but yup: 3x three pin connectors, PDIP package and you're done. If you're careful about which GPIO pins you pick, you can even use the connectors for your ICSP - just build an adapter wire.

Three connectors is a bit bendy for bug-building -- it's possible, but would be akward. If you use perf board and make the difficult connections using wire wrap, you shouldn't need more than 4x7 holes, which could be shaved down to .5x.8" perf board. A bit bigger if you want single layer traces instead of wire, a bit smaller if you went double layer & SMT.

As for grounded pins, that's what I mean, but you might just solder it with a tiny loop of wire wrap. Cut it later if you don't need it. Just don't forget to set the weak pull-up for those pins.

Also, unless you've wired your plane in an odd way, the power rails of this circuit are pushing all the power to run the servos and Rx, sourced from the ESC (thankfully, not your motor). Make sure you've got a decent amount of metal connecting these or it'll act like a resistor (or worse, a fuse)
 
#11
Sorry, the circuit is here: http://www.bhabbott.net.nz/Reverser.html

I've got some circuit sheets and ferric chloride. I was going to try out the toner-resist method once I pare down the circuit.

My plane is wired in a strange way. The battery powers the ESC and the receiver directly via a bridged 2x3 pin header, but only signal goes from ESC to receiver.

I did it this way because of the swappable pods (glider/motor). When the glider pod is on, I can put 2 batteries into the 2x3 header--also needed for ballast. When the motor pod is on (inserting the motor pod plugs the rewired ESC connector in), the ESC plugs into the top set of pins and the battery into the bottom pair (one pin is removed from the bottom set of pins to eliminate confusion).

I was going to do a wire jumper in addition to the copper of the circuit board. So far, with all 6 servos going crazy, total amp draw from my 1S 400mAh 30C battery is less than 1.5A (HobbyKing HK-5330 servos and FrSky VD5M micro receiver). The XP7A ESC and AP05 5000KV power system is waiting for me to get to the post office to pick up so I don't know what the amp draw is there--up to 7A though.

Thanks for your help Crafty Dan. Your guidance is greatly appreciated.

Cheers!
Randall
 

Craftydan

Hostage Taker of Quads
Moderator
Mentor
#12
Randall,

No problem. I'm new to the hobby, but EE (and ironically programming airborn "radios") is my day job. I'm leaning a lot about RC A/C from you guys here, it's nice to be able to lend my nich.

you're trying to get an N channel RX to drive N+1 channels -- Of course this it's wired up diffrent.

Ah! you're only running 1S with no regulation. hmmm. I'd put a Thalium cap (~1uf) between Vdd & Vss, just in case there's too much back EMF from the ESC. Watch your polarity. You're probably OK, but it's insurance for very little weight.

If only signal is going to the ESC, then you *could* be down to one wire on that, but for sanity's sake run a ground through, just make sure the ESC has a seperate, healthy pair of leads back to your battery bus. Can't see the circuit until I get home, but if your trimmed PCB is bigger than .5x.8", you can do better w/ perfboard and wirewrap. hard to get this smaller.

you can cut corners on the rails, but unless the servo is also wired funny, make sure it's got a good trace for power. Since the ESC power isn't going through here, the extra splice of wire is probably overkill. Just put the connectors side-by-side on the board, make the trace between wide and after tinning the trace, add on a touch more solder to up the crossection.

BTW, a quick check on the spec shows the 675 is only safe down to -0.3v -- I point this out because if you hook your RX lead up backwards, -3.7v will probably fry it. Other than keying the lead simliar to what you did on the battery bus, I'm not sure there is a light way to avoid this.
 
#13
I'm not too worried about polarity. The ESC connector will be glued to the motor pod--and it only goes on the glider one way.

I think I actually have a Thalium cap about that size. Good idea.

To clarify, this is how I have the battery/ESC/RX wired...
elf-wiring-diagram.png

I'll have to figure out what my circuit board is going to look like.

Thanks for your help Crafty Dan.

Cheers!
Randall
 

Craftydan

Hostage Taker of Quads
Moderator
Mentor
#14
Ok, got a look at the circuit.

The good news, Bruce has more than done his homework. That's a solid design with protection built in. You could still put something together w/o protection that's a hair bit smaller but you run risks of frying your board for the cost of a couple of grams (yeah, you can still rip out the trim pot and LED w/ minimal consequence). You'll want to pester Bruce for a BOM to get the specific parts for his layout.

The bad news, that's a double layer SMT board. They're really hard to get alignment in a toner transfer. It can be done, but be ready for headache. If you want to go this route (looks to me like a good idea), there are some toner transfer papers designed to separate from the board with water (dry paper into printer, iron dry paper onto *CLEAN* board, wet the paper and it falls away). I've used this stuff before. There are other ideas for materials available. Whatever you use, be prepared to do multiple trials, and use multiple drilled registration marks. Also check out the HCL/H2O2 etchant recipies - FCL is a NASTY etchant.

BTW, tantalum caps (not thalium -- where'd I get that from?) are polarized -- if you ever use them, worry.
 
#15
Once the weather started to get warmer, I noticed a bunch of tail wagging on launch. The 3 strips of 1mm x 3mm CF I glued together (in a triangle) are just too flexible.

I opted for a 4.8mm CF tube I found at my LHS and made a new tail module. I also went for the full size (according to the Mini-ELF plans, anyway) 256mm horizontal and 158mm vertical stabilizers. No more slop.

Mini-ELF-v3-tails.jpg
The old tail is in front for comparison.

I went out this morning (at 5am) and tossed the 1m wing/new cruciform tail config around for an hour. It's predicted to be 38 Celsius today, so I wanted to get out before it got too warm.

Today's the first day I've come back from the field and not needed to adjust anything on the plane. It flies straight and level with no input, now I just need to cover the pod and develop my technique.

I called it quits after 5 successive launch and catch (without moving my feet) flights. I love my powered planes, but there's no greater feeling of peace for me than me, my DLG and an open field.

Cheers!
Randall
 
#17
how did you make the wings on both of them
Hi Paul.

The 800mm wing is 2 layers of 3mm depron with a 1mm x 3mm carbon fiber spar at 50% of the chord on the top layer. I cut out the template for the full wing, then used one half for the 100% layer and the other half cut to 50% of the chord.

The 1m wing is 3 layers of 3mm depron with a 1mm x 3mm carbon fiber spar at 50% of the chord in the middle layer. The top layer is 40% of the chord and the middle layer is 70% of the chord. I hollowed out the middle layer of depron in a ribbed-style to save some weight.

The layers are laminated together with 3M Super77 spray adhesive. The spar is glued in with Foam-Tac (Beacon). I used a piece of 1.5mm x 4.5mm CF strip supported with pieces of 1mm x 3mm CF strip to withstand launching. On the leading and trailing edges, I affixed a strip (cut in half-width) of packing tape. The dihedral is supported by 2 pieces of 3-ply 1mm plywood cut to a 20-degree angle on either side of the CF strips.

There are pieces of 1mm plywood on the bottom of the wing where the mount holes are and I used the tube from an old gel ink pen glued into the wing (supported by the plywood).

Cheers!
Randall
 
#18
Great job randall_l!

Is their any chance at all you could do a build log! I would love to build one of these. How did you do your original pod it looks great, and what are you using for your fuse?