Twitchity quad with power board - test build

jhitesma

Some guy in the desert
Mentor
Ok. So it's a bit of an open secret that I've been working with Twitchity to design a power distribution board for his new quad design. The quad itself is still an experimental design and the PDB is even more experimental. So consider this a beta quality build :D

I had planned on doing a really nice build log with detailed photos...but...well. First I got impatient waiting on the first round of PDB's and put my hardware onto my ubmq experiment. So most of the fiddly work was done already by the time I got the frame on my worktable. And with no wiring harness to build most of the rest of the fiddly work is eliminated (I HATE building power harnesses - the PDB is SO MUCH nicer!)

I did however remember to setup my mobius for a timelapse. So here's a 30x sped up view of me building this while my Daughter napped this afternoon :D



Hardware for the build consists of:
Twitchity Quad frame - not sure if he's ready to offer these yet or not but either way they're not in stores, contact Twitchity for details if you're interested in one.
Jhitesma/Twitchity PDB board for Twitchity Quad - More on this below.
DYS 1806 motors I got mine for just under $12 each from banggood.com and they arrived in under 2 weeks...but one had grit in it and had to be cleaned extensively before it would turn. Just in case I ordered a spare from MultiRotorMinis who had in the meantime pricematched Banggood and being in the US got me my spare motor in just 3 days (Maybe 2...they're pretty centrally located being in Texas and ship same day if you get your order in early enough.) Banggood has dropped their price even lower, but despite their "quick for China" shipping I'd spend the extra $2 per motor to get them from MRM if I ordered more since the shipping is so much quicker and you're dealing with a US shop if something goes wrong like the one with grit in it.
BlueSeries 12a ESC's from RTFQ running Simonk 5-15-13 for mini motors. I used these because...well...because I got them free when my friend upgraded his warpquad to KISS esc's.
For the FC I'm using an afromini32 ordered direct from abusemark. Despite going with the cheap shipping Timecop still shipped same day and the mini arrived just 7 days later. For $20 this is a heck of a great FC. It's tiny, light, cheap and is actually more capable than an afro Naze but lacks the mag of the Funfly Naze. It is not however beginner friendly, it's tricky to wire, tricky to mount and has no built in USB so it's tricky to configure and tune as well. I love it, but it's definitely not for everyone.
For the RX I'm using a LemonRX DSM2 sat. For about $5 this is a wonderfully simple and light choice. I did have to spend a bit more though to get the 8ch RX and sat set since I didn't have any sat capable DSM2 RX's which is needed for binding. Now that I have it though I can do future builds with just a $5.50 sat.


Next post...the actual build begins!
 

x0054

Senior Member
Hey, if you do not mind me asking, what do you use to design your PCB boards and who do you use to print them?

I am also working on a quad design of my own, and I want to make a custom PDB for it. Ever since building my first mini quad a few weeks ago, I have been thinking about making a PDB like board, but one that would incorporate direct pin mounts for AfroMINI and MinimOSD, as well as space to double sticky type and directly connect FRSky DR4 receiver(or similar) and 4 ESCs. The idea would be to eliminate almost all of the wires, other then the once going to the motors and the battery.

Anyway, I designed things in AutoCAD before, and finishing up my design at the moment. But I never had to design a PCB board. I am learning, but if you have any tips on good software and service to use, I would appreciate it :)
 

jhitesma

Some guy in the desert
Mentor
So...the build. The quad came packaged in a nice padded envelope with all the pieces wrapped securely inside, remember this isn't retail or even really publicly released yet so no instructions and no frills here!
View attachment 29782

He had to tease me by putting the PDB deepest in the packaging forcing me to unwrap every other bit before I got down to it. Bad enough he taunted me with photos for a week this was just salt in my wounds even if he claims to have done it for "protection". Once finally unwrapped the parts look great:
View attachment 29781

The big thing Twitchity is trying to achieve with this design is to go for as low profile of a build as possible. Instead of mounting the FC on top of the quad he envisions mounting it even with the top plate. Since my plans were to use either the afromini or a homebrewed MW controller that didn't really help me much :D But it does explain a few things about the design like why there are 3 screws for each arm instead of two. Where the inner screw would normally go he had to make room for the FC. Hence the notches in the arms and the odd 3 screw arrangement on the arms. It sounds like this may change as fitting a FC in there is still tricky and he's not sure how well the arms will hold up - I hope to help answer that question :D

A little backstory here - Twitchity contacted me to see if I'd be interested in a free frame to document the build of. So - yes I am a bit biased here and did get this frame for free. The rest of the build was up to me though. I had mentioned some ideas for how he could make a PDB using his CNC machine since PCB material is G10 just like he's been using for his frames so far and once I agreed to build one of his frames we got to talking more about having a PDB made in a board shop. I got him to send me the design file for the lower plate in a format that enabled me to open it as a board outline in Eagle. Before he had his order of PCB material to experiment on I had already drawn up some PDB design ideas and uploaded them to oshpark to see if they looked like they could work. We both started to get excited at this point and within just a few days I uploaded this design and he ordered the first 3 boards. While we've been waiting on the boards I've been working on the next step and trying to route out a MW based controller and a power regulator similar to what MotoMoto has done for the WarpQuad. For now we just stuck a few LED's on the board just to test things out, so I choose ridiculously small LED's and resistors to test how well OSHpark could do on parts and traces of a size that would be necessary for a full flight controller. Besides, the bare boards cost the same no matter how complex the circuit is so it seemed wasteful to just put down power traces to feed the ESC's.

The controller is still a long ways off. In fact I hesitate to even mention it because I hate to get anyones hopes up about it. I'm facing some serious issues that are likely to keep it from ever being a viable product and at best a semi-viable DIY project. I'm sure there are ways to overcome them, but on my budget and with my limited free time not to mention my lack of any formal training in EE or board design...I'm not even going to guess how long it will take to overcome them. And at the rate things move in multirotors by the time I do work them out I'll probably be working on a design that's horribly outdated :D


One thing I should clarify. I'm going to be making a lot of comparisons to the 230mm warpquad because that and my ubmq are the only real mini's I've got to fly so far. My Syma x10 is about the same size...but it's a toy compared to these things. Even so I want to be careful because currently being made of G10 this is NOT meant to go head to head against the warp quad. The eventual CF version may be worth of going against the warp. But I did not enter into this build expecting warp quad level performance. Instead I was interested in seeing just how good of performance I could get for minimal cost. Mostly this is due to my own budget but partly it's since G10 isn't as stiff as CF. It's still a great material that I think is not currently given the credit it deserves in our community. I believe G10 is stiff enough to give very good performance - and the stiffness it lacks can be beneficial in crashes. And given how much cheaper G10 is than high quality CF I personally believe I would rather have a nice G10 quad than a cheap CF quad. So I'm very curious to see if my opinion of G10 is over inflated or not.


Anyway. Twitchity sent me a full frame and hardware set just in case the PDB was a total failure so I could still build it up with a Flip Pro I have on loan from a friend. (And I may still rebuild it with that just to show how it can be done at some point.) I wasn't sure just which pieces I'd use and didn't decide until I was deep into the build. So first thing I did was measure and weigh all the pieces:

The arms like the rest of the frame are made of G10. Twitchity plans on this quad being CF eventually but G10 is a lot cheaper to prototype with! He used 3mm G10 for the arms, I would have preferred to see 4mm or 5mm, but that would be heavier and I'm not sure it would be stronger enough to be worth the loss in stiffness for crashes. I think Twitchity made a good call with 3mm but I'd still like to try some 4mm arms just to see how they'd compare - I may change my mind after a few crashes or I may be begging Twitchity to send more arms. Time will tell :D
10614500_10152224572111805_404927757_o.jpg

Each arm weighs 10g. I was a little surprised by this, my wooden 1/2" poplar arms were a tiny bit lighter...but not much.
10642548_10152224568341805_1166617148_o.jpg

As expected four arms comes out at 40g. I wish I had a more accurate scale with a few decimals on grams. I wouldn't have been surprised if this was 1g heavier or lighter but it looks like they're actually pretty darn close to 10g per arm.
10627536_10152224571416805_1102986875_o.jpg

The solid base plate came in at 13g.
10621250_10152224570321805_897569657_o.jpg

I was surprised to see the PDB come in at a gram less than the solid plate. It is made of very slightly thinner G10 (despite me earlier saying it's the same really it's not.) It does however have a thin copper coating on each side - but the current circuit design removes most of the copper (my next version will leave most of it as ground and power busses but more on that later.) The copper is thin, but with it left on the board would probably be the same thickness and weight as the plain G10 board.
10565824_10152224568636805_108547947_o.jpg

The upper plate comes in at just 9g. If I can use it instead of the solid plate for the top I'll save 4g. (Spoiler: I'm not currently using this.)
10627460_10152224568891805_228947695_o.jpg

The protective cap is 4g which is the same as the very similar but slightly thicker (IIRC) CF cap for the WQ measured on my scale.
10642932_10152224569661805_1105818270_o.jpg

Bare frame with PDB and official top plate comes in at just 73g with hardware. Not too bad. Lighter than I expected honestly.
10563303_10152224589516805_1185300014_o.jpg

And tossing on all the bits I would be transferring from the ubmq shows the AUW should be in the sub 250g range I was hoping for.
10639028_10152224589271805_228526619_o.jpg



So starting the build I dug straight in on what I'm most excited about - the PDB I designed! The 0603 LED's and resistors were great for testing how OSHpark does on small traces and pads...but were not much fun to hand solder! Even with my magnifier these were a pain - I can't believe Twitchity did his unaided! I had ordered some 260ohm SMT resistors the day Twitchity ordered the boards...but they left LA on Aug 7th and haven't been heard from since :( I contacted the bay seller and they sent me a message saying they'd resend them - but then a few hours later refunded my money. Not sure if the replacements are coming or not at this point. Figured I'd just steal some off LED strips I have on hand...but turns out none of the strips I have use resistors this small or of the right value for the LED's I ordered. So I used the fullsize resistors Twitchity had tossed in that he used in his build.
10642993_10152224653731805_497631232_o.jpg

Since the boards are purple we wanted to go with purple LED's on these first builds. But purple LED's are hard to find. At least in small quantities and in this size. But it turns out they're used for modding xbox controllers so I was able to pick some up cheap on ebay. I'm not impressed by them though. I don't have enough specs on them to know if the resistor value is way off for them or what but I'm only getting a dull glow. Not nearly as bright as Twitchitys appeared in the photo he shared with me. I figure it could be the small size but this is the same size as the status LED on the motowii though and they're not nearly as bright as that. I may dig in my resistor collection and try some different values to drive them a bit harder and see if I can get them brighter. I've got enough for 2.5 more boards so if I burn them out it's no big deal.
10621391_10152224673756805_99778999_o.jpg


There are only 3 of these PDB's in the whole world at this time. I designed it and we had them made by oshpark. I shared the design so anyone who wants one could order them for themselves from OSHpark here: https://oshpark.com/shared_projects/2wou3kC8 - BUT...you have to order 3 for $45 and they're just bare boards. The only components on them are 100% optional and consist of 4 LED's and 2 resistors bot 0603 sized which is TINY, like smaller than a grain of rice tiny. And I messed up and used too small of a hole for the battery wires so you have to get creative attaching your battery wire. Bottom line...don't order them. This was an alpha test mainly to see what the quality from OSHpark was and to confirm that we could make a shaped board with the holes and cutouts positioned accurately enough to be used as a structural member of the frame. In that regard they're a HUGE success. The boards are 2mm G10 just like Twitchity is using for the plates in the standard frame design and we confirmed that the machining while not 100% perfect is accurate enough for this to work.

I'm going to make a revision of this board this week that will fix some issues with this prototype. Like the power holes being way too small, and some improvements like a different pad size for the LED's and resistors so larger devices can be used. Eventually Twitchity and I hope to offer ready to use boards...but I plan on sharing any designs that can be produced by OSHpark on OSHpark for those who are up to building the board themselves and don't mind having to buy 3 of them :D

If anyone is interested I'll share more details about the board design process as well. Just let me know.
 

jhitesma

Some guy in the desert
Mentor
Not sure what happened to the first two attachments in that last post...guessing I hit a limit and the board didn't let me know :D

Anyway here they are again:

10631490_10152224517351805_1076979435_o.jpg

10614541_10152224519721805_144842303_o.jpg
 

jhitesma

Some guy in the desert
Mentor
That was fun to watch! I also like your chairs.

Those chairs...oh those chairs. We found them at the local Habitat for Humanity "ReStore" store a few years ago. We were having a party that night and one of my outside speakers died. So I rushed up to the store to buy some cheap replacement speakers and on my way to the electronics section saw those chairs for just $40 - but with a tag saying they were reserved for someone :( Found some cheap blown speakers for $5 (for outside I just take cheap speakers, remove the speakers and bolt in cheap boat speakers in the housings) and was looking at the chairs again thinking "May my wife would love these too bad they're taken". When a store clerk saw me looking at them and said "You know, that woman is always reserving things and never comes back for them. If you're interested in the chairs you can have them."

Quick text message with photo to the wife and they followed me home :D

But...they're not very well made. They had no screws or nails in them and were built just with wood dowels to hold them together but most joints weren't glued - only the dowels. And the construction is very primitive. I've made some modifications to them to help them last a bit better...but overall they look nice but aren't very good chairs so they don't get very much use - and when they do they get frequent repairs :(

But thanks for noticing :D
 

jhitesma

Some guy in the desert
Mentor
Hey, if you do not mind me asking, what do you use to design your PCB boards and who do you use to print them?

While you were posting I was working on a post saying if people are interested I'll share more details :D It's getting late and I'm not even going to finish the posts I had planned for tonight...but...

It's a mix of the free versions of Eagle and Diptrace that I'm using right now. I'm still looking for a better solution though. I really don't care for Eagle much even though it's become a go to standard in the DIY community. And their free, Hobbyist and even lower end commercial offerings are all too limited to do a board for a H quad or hex due to their size limitations.

Diptrace I like a lot better overall. But I've yet to get it's import to work at all. So I have to import Twitchitys files into Eagle and then run a special script in Eagle that writes a Diptrace file which I can then open in Diptrace to edit. This is one of the big issues holding back a hex PDB for Twitchity right now and something I'm still looking for a solution to. Haven't run out of ideas to pursue yet - but haven't found any ideas that work yet.

I am also working on a quad design of my own, and I want to make a custom PDB for it. Ever since building my first mini quad a few weeks ago, I have been thinking about making a PDB like board, but one that would incorporate direct pin mounts for AfroMINI and MinimOSD, as well as space to double sticky type and directly connect FRSky DR4 receiver(or similar) and 4 ESCs. The idea would be to eliminate almost all of the wires, other then the once going to the motors and the battery.

Not sure how realistic that is to do with the afromini due to it's pad layout. The M1 pin is a flat pad on the bottom of the board, it's not on the pin header holes. So you'd have to tweak the software to move M1 to a different pin or come up with a way to solder that flat pad. The actual circuit for a MW or Naze isn't that complex - but it's not exactly a beginner level project to try and route either.

Anyway, I designed things in AutoCAD before, and finishing up my design at the moment. But I never had to design a PCB board. I am learning, but if you have any tips on good software and service to use, I would appreciate it :)

Honestly PCB layout software is just different enough from everything else that it's going to be frustrating and difficult to learn no matter what your background :D I learned AutoCAD in the late 80's on an old original IBM PC and have played with various other CAD packages since. I spent a good portion of college in Photoshop and Illustrator (as well as Flash and Director - I majored in photo illustration but only because "multimedia" at the time was taught as interactive CD's at best [they were still teaching linked slide projectors and audio tapes!] and I was already building interactive websites in my sparetime and knew the CD format was dead.)

But the packages for laying out boards is just different enough from all of those that it's like starting from scratch. Knowledge of electronics is more helpful honestly since that's the kind of mindset the software is aimed at. I can share more later if you're interested.
 

jhitesma

Some guy in the desert
Mentor
More about the build to follow. But it's 12:30 already and I can't keep posting :D

I was going to save these but since I can't finish my intended posts here's videos of the maiden and second "flights". I didn't get video of the 3rd flight but it went much better since I set throttle expo down to 0 which finally got rid of the "fall off the cliff going off hover" problem I've been experiencing.

Maiden "flight" including glimpses of the recovery:

Second attempt:
 

jhitesma

Some guy in the desert
Mentor
Do you ever sleep? :)

I think I remember sleep. That was that wonderful thing I used to do before having a kid right?

Still have a stack of photos to post...but may not get to them until this evening. Going to go do some test flying shortly, have to help a friend setup a home gym afterwards, then have to do the weeks grocery shopping (I do the shopping and cooking due to my wife's schedule and because I'm the better cook) and then we've got a birthday party for my daughters best friend this afternoon. So I probably won't have time to post until this evening at best. We'll see.
 

Cyberdactyl

Misfit Multirotor Monkey
Entertaining thread so far. I have to admit I did a bit of fast forwarding during the roof climb and time lapse. :eek:

Hope you get it running good. I can see the potential ready to burst free. :)

Oh, if you weren't. . . wear some eye protection when you hold the quad's prop plane in line with your eyes and run the motors up, just sayin' :) :black_eyed:
 

jhitesma

Some guy in the desert
Mentor
Yeah, I pretty much always have sunglasses with shatter proof lenses on when I'm outside. Just part of living in AZ :D

Did do a few test flights this morning...but ran out of props distressingly fast. Though I had more of the HQ 6030's but it turns out while I have quite a few I don't have enough of the right rotation/color to make a full set :( Kept breaking orange R's on the wood quad and didn't realize I was out of them. So going to have to order some more since I don't want to mess with the gemfans I still have for tuning.

It was doing pretty well, will have video later. Handled some hard crashes into pavement and grass very well. Have a bit of a high frequency wobble though that doesn't seem to go away with higher or lower PID's so I'm worried it may be caused by the arms not being quite stiff enough. Need to do more experimenting. But have to clean mud out of a motor first and I'm down to the crappy gemfans so may not make much more progress until I can get more props in from gothelirc in a few days.

Will still have more photos/videos later today. Just had to get online for a few minutes to put the smack down on a hacker trying to launch SQL injection attacks against a site I built. Fun way to spend a Sunday....
 

Tritium

Amateur Extra Class K5TWM
Man, you are way ahead of me on this build. So far I have carefully sanded all the parts like the Warp Quad and prepared my esc's. I move quite a bit slower. I am a bit worried that the traces on the PDB are not big enough for higher current draw, but that may just be my tendency to overbuild things, time will tell ;). What battery are you running?

Thurmond
 

jhitesma

Some guy in the desert
Mentor
Man, you are way ahead of me on this build. So far I have carefully sanded all the parts like the Warp Quad and prepared my esc's. I move quite a bit slower. I am a bit worried that the traces on the PDB are not big enough for higher current draw, but that may just be my tendency to overbuild things, time will tell ;). What battery are you running?

I admit I rushed :D But I also had a head start since I already installed all the bits on the wood quad to test them out ;)

I admit I didn't bother with sanding all of the parts, Just the arms since I didn't want the heat shrink on the ESC's to split when I shrank it like it did on the wood booms I didn't sand :D Since Fiberglass is non-conductive I'm not as worried about it.

I used an on-line calculator to calculate the trace size for 20amp capacity based on the specs of the copper layer provided by OSHpark - at least the motor power traces. Then I rounded up just to be safe. So they should be ok. At least so far they seem to not be an issue, but I plan on trying to get some temp readings off them after flying once I get it tuned well enough to fly through a pack :) Though thinking about it again now I may be off on that number. I may have rated it higher since some parts of the traces are connected to more than one motor. I do know I put quite a bit of thought into making sure they would be heavy enough for the power demands of 2204 motors. The next revision I plan on doing shorter traces and using large copper pours for the majority of the power buss. I had planned on doing it for this one as well, but we were really itching to do a test order and I wasn't sure how to do polygon pours in diptrace so I put that off for rev 2. I'm hoping to have a rev2 version of the board ready to order some tests of by next weekend...but we'll see how this week goes.

I'm using the 1.3 Turnigy 30c and 1.3 45-90c Nanotechs right now. To be honest though I haven't stuck the watt meter on it yet to see what the full draw is. The warp flew OK on the 1.3 30's but definitely was pushing them to their limits at full throttle. So I figure on the smaller motors they should be safe on this. I'll try and get some watt meter readings later today just to be sure.

Hoping my wife take my daughter to this party solo so I can try and do a bit of video editing on the limited testing I was able to do this morning. It looks better on video than it did in person. The high frequency oscilation I was getting isn't really visible at all on video.


What motors are you going to be using on your build Tritium?
 

jhitesma

Some guy in the desert
Mentor
Hmm...you know...looking at things again I may have made a big mistake there Tritium....it looks like I may have gotten mill's and mm's mixed up and the traces may be narrower than I meant to make them....So far I haven't burned one out...but until I can investigate further I'm going to suggest keeping a close watch on them. Need to check a few things that aren't handy to check right now to be sure...
 

jhitesma

Some guy in the desert
Mentor
So....Instead of finishing with the material I wanted to post last night I made a bit more this afternoon investigating some motor issues, think I found the source of my high frequency oscillation:


After recording that I saw that harbor freight had dial indicators on sale this weekend and were open for another hour. Ran up and grabbed one - sure enough I've got .100" runout on motor #4. #3 has .010" and the other two have none that I could detect.
 

Tritium

Amateur Extra Class K5TWM
Sorry been off line a few hours now. Lightning got my 2 meter J-Pole and took out my DSL modem, vaporized my router and got my motherboard integrated network interface on my PC. I have partial connectivity now but no wireless till Thursday till I can get another router. I thought the traces looked pretty narrow and mils / mm is definitely very different measurements. I assume your running 3S ?

Thurmond
 

jhitesma

Some guy in the desert
Mentor
Ouch, love my J-pole's (made out out of 1/2" copper and another out of 3/4" used one for APRS and the other for voice for years. Still haven't decided where to mount them in the new house yet...7 years later :black_eyed: )

Yeah, mils / mm are a lot different - but when you're doing your development late at night those M's can look awfully similar. I am running 3s So far no major issues, And I didn't trust the calculations at the time either and that was why I made them wider than it seemed to call for. But they're still not wide enough at some points. Not overly concerned since you me and Twitchity are the only guys with these right now and the next version I'll be using much heavier traces along with a few other changes.

Oh - and I rushed that last post...messed up my measurements again (I keep telling people I'm not good with numbers but they never seem to believe me since I'm a "computer guy".) The bad motor is .010" out and the #3 is only about .003" or so The others are within .001"
 

jhitesma

Some guy in the desert
Mentor
Thanks Mustang, I forgot to mention that I did measure that while I had the dial indicator setup and it sure looked like M4. Now to find someone in town with some M4 nylocs :D

Also, just found something interesting abou the DYS 1804 motors I'd like to share. I started swapping the one I got from MRM for the noisy one with a full .070" of runout (once it was off the quad I was able to mount it in my vice and get a much more accurate reading of it.

Since I had the indicator handy I figured I'd take a look at the new motor before mounting it up. The bell is very accurate and has less than .001" of runout - probably less than .0005" from the looks of it, that needle barely moves and I suspect most of the movement I'm seeing is from my indicator base not having a good metal plate to mount to so it's not as rock steady as I'd like.

But when I put the prop adapter on and measured it...I found I was getting .006" of runout on it right out of the box. I was seeing less than that on some of my other motors so I was kind of surprised. What I did was loosen the screws and then carefully snug them down again to try and keep from torquing any one too much faster than the others. But again I got .006" of runout. Hmm.....on a whim I took out all four screws and turned the adapter 90 degrees. Carefully torqued the screws again...and now I'm getting just under .001" of runout like the other motors.

For a <$12 motor this is hardly a complaint in case anyone is reading it as such. .005" isn't much to quibble over. But on these mini's every little bit can help so if you're looking to build the best build you can a $15 dial indicator and $10 stand for it may be a wise investment! Only takes a few extra minutes to measure the shaft and adjust to get as little runout as possible and should reward you with slightly smoother motors if you're feeling particularly OCD ;)

On the noisy motor I'm going to replace the bearings, for sure, but I may try rotating the adapter on the other motor that's a tiny bit off to see if it helps any.