Time to try a tri

[jhitesma]

Well, looks like swapping out motor #3 is the next step.

It's more flyable on stock CF 1.9.0 but still doing most of the same things. Takes a lot of forward pitch to level it out, horizon was able to compensate and make it almost flyable but it was pulsing like x0054 was experiencing on his that he traced back to vibration.

So tonight I'll have to cut off the shrink wrap, warm up the iron and try some of these other motors to find which ones are smoothest. And find out what size bearings these motors take and see if I don't have a few or can pick up a few replacements.

Note: With stock CF 1.9.0 and all the board rotations set to 0 the 3D view responds as expected. So I'm still wondering if there's something wrong with the v3 custom version since the yaw was backwards with that. But it does seem that vibrations may be the big culprit here.

 

[x0054]

Placing the AfroNaze on some foam has helped a lot. Most of the oscillations are gone. Give it a try, it could help. Though your instant pitch foreword is such a strange issue. Have you calibrated the ESC's? That's the only thing I can think off.

As for the David's Cleanflight vs normal, I also noticed a problem with orientation. Also, at least on my AfroNaze, it would not save the mode settings for some reason. I would setup the mode switches, save the settings, go fly, unplug the battery, plug it back in, and the mode settings would be wiped. I tried it at least 5 times. I though my afroNaze was toast, but then I noticed that all the other settings would get saved. So I flashed it with 1.10 Cleanflight, and everything is working fine, and mode settings are being saved now.

I didn't notice any difference in the flight performance between David's Cleanflight and regular, so I am going to stick with regular for now. Maybe they already merged the Yaw changes into the main branch, I haven't checked the commits.

I am still not happy with this solution. I paid good money for the integrated Naze, damnit, and I want to make it work

 

[x0054]

Oh, in my case I had to rotate the board 180 on pitch in Dave's copy, and 180 on yaw in with normal clmeanflight (my board is backwards). So, in your case, if the board is orientated correctly, you would want to do 180 on roll only in Dave's copy, I think.

 

[marloth]

What mode are you in?

Just a few questions that may prompt different thoughts from travel shooting perspective. What mode are you flying in? What I found was that in accro mode there is no drift once the copter is calibrated, but that in angle mode there was drift in pitch or roll to be corrected. Check in CLI by typing "get acc". This will list all parameters with acc in its name. The parameters acc_trim_pitch and acc_trim_roll should be zero or close to zero. If you have a slight drift in angle mode, these should be tweaked to compensate the drift. If somehow these values are very large, there may be excessive drift. This is true only for angle mode, or possibly also for horizon mode. In accro, there is supposed to be no drift.

You could adjust these drift values from your radio:disarm, throttle max, stick left to max and LED on naze will flash for every increment on acc_trim_roll (negative, and positive when stick right). Same for stick fore/aft (acc_trim_pitch). Throttle back to zero and values are saved. Arm and test. But first check the values in CF, as the amount of "drift" you have cannot be adjusted by these small corrections. You can set them to 0 by CLI command "set acc_trim_pitch = 0".

Another thought: on the sensors tab in CF, are both and only the gyroscope and accelerometer checked? Not sure what would happen if this is not the case.

Final thought: I once fried a standalone naze32 board. since then I was able to connect the board with CF, but the 3D image appeared to move very slow, like I saw in some parts of your video. The net result of my naze32 being fried was that I had no more power on my receiver, which is clearly not the case for you. In any case, some damage may exist on the sensors on your board that stuffs up the control laws. Easiest way to check is to try a new board, which will need some funding ...

Good luck with it.

 

[jhitesma]

Placing the AfroNaze on some foam has helped a lot. Most of the oscillations are gone. Give it a try, it could help. Though your instant pitch foreword is such a strange issue. Have you calibrated the ESC's? That's the only thing I can think off.

It is on some foam, interestingly enough it seems the foam in Switzerland is the same as the foam I get at the local craft shop so I haven't swapped nilsen's original foam for my own yet

It's actually instant pitch back. I have to give it almost 3/4 stick forward pitch to get it to start to fly. ESC's have been calibrated - I redid it again after reflashing to stock 1.9 CF just to be sure. Sticks all center at 1500 with full 1000-2000 range. Accels calibrated and level.

With stock CF it took a lot less forward pitch to get it to fly than it took with David's modified version - and it flew with all of the rotation settings at 0. But even in acro mode I still had to give it a lot of constant forward pitch to keep it level and I was getting the pulsing in altitude that you were - which is why I'm not giving that rough motor the stink eye. Didn't get a chance to swap it out last night - will try that next though.

 

[jhitesma]

Just a few questions that may prompt different thoughts from travel shooting perspective. What mode are you flying in? What I found was that in accro mode there is no drift once the copter is calibrated, but that in angle mode there was drift in pitch or roll to be corrected. Check in CLI by typing "get acc". This will list all parameters with acc in its name. The parameters acc_trim_pitch and acc_trim_roll should be zero or close to zero. If you have a slight drift in angle mode, these should be tweaked to compensate the drift. If somehow these values are very large, there may be excessive drift. This is true only for angle mode, or possibly also for horizon mode. In accro, there is supposed to be no drift.

I fly almost exclusively acro mode usually. All of the testing in the video was in acro mode. I did do a test with horizon off-camera and it actually helped, but I had almost no pitch control and it was more unstable than in acro mode.

Another thought: on the sensors tab in CF, are both and only the gyroscope and accelerometer checked? Not sure what would happen if this is not the case.

That just decides which are displayed in the graph on that tab. In acro mode only the gyros are used.

Final thought: I once fried a standalone naze32 board. since then I was able to connect the board with CF, but the 3D image appeared to move very slow, like I saw in some parts of your video. The net result of my naze32 being fried was that I had no more power on my receiver, which is clearly not the case for you. In any case, some damage may exist on the sensors on your board that stuffs up the control laws. Easiest way to check is to try a new board, which will need some funding ...

Good luck with it.

I've fried a number of Naze boards. In fact every Naze board I've owned I've fried so far (yet I've yet to fry another other boards...one of the big reasons I'm not a huge fan of the naze overall.) Admittedly one was fried before it was given to me

The other two both fried from using the serial port while it was connected to USB (stupid stupid design thanks to TC cloning a poorly designed Chinese board into a smaller form factor - the F1 chip can talk USB natively there's no need for the external serial to USB chip or to waste a serial port on USB...other than to keep backwards compatibility with a Chinese board almost no one has heard of that uses the USB pins to drive LED's because they apparently figured no one would actually want to use USB to connect to their board.)

Plus the Naze uses the Prolific Serial to USB chip which has horrible drivers that my computer absolutely hates. I've always had issues with prolific serial to usb converters. Which is what causes the slow response sometimes. Their windows drivers are just junk.

But I will take another close look at the board. I did have one Naze where a small ground trace burnt through and it acted goofy until I noticed it and repaired it with a small jumper wire: http://forum.flitetest.com/showthre...S-experiments)&p=191010&viewfull=1#post191010

Though that board was being way goofier and couldn't fly at all.


I'm really starting to suspect the big issue is that rough #3 motor now.

 

[jhitesma]

BTW - I really do appreciate all the feedback and ideas even if it sounds like I'm shooting a lot of them down! It's hard to express that in written text without just coming out and saying it sometimes so I thought I better mention it.

I've been fighting with this thing in private for over a month so most of the ideas that have been presented I've already thought of and tried but just didn't bother to outline. If I had written up everything I've tried this would have been a three page thread before anyone else could reply

But that doesn't mean I don't appreciate the suggestions even if they're things I've tried already since when you get stuck like this sometimes just hearing your own ideas repeated back at you can give you a fresh perspective that can lead to a breakthrough.

Now I just have to not be lazy and get that motor swapped

 

[nilsen]

BTW - I really do appreciate all the feedback and ideas even if it sounds like I'm shooting a lot of them down! It's hard to express that in written text without just coming out and saying it sometimes so I thought I better mention it.

I've been fighting with this thing in private for over a month so most of the ideas that have been presented I've already thought of and tried but just didn't bother to outline. If I had written up everything I've tried this would have been a three page thread before anyone else could reply

But that doesn't mean I don't appreciate the suggestions even if they're things I've tried already since when you get stuck like this sometimes just hearing your own ideas repeated back at you can give you a fresh perspective that can lead to a breakthrough.

Now I just have to not be lazy and get that motor swapped

Ah no worries man, I hope you get it flying better than I

I'm fairly certain that Naze is fine, I really was flying it until I crashed it due to bad tuning and horrible vibrations.

Is the motor screwed or just needs some TLC? They served me really well, despite being horribly balanced they weren't all bad.

 

[x0054]

Have you tried running it on a level surface, without props on, while connected to the computer? This way you can see what's going on as you throttle up? Maybe you'll see the 3D model tilt as you throttle up, due to motor vibrations, or voltage drop, or something like that. Just a thought, feel free to dismiss if you already tried it

I had this exact thing happen on a quadcopter with TauLabs code, but in that case the PIDs where just way, way off, specifically the I value. If you are using David's PIDs, they should be reasonably close.

Any way, hope you get it flying, this tricopter is really fun to fly. I think I am pretty solid on what my problem is, I just need now to figure out how to solve it. This frame appears to be very susceptible to vibrations. I think it's the light weight of the frame combined with it's size. Those motors have lot's of leverage with those long arms, and not that much mass to move. In a normal frame the battery would eat up a lot of the vibrations, but in this one the battery is not directly connected to the main frame, so the vibrations are more pronounced, and cause more problems. It's a tradeoff that I am sure works really well when it's perfectly tuned.

Hey, maybe that's something for you to try as well. Strap the battery directly to the frame, not the battery tray, and see if it changes or improves things. I am going to try that tomorrow, just to see what happens.

 

[jhitesma]

Is the motor screwed or just needs some TLC? They served me really well, despite being horribly balanced they weren't all bad.

Motor isn't bad...but it's definitely in worse condition than the other two. It's really noisy and if I watch the gyros in CF while powering the motors individually there are a lot more vibrations from that one than the other two. Since you sent the other 4 though I'll just try swapping one of them next - worst case I toss this one one a plane

 

[jhitesma]

Have you tried running it on a level surface, without props on, while connected to the computer? This way you can see what's going on as you throttle up? Maybe you'll see the 3D model tilt as you throttle up, due to motor vibrations, or voltage drop, or something like that. Just a thought, feel free to dismiss if you already tried it

I did try that and it doesn't look like anything unusual is happening. However with DW's version of CF if I held it over my head and tried to tilt it forward the tail motor would kick in instead of the two front motors - and the reverse if I tilted it backwards. Even though the 3D view was reacting correctly. I haven't repeated that test with the stock CF yet, but since it behaves better with stock CF I'm really suspecting the rough motor.

Any way, hope you get it flying, this tricopter is really fun to fly. I think I am pretty solid on what my problem is, I just need now to figure out how to solve it. This frame appears to be very susceptible to vibrations. I think it's the light weight of the frame combined with it's size. Those motors have lot's of leverage with those long arms, and not that much mass to move. In a normal frame the battery would eat up a lot of the vibrations, but in this one the battery is not directly connected to the main frame, so the vibrations are more pronounced, and cause more problems. It's a tradeoff that I am sure works really well when it's perfectly tuned.

Yeah, this is only my second time dealing with lower kv motors like this. The first time was with brand new sunnysky's on a 550 sized setup a helped a friend with and they were silky smooth so we didn't have any issues. I'm really thinking it's the rough #3 motor causing the problems. Nilsen's last crash must have been a good one I figure

Hey, maybe that's something for you to try as well. Strap the battery directly to the frame, not the battery tray, and see if it changes or improves things. I am going to try that tomorrow, just to see what happens.

Actually I don't have the battery tray and have just been strapping the pack to the frame. The tray was damaged in nilsen's crashes so he didn't send it along. I've been thinking about making a new one though and moving the pack down to there.

 

[jhitesma]

Well, last night I finally fired up the soldering iron and swapped motor #3. Once it was off I took a look at the bearings and the top one is REALLY rough. [Side note: When I don't have a spare straight shaft of the appropriate size I like to test bearings by taking the bell of the motor off, then putting it on backwards. So only the shaft goes into one bearing and the magnets aren't making contact. Lets you get a good feel for just one bearing at a time. Won't work on all motor designs but worked on these.)

Looks like they're just 3x8x3 bearings: http://www.avidrc.com/product/1/bearings/227/3x8x3-Metal-MR83ZZ-bearings.html Not bad at only $1 each so I'll probably order some at some point. I noticed that most of the spare motors feel a little rough as well but picked the smoothest one and put it on.

Haven't had a chance to try flying it yet...but it feels a lot better and I'm seeing far less vibration in CF when testing the motors.

I'm also thinking about swapping out the collets for bolt on prop adapters since these motors have 4 threaded holes on top...but I'm not sure where to find properly sized bolt on adapters. The holes seem to be on 12mm centers. Anyone have any sources? I see that getfpv has a bunch of adapters but didn't see any that looked like the right size and most don't say what size only what motor they fit..but there's got to be some out there that will fit these. These are the motors: http://store.jdrones.com/AC2830_358_Motors_p/ac2830358.htm

jdrones calls them 2830's but they're measuring the bell. Measuring the stator like most companies now measure motors they're 2212's

Hopefully tomorrow I'll get to give it a test flight now that it's considerably smoother.

 

[x0054]

Those collets, at least in my case, are in part the cause of all the problems I have been having with my build as well. I noticed yesterday that even though the props are centered on the shaft, they are not perpendicular to the rotation of the motor. Meaning, as I spin the prop, one blade passes a few mm higher and the other a few mm lower through the same space.

As a result I am also looking for prop adopters for these motors (http://rctimer.com/product-130.html). Just in case some kind sole has a lead for me as well

I think you are going to have much better luck tomorrow. At least in my case the vibration plays a huge role. Also, when you are testing tomorrow, try setting gyro_lpf to 10 in CF. It's not a permanent solution, because the tri will fly really sluggishly with with gyro_lpf set so low, but if your problems are due to vibrations, it will make it at least flyable. That's what happened in my case, at least.

Good luck.

 

[cranialrectosis]

Been out a few and WHOA, Jason with a tricopter!!!!!

And then DAVID! Holy cow. It's like Edison resurrected to help Tim the Tool Man Taylor with a light switch.


I had two tricopters and found that Naze32 Baseflight SUCKs with a tricopter. I never got that to work.

Cleanflight SUCKS with a tricopter until you find LuxFloat. Dial the yaw P gain WAY back and drop yaw D to 0.

The issue with tricopters on a Naze seems to be wobble. Bad enough and it will instaflip you. Gains are the way I have been able to counter that assuming the motors, servo and tail pivot are in good shape. I can't tell you why LuxFloat worked for me, but my Twitchity Tricopter was unflyable until FGA pointed it out and Narco published some PIDs.

I never had any issues with my BatBone and a KK2. That was just easy.

I also note that with the Naze I really feel it when I strip a gear on that servo. The wobble is present every time you move past that point.



And you thought that hex was a challenge.

 

[x0054]

Hey, can you post your PIDs for LuxFloat on the Twitchity Tricopter. I have a 270mm Tricopter, that flies really well with APM, but I really wanted to try Naze on it, because my APM board is a bit broken, has intermittent gyro issues.

Thanks!

 

[Cyberdactyl]

Yea, my only experience with a Tri has been the KK2. Works like a charm. And with the newer Steveis >17.X it is close to perfect right off the bat.

I was hoping a new motor in that upper left position would make all the difference, but now with Cranial's comment, I'm not so sure. . .

 

[cranialrectosis]

I'm not by the Naze so I can't just pull them.

My starting PIDs were these:
http://forum.flitetest.com/showthread.php?16762-Cleanflight-What-s-your-favorite-PID-controller&p=188843&viewfull=1#post188843

More recent ones are on the front of my thread here:
http://forum.flitetest.com/showthread.php?14472-CranialRectosis-Twitchity-Mini-Tricopter-build

These should be what I am flying today. If not they will be pretty close.

 

[cranialrectosis]

Yea, my only experience with a Tri has been the KK2. Works like a charm. And with the newer Steveis >17.X it is close to perfect right off the bat.

I was hoping a new motor in that upper left position would make all the difference, but now with Cranial's comment, I'm not so sure. . .

At least on my little 220mm tricopter motor jitters, servo broken teeth, really bad props count and count big. Any little glitch on that tricopter matters MUCH more than my similar sized quads. Replacing that motor may turn the trick.

Quadcopters are incredibly more simple and IMO give you a lot more wiggle room on little glitches like balanced motors etc. My motors are little 2204s. 2212s will have a profound impact if they are out of balance.

I think vibe impacts the tail mount or the servo by shaking it, magnifying the wobble by 100 and causing the tail to do all sorts of crazy crapola.

That's my theory anyway!

 

[jhitesma]

Unfortunately I had to deal with a computer and a door lock issue today and didn't have time to try to fly over lunch Probably won't have time this evening either as it's book fair at my daughters school and I promised to take her. So it will probably be dark out by the time I'm back at the house. Winter means nicer weather finally...but a lot less daylight

Dying to give it a try with the new motor.

I've been sticking with David's suggested tune since other than a smaller battery the setup is pretty much what David suggests so I figured it would get me close. If I'm still having problems I'll give lux a try...but man I have had no luck at all tuning lux on my quads so far. Don't know buy but it just gives me fits.

I may try dropping the gyro_lpf as well...but since I really don't want to go lower than 42 I've been hesitant to even try lower values...though that is a good point that it will help confirm that vibration is the issue.

So far the only bolt on adapters I've found that might work are the ones RTFQ sells for their "red special" motors. But with no measurements given they're a gamble...and I really don't want to order from RTFQ again after my last experience. Still at $2.50 each it's not a big monetary risk to try them...

 

[jhitesma]

Quadcopters are incredibly more simple and IMO give you a lot more wiggle room on little glitches like balanced motors etc. My motors are little 2204s. 2212s will have a profound impact if they are out of balance.

I agree that quads are way simpler So glad I started with them and then tackled the hex before jumping into a Tri. The hex wasn't that bad other than being more connections to setup.

I would however say it's not so much the size of the motor as it is the speed of the motor and the frequency of the vibrations.

High KV motors have higher frequency vibrations which don't affect the gyros as much due to the sensitivity of the gyros. I have 2212's on my knuckle quad and even though they're beat up now and one or two have some bent shafts it still flies great and stable despite visible vibration in the motors. But those are 1400kv motors compared to the 850kv 2212's on this Tri.

That's part of why KK boards (especially the older ones) are easier to setup on the Tri's - they used analog gyros which were even less sensitive. Plus there's a difference in how analog vs. digital filtering works (that's a bit too much for me to go into right now) that effectively means analog filtering deals with frame vibrations easier than digital filtering. The MPU-6050 is a more sensitive gyro that can give a more locked in feel than the old analog gyros - but the tradeoff is that lower frequency vibrations can really mess with it.

I've been kind of tempted to put my 1400kv motors from my quad onto the Tri just to see how they do on there because of this.

Oh - also I may try reflashing the ESC's...they appear to just be AFRO 20A's that must have been rebadged by jdrones. The layout on them matches the afro 20 layout exactly. So they should run BL heli with oneshot and damped light quite well.

[Kind of funny side note. Even though TC resisted oneshot for so long and is still apparently convinced that it can't make a difference despite he did make a very good choice on the design of the Afro ESC's and put the input signal on a pin that supports hardware PWM - something almost no other atmel based ESC does. This is nice because it means they're one of the only atmel based ESC's that don't twitch at idle when running oneshot because the hardware PWM doesn't alias like the software does on the short 125us signals from oneshot.]