A new take on my UBMQ

[jhitesma]

If you don't have yours built by the time I finish the pod I'm going to have to send you some hardware with it to push you along

I JUST finished swapping the ESC's. My torture testing two Saturday's ago failed to break anything on the frame...but did manage to destroy one of the old "simon series" (even though it's a bastardized version of blheli on them) emax 12a ESC's. The only spares I had on hand were some KISS that I got free or cheap from people who had burned too many up and didn't want to use them anymore. 1 original ultra tiny pad 18a and 3 larger pad v1.1 18a are now in place. Kind of destroys the budget part of this build...but they were cheap since I got them used




Passed the smoke test, and as always with KISS esc's I love how much quieter the motors are now. Can't wait to get some props back on here and resume bashing


Progress on the pod continues, I've been uploading it to the git repo as I go since I'm working on it from 4 different computers I ended up going back to a larger less pointy and slightly taller top, but am keeping the full size that covers all four screws. Should make for a cleaner build and gives me more room in there for the VTX and cam.

I then took a break from the pod the past few nights to mockup a board cam. I know the HS1177's are popular right now but I don't have any and don't have the budget to buy any right now. I do however have several board cams on hand going unused. And while I was able to find a decent (but not well scaled) HS1177 model on thingiverse I haven't found a good board cam. So I drew up my own in openscad: https://github.com/jhitesma/openscad_board_cam (If you go into the generated files folder and click on the stl github will render it for you) I'm not quite ready to upload it to thingiverse yet...but it's more than good enough for what I need. Giving it extra details like the knurling on the lock ring and shape of the lens has been good practice in openscad.



I didn't bother to model the components and connectors on the bottom since they change from camera to camera...and doing so would have made this MUCH harder to print since there's be no flat surface to rest against the build plate. Playing with it I think I know how I'm going to make my motor mounts now too. They'll go on the sides of the cam and put the pivots in front of the board on the sides. Just trying to finalize how I want to design them since I'd like to make something that can mount and be adjusted without tools once it's screwed to the cam. Got some ideas on how to do it just have to test them out.

And...today I got a new treat for my printer - some Taulman 910 nylon filament I've been wanting to try printing with nylon since before I even got the printer and it was the main justification for the all metal hot end that can handle higher temps. It's expensive though...same price as high end PLA/ABS/PETG but comes on a roll half the size. However it should be even stronger, and it can be dyed once printed. So I may try reprinting the center and motor mounts in that once I get it dialed in...have to wait and see how well it prints for me first. Some people have horrible warping issues but I've heard the 910 is pretty easy to print with.

Given how bulletproof this quad has been so far though I'm really looking forward to getting the FPV pod built - should give me a nice FPV platform I'm not worried about bashing

 

[cranialrectosis]

Ha! KISS on the UBMQ...

No matter what you build with, jhitesma, it becomes a Cadillac on your bench.

 

[jhitesma]

Ha! KISS on the UBMQ...

No matter what you build with, jhitesma, it becomes a Cadillac on your bench.

Well...for certain values of "Cadillac" maybe:

( Google is reading my mind today...I only had to type in "joh" and that showed up as the first search results )

 

[jhitesma]

Well...she's dead again

Ran it through an autotune this morning and did not get the results I was expecting. Almost identical to what I had before. Which makes no sense. Going from those old "simon series" ESC's with no active breaking to KISS I should have seen a pretty major decrease in Tau and increase in gains. But instead I saw a barely noticeable difference and my hover throttle went from 24% to 44%. Plus it just felt "wrong" in the air.

But I was running some beat up bent up DAL tri blades...so swapped to a fresh set of the same props and re-ran the AT (always from defaults, never stack autotunes) Results were even worse, tau doubled - worst AT results I've ever had on anything

Swapped back to my good old FC 6045 props and ran it again...back to where I started a month ago with a 42ms tau. Still pretty lousy - not at all what I was expecting with KISS ESC's on there.

But applied it, bumped my rates and gave it a fly. And while the tune felt good the quad overall just felt WAY down on power and not right. Switching in and out of leveling mode it would suddenly lurch to one side. Acro felt ok though so flew it around a bit. Did a few flips. Then got some altitude to try a more aggressive set of flips - and it locked up full throttle while upside down flying into the street at full throttle from about 50' up

I've never bent the shafts on 4 motors at once before.



Somehow they still kind of mostly turn...but they're trashed.

I'm guessing the crash that killed an ESC two weeks ago also damaged at least one of the motors. So looks like it's time to get some new motors. The YKS took a pretty good beating but pavement at high speed was too much for them.

On the upside, the 3D printed frame shows no signs of damage at all. So at least this frame is proving to be darn near bulletproof at this point.

 

[LitterBug]

OUCH! Glad I have grass to "land" in.

Cheers!
LitterBug

 

[jhitesma]

Well contest results were announced today: http://www.instructables.com/contest/drones2016/

I actually found out I was a finalist last Friday but have been waiting to see the full results before I said anything. So I knew I won something just didn't know what. Turns out I got a shirt. Beats not winning anything and I can always use another shirt Sure would have loved a gopro or even had some fun messing with the RC bird...but can't really complain.

New motors are ordered on on their way, scheduled for delivery Saturday though I ordered some RS2205 red bottoms since they seemed like the best bang for the buck right now. And banggood has them in the US warehouse on sale for <$16 each shipped. They listed the free shipping as USPS first class 2-6 days and the motors as in stock ships in 24 hours. Usually I've had great luck with their US warehouse and get things just 2-3 days after ordering since they're here on the west coast. But...my luck ran out. I ordered motors Saturday but they didn't ship until yesterday and instead of USPS priority they're sending it fedex "ultra goofy slow hand off to USPS for the last mile" which I hate Fedex tracking shows the motors are here in town already but fedex being fedex I know the chances of them actually getting to me before Saturday are slim to none. (I've had fedex sit on things in town 4 days because they got here before the expected delivery date.)

Oh well, hopefully I'll have time this weekend to get them installed.

And guess I need to get back to finalizing the FPV pod now too!

 

[Snarls]

Congrats on winning something. Unfortunately my gimbal did not make it as a finalist. I suspect I submitted to the contest too late and did not get enough votes. Some projects submitted after mine got featured in the email newsletter which I assume boosted their views and votes so I am little bitter than mine got left behind. Still can't complain about a free contest. A did get my instructable featured so at least I get three months of premium membership for free.

Can't wait to see you back up and flying with some redbottoms. I need to see more flight videos!

 

[jhitesma]

Slow but steadily progress continues. While waiting on new motors I continued working on a design for a FPV pod:



Third version was very similar to the 2nd, but 4th was a bit of a departure. I had been designing it around a HS1177 or one of the new runcam cam's...but just can't justify buying one right now since I still have a few unused board cams laying around. So I designed a carrier for the board cam and started designing around that instead. So things had to get a bit bigger.

But I finally got some actual mounts added inside that a camera can mount into and have finally made a version that can hold a camera successfully:



I also offset the top of the pod a bit to get the camera a better view. Went a bit too far though...didn't account for how the camera mounts quite right and it sticks out more than I planned. Still I'd fly this except there's still not VTX mount or antenna mounting point.



So I need to deal with that. And I'm a bit torn. Because getting in there to change VTX channels is NOT going to be easy. This latest design eliminates the CNC cut top-plate from my original design and just uses the POD for strength on the top. Which hopefully will be ok.

You may also notice there are some new motors on there

I sprung for a set of RS2205 red bottoms for it - bit nicer than I had expected to be flying! But seemed to be the best bang for the buck so why not. Only now with these hot motors I was worried about the 18a KISS not being able to keep up with some of the props I want to try and with wooden arms risking fire seems like a bad idea. Thankfully my health insurance decided to send me a $50 amazon gift card for being a non-smoker (about the only halfway decent thing they've done in years) so I picked up a set of littlebee 30a blheli_s ESC's. Which may not have been my best decision.

I almost went with 20A or 25A ESC's because I knew the 30a's would be a tight fit. But I really wanted the safety since I've seen these motors documented as pulling 27a with some props. Except my fears were very well founded.



That is a VERY tight fit. I might be able to make it work...but I'm thinking about maybe mounting them on the sides of the arms now instead. I really don't like that solution...but it's going to be super tight fitting them on top. So it's been sitting here for a week waiting on me to make a decision on how to mount these I could make the arms a little longer....but I like how compact this is. Just don't know. Need to make a decision soon I'm itching to fly this guy now that it's got an upgraded power package

 

[LitterBug]

Glad I don't have a 3D printer because I build too much as it is. Have you done anything with the Lunar Lander lately?

Cheers!
LitterBug

 

[jhitesma]

No, I keep meaning to but just haven't got to it. I have the new shroud printed and ready to go but haven't even installed it yet

And this week I'm not going to get much done since my wife is going to KY for a school reunion and I get to play single parent.

Really wish I had two of me so I could get more stuff done

 

[jhitesma]

So this project isn't dead....but I've been swamped with other stuff and haven't had much time to work on it. Plus I've been debating how to deal with fitting the 30a esc's on it.

Finally sucked it up last night and decided to mount them on the sides of the booms, actually worked kind of well. I can almost claim I did it that way by choice - makes for super clean power wiring with the pad layout on the DTFc and has the ESC's on the side away from prop rotation so they're theoretically a bit more protected from prop strikes. See, I did that on purpose...really. Not just because I didn't want to cut the motor wires because these motors are probably going to go on a better build soon



I also finally picked up some TPU for my 3D printer and tried printing the pod in TPU. Came out pretty good. I don't have TPU 100% dialed in yet, it doesn't retract well, but strings a lot so it's kind of tricky to get clean prints with it. Still the pod came out fairly decently:



And it's much stronger than I expected. I can jump up and down on it and it will squish - but bounce right back. I think nylon still wins out over TPU for tensile strength but it's pretty close. Nylon is stiffer though. However for TPU to work I'll have to completely redesign the pod due to a few issues.

First, the nut traps that hold it onto the frame are useless in TPU, the nuts just spin even though they're a tight fit since the TPU just squishes out of their way. Second the bottom "plate" of the TPU isn't stiff enough to replace the top lexan plate so I'd have to keep that in there. Third the cam mounts are too flexy - while it does make it much easier to install the camera it doesn't hold it stable enough and the camera can fall out. Fixing the camera mounts would be easy - I could just tweak the design to tie them into the shell of the pod. But fixing the nut trap and base plate issues...not so easy. Base plate I could work around...but the nut traps...without them there's really no way to assemble this.

So for now I'm going to stick with the PETG pod since it seems like the best so far. Just need to figure out how to fit a VTX in there now...and really the big problem there is how to do it in a way that will still give me a chance of changing channels without spending an hour taking it apart and putting it back together again

But, for now it's back together as an LOS acro quad though upgraded from flysky PPM to a spectrum sat ($5 banggood clone version at least) for slightly less latency in the control.

Tested the new experimental improved dRonin autotune on it just now and it feels great. The blheli_s esc's and RS2205's are a big improvement over the old "simon series" emax and YKS motors - but I knew they would be....just about anything would be My tau dropped from 45ms down to 22ms and it's feeling great in the air. Kind of neat to put it into leveling mode and smack my sticks then watch it "oscillate" as the sticks settle down on their springs At first I thought it was a bad tune but then I realized it was just that it's locked in enough to track the tiny vibrations of the sticks self-centering. Look forward to giving it a good test thrash tomorrow.

Oh, and the new dRonin autotune? It's not a huge change. It's actually a lot milder while it does it's wiggle dance. But the way it collects data has changed under the hood and it should give more consistent results and should make a lot of the setups that failed on autotune before work now. Plus the new improved system identification it builds is laying the groundwork for a next generation non PID based control scheme...but that's still a few releases away - maybe next year sometime if all goes smooth

Only downside to the new autotune - won't work on F1 boards (naze32 and CC3D) they just don't have the power to handle the new stuff. But for F3/F4 based boards this is looking to be a nice improvement!

 

[jhitesma]

Well the new motors and ESC's are great. Easily the nicest powertrain I've run. Ran the new autotune to test it on this and my tau went from about 45 on the old setup down to 22 on DAL j5045 props. That's not bad at all. If I ever stepped up to 4s this would probably be a 12-15ms tau machine despite it's weight.

I sent my test data from the new autotune into icee and he ran some analysis on it:
http://nbviewer.jupyter.org/url/jar.lyle.org/~mlyle/angle.ipynb

That's probably a bit tricky for most people to follow...but basically it's walking through the analysis that the GCS usually does to prepare for calculating PID's. That doesn't include the actually calculation of PID's, just calculation of the parameters that are used to calculate PIDs. The new improved autotune code records a lot more data about what happens during the wiggles so it can be analysed more carefully, it also makes some minor changes in how the wiggles happen and some major changes in how the data from the wiggles is recorded. Overall these changes should make autotune in the next release of dRonin more reliable and capable of tuning setups it failed on before as well as not falling into the "too low of tau" failures that it sometimes hit before on setups that were initially overtuned.

What was interesting is that in analysing the data there's significant difference in tau between Roll and Pitch on this setup for some reason. Which doesn't make a lot of sense since the quad is symmetrical! There is some cross-axis coupling because of the angled arms but that shouldn't cause roll and pitch to be so different from each other. Really not sure what's causing that at this point. My best guess is it's due to the battery being comparatively long and skinny so affecting balance in one dimension more than the other - but that seems unlikely.

Anyway, today I did a few more tests of the latest changes to the new acrodyne flight mode. For those who haven't heard me talk about this before it's a new mode I'm really excited about. Basically it's acro mode with dynamic rates. So when you cross a certain threshold and go to higher stick positions your rate will start to increase. I'm flying with 830/35/140 which means my maximum rate is 830 degrees per second but around center stick they're only 140 degrees per second, the 35 sets the shape of the expo slope. This gives me nice tame center stick controls but the ability to easily do flips at full stick and fairly easily do double flips.

At least in acro mode that's what it means When I switch over to acrodyne that all still applies but 830 is no longer the max, instead the max will climb as I stay at high stick rates. With those same expo settings I was able to do 3 and 4 rotation flips as easily as I could do doubles in regular acro mode. And I could fairly easily do crazy 5 to 6 (it gets hard to count) rotation flips and rolls That was fun.

Unfortunately it was too much fun I had to see how many rotations I could do while below the roof line. 5 and 6 rotation flips below the roof line are fun but risky! Managed to slam it into the ground again near the end of my pack and...well...the UBMQ has seen better days:



Ouch. It's hard to see in the photos but the whole thing is cupped up in the center.


This is the first time I've managed to break an arm on this setup. On the original UBMQ without the 3D printed center I broke arms almost every crash. So the 3D printed bit does seem to add quite a bit of strength. And it looks like it didn't actually break this time either...even though the top lexan did and something is pretty bashed up because the 10 degree tilt is pretty much gone

Thankfully the new motors seem ok. One is a little less easy to turn and has a bit of a squeak so I may have trashed a bearing But all of them still seem to be straight with no noticeable wobble to them. Impressed with how tough these RS2205's are. That was a hard full throttle slam straight into the ground - the same kind that bent all 4 YKS motors just a few months ago (Though thankfully this time I hit soft dirt and not hard pavement!)

Not sure if I'm going to rebuild right now or not. I still want to do the pod for this at some point...but just not feeling it right now. Instead I may move these bits over to my old nighthawk 250 frame that's hanging naked on the wall because I've got some other stuff I want to test out and a frame with a bit more room would be nice for those tests....