Neurofuzzy helicopter problems

In my last post, i had vibration issues, remember??

So, what did i do to resolve that??

- First, we design a better place for the sensor to be, it uses elastic bands, that improve a lot the readings, but we still had glitches on the angles.

- We change from DCM to Kalman, because the performance of Kalman was better, and faster.

- We use a second filter, we implement a simple digital low-pass filter just after the analog readings, it help us to smooth the sensor jitter.


With this 3 upgrades, the angles were pretty acceptable even in max acceleration, ill post a video in a while.

Now i have a new problem...

I was using the training blades, that reduces power by 30%, and i was needing all the power, so, i change them to the good o normal blades, and i did a flight to test them, the results were bad, the helicopter was uncontrollable, very unstable and veryyyyyy difficult to hover.

I proceed to balance the blades, i follow the instructions from this video: https://www.youtube.com/watch?v=RMer3jIm9pA

We manage to balance "correctly" the blades, so i proceed to test again.

The same result....

And a curious data about this issue is that the motor and the battery were very hot after the test flights, veryyy hot, and thats very strange, it never happen with the training blades, i presume that the motor is requiring to much current, and the battery is getting hot because of that, so i think something is wrong with my heli, but i don't really know whats the problem...

I know that i have to make a tracking balance of my blades, but i don't really know how to do it...

Anyone has an idea??



In this picture we see a NI acquisition usb card, that helps us to find the spectrum of the frequency, for finding the correct parameters for the digital filter.

In control news, we are currently designing the neurofuzzy control, and it will take a couple of more days, its a easy task, because we have nice knowledge in that subject, ill post somethig about that later.

But im very stress because i cant make flight tests!!! hahahaha

Thanks!!

Cheers!!!
E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones

Comments

  • Developer
    Canopy will help only a small amount. You are learning thru experience! Any blade problems will upset some aspect of flight, tune in manual mode to achieve at least 1 second hands off hover.. Be sure fly bar is set up correctly too, balanced and both blades set to 0°. You could add extra weights next to paddles to help stabilize hover at the cost of high performance 3D flight. Good balanced carbon blades will translate to better smoother flight..
  • Sorry, i didnt saw the last post...

    Im only reading pitch and roll, yaw is for now out of the picture....
  • @Andrew Whipple

    Hi man!!

    10x for all the advice...

    1.- I just balance the blades, i just need to do the tracking, its my first time in both tasks.... hehehehe
    2.- ok, balance is done, tracking tomorrow...
    3.- my blade clamps are not to tight, but the blade is kind of strange, its enters the clamp with dificulty, it appears like the screw that holds the blade is too tight, but the reality is that the screw is very loose, perhaps a error in manufacturing... "you get what you pay for" hehehehe
    4.- KF and DCM both were tested on training blades, i will test both algorithms with the new blades, but i have to fix the problem first.
    5.- The only thing i notice with the electronics, is that my LiPo, fully charged, last only 11 min without electronics and with the electronics it last like 9-10 min. The problem is that i dont have too much money to invest i a better helicopter or parts, this helicopter was actually a gift, or like a scholarship, hehehehe....
    6.- I might think that the cables are creating a little bit of problem, and the extra weight (electronics, more cable, extra landing gear), ill try using the canopy... its been a lot of time since i dont use the canopy, because is easier to get access to electronics this way, but ill stop being lazy and i will put the canopy.

    10x a lot man!!

    Suerte para tu helicoptero!!

    Cheers!!!
  • Oh yeah, and how does ur heli manage without magnetometers? I always thought they were required for yaw correction.
  • Hi dronies! and Aldo Vargas,

    This is VERY important, I am not new to rc helis at all, even though your blades are brand new they are often not balanced. There is a lot of things that have to go right in order for a heli to fly with little or NO vibration. Usually as with most things the more you pay for a heli(or anything) the better they fly(or operate). On Youtube there are plenty of rc helis that are almost vibration free due to their expert setup. The several things you have to do are:

    1. MAKE sure that your blades are COMPLETELY balanced because often the blades you buy from a hobby store are not completely balanced (or you just got a bad pair despite what manufacturers do to prevent this). I can't tell so excuse me but for future reference if you accidentally removed some of the tape on one of the blades that try to balance them that screws you up! I add small pieces of electrical tape to counter-balance the blades.
    2. REMEMBER that tracking is only to be done AFTER your blades are balanced or else this step a few have mentioned is useless because the offset weight will have more inertia than your rotor head tracking can handle and correct for.
    3. When my heli has vibration usually the first thing I make sure of is that your blade clamps are screwed in good but not tight at all so the blades can move freely inside them. This was one of my many mistakes first getting into rc helis!
    4. ALSO this just popped into my head but if you switched the blades and the filter to the KF at the same time before testing the lighter same size blades should provide a faster response to the pitch angle you were looking at than the heavier same size blades creating the illusion that the DCM code was slower than the KF which should not be the case as I agree with William Premerlani on the comments on execution time.
    5. SOME ESCs that come with RTF rc helis are not good quality. These not bad but not great ESCs burn out faster than higher quality ones, again were back to the "you get what you pay for" slogan, BUT between blade changes and Moving of Electronics and what not your ESC could have built up static electricity and damaged some of the internals of it causing it to overheat your connected electronics by wasting energy stored in your LiPo (hopefully i love mine:) . I have a Belt CP V2 currently and it flies great but I've heard that the $400 ones with carbon fiber everywhere and CNC rotor head & ect. and i plan to upgrade several parts to CNC and carbon fiber to make it more durable and create smoother flight. I'm gonna use a similar setup with either the AruIMU or 9DoF Razor board! so i' excited!!
    6. I KNOW another one! but you of course want to make sure all you electronics are spread evenly and not creating havoc to your heli even though you heli LOOKS GREAT dude!!

    You can Google this stuff if you need tutorials on it and this will help you out a lot if you follow the tutorials directions right!

    Sincerely,
    Andrew Whipple
  • Admin
    Hi Vargas ,
    Try the tracking correction like Doug mentioned, yes you can setup the heli on table just tie it to some thing heavy and wide and test instead of trying to hold it ( dangerous ). Increase your pitch angle and reduce the throttle curve ( since you mentioned that you need to run at full power just to hover) , after clearing tracking issue if any. I liked your setup of vibration damping and clearly proves the fact that simple ideas work too. I am/will using RC car springs( instead of rubber bands as mine is nitro and so no good for rubber. Good luck,
  • Hi again Bill!!

    Do you want to take a look on my code for DCM?? and Kalman?? which one do you recommend more? Kalman is working better, but if i can save more processor by tuning DCM i think is better....

    Cheers!!
  • T3
    Hi Aldo,

    Thanks for the reply and the information.

    A linear Kalman filter should be ok if you keep the pitch and roll angles small. So, you do not need EKF. You would need EKF if you want to have everything work in inverted flight.

    I am puzzled about your comments about the slow response of your DCM implementation. In all of the firmware that our team has written (roll-pitch-yaw demo, MatrixNav, AileronAssist, MatrixPilot) the response of the DCM algorithm is nearly instantaneous. It is on the order of one or two calculation steps ( 0.025 to 0.050 seconds). It sounds to me that perhaps that your DCM implementation was not using the right gain factors in converting the gyro signals to an update matrix. So, what would happen is that it was the drift calculation that eventually drove the matrix to the correct values, which might take several seconds. I saw the same symptoms when a gyro gets disconnected (I had board with cold solder joints on the gyros).

    Though, I guess at this point it does not matter, if you have Kalman working ok, than that is great.

    Best regards,
    Bill
  • First of all, thanks to all!!! very good responses!!! 10x!!

    @William Premerlani:

    Plenty of your answers are in my first blog, have a look: http://diydrones.com/profiles/blogs/stabilization-of-rc-helicopter

    1.- Razor 6DOF, gryros: LPR530AL and LPR530AH; accel: ADXL335
    http://www.sparkfun.com/commerce/product_info.php?products_id=9431

    2.- Im using a Kalman filter, and a digital low pass filter.

    3.- I want to keep angles small, because i just want to stabilize the helicopter, so, my angles are between -30 --- +30, in pitch and roll. (i have a lot of problems with vibration, if you see my video, youll get an idea http://diydrones.com/video/instrumentation-with-elastic sorry, its in spanish... if i use a EKF could i get better results????)

    4.- When i was using DCM (i know there is less calculation with it) i got the angles like with a delay, like more "slow", when i switch to kalman, the angles change very fast, well... more faster than with DCM, if you want i can make a video showing the differences, let me know.

    @Morli

    Hi! the blades are the same size, just 5 grams less heavier...

    @Doug Weibel

    I cant hover the heli!!! its uncontrollable!! hehehehe, can i just place it in a table and make the tracking there? holding the helicopter...


    Thanks a lot to all!!!!
  • Developer
    Tracking the blades is very important and if they are not in track the heli will use more power (hot motor) and have bad vibration. It is not uncommon when changing blade types to have to reset the tracking.

    You can find many good references on the web on how to track the blades. A synopsis is:
    1 - Color the ends of the two blades with some magic marker, lighweight paint, or very thin tape so that you can tell one blade tip from the other.
    2 - rebalance the blades if necessary (paint, tape)
    3 - Hover the heli so that the blade disk is at eye level. If you are not a good pilot get a helper so you can fly and they can look closely.
    4 - If the blades are not tracking one tip will be higher than the other. Note which is higher and lower.
    5 - In the head linkages identifiy which links from the mixing arms you can use to increase or decrease the blade pitch. On the link affecting the blade that is higher adjust the ball link one turn in the direction to decrease the blade pitch. On the link affecting the low blade adjust the ball link one turn to increase the blade pitch.
    6 - Repeat steps 3 and 4. If the high blade is even higher you are adjusting the wrong way. If the high blade is still high, but there is less separation then repeat step 5. When you get to the point where repeating step 5 swaps which blade is running high then you can adjust just one ball link.

    You should be able to get the blades tracking so that the tips are at the same level within the thickness of a blade. Do not adjust a ball link a half turn as they are made to go on one way. Forcing them on backwards is bad.

    Good luck
This reply was deleted.