FunJet (Delta Wing) asymmetric elevon response

[Phil Giacalone]

Hi,

I'm setting up a FunJet (Delta Wing) and have everything just about
ready to go for a T3 run. However, during ground tests in stabilized
mode I'm seeing a slightly asymmetric response in the elevons.

During pure roll, one of the elevons moves up more than the other
moves down (~25% difference). During pure pitch, the elevons moves
than the other (~25% difference).

Is there something in the options.h file that can correct this?

Any fast help would be appreciated since the T3 competition deadline
is tonite.

Best regards,
Phil

[Philip Giacalone]

Here is my options.h file related to my prior post -- FunJet (Delta Wing) asymmetric elevon response

[ben levitt]

Just to check, do you have the same asymmetric response when in manual mode?

If so, you probably have some accidental physical asymmetry set up.  At rest, are your servo horns sticking out perpendicular to the line of flight?  Do your servos move the same amount forwards and back?

Ben

[Phil Giacalone]

Hi Ben,

Thanks for the fast response.

There's a slight asymmetry in manual mode, but it seems to be due to
flex in the push rods and some space in rod holes. The servos do not
appear to be binding. The asymmetry in manual mode looks much smaller
than the UDB asymmetry. Especially in roll.

Here's a bit more observation.

- Roll left, response is more right elevator up than left elevator
down.

- Roll right, response is more left elevator up than right elevator
down.

- Pitch up, response is more right elevator down than left elevator
down

- Pitch down, response is more left up than right elevator up

The roll asymmetry (both left and right) is much more obvious than in
pitch (both up and down).

Best regards,
Phil

On Dec 11, 10:48 am, ben levitt <ben...@gmail.com> wrote:
> Just to check, do you have the same asymmetric response when in manual mode?
>
> If so, you probably have some accidental physical asymmetry set up.  At
> rest, are your servo horns sticking out perpendicular to the line of
> flight?  Do your servos move the same amount forwards and back?
>
> Ben
>
> On Sun, Dec 11, 2011 at 10:39 AM, Philip Giacalone

> <scienceth...@gmail.com>wrote:

[William Premerlani]

Hi Phil,

There is nothing in the options.h file to adjust for elevon asymmetry. I am hoping that you can fix that mechanically.

Just a reminder on a separate issue....there are three bits of information that you need to set to get deflection directions correct for elevons. In addition to aileron and elevator channel reversing, you also have to set ELEVON_VTAIL_SURFACES_REVERSED.

The way that I suggest you set these three things up is to first set aileron and elevator reversing to get the elevons moving in the correct direction for pitch. (I can see from your post that you must have done that.) Then, set the surfaces reversed sign to get the correct sign for roll response.

Good luck to you, I hope that you get everything worked out in time for a T3 entry, but please take your time and be careful.

Best regards,

Bill 

[William Premerlani]

Hi Phil,

I will do a ground test with your options.h file and get back to you in about 60 minutes. In the meantime....

With respect to roll response, I think there might be some roll-elevator mixing going on, the response that you describe is what you would want to happen in a turn. You need a bit of up elevator when the aircraft is rolled. This is controlled by the parameter ROLL_ELEV_MIX, which I see that you have set to 0.5, so I think that the controls are doing what you want with respect to roll response. If you want to check that, set both RUDDER_ELEV_MIX and ROLL_ELEV_MIX to zero, and see what happens.

With respect to pitch response, I can not think of any reason why it would do that. But I note that you said that is not as pronounced.

Best regards,

Bill

[Riccardo Kuebler]

Hi Phil,

I don't understand if you mean when you move the sticks or when you check for UDB stabilization.

So in the first case you could check some few other things :

- do you have same throws programmed on your transmitter ? I mean both up and down the same amount and each channel the same

- and also, do you have any aileron differential set on a value other than 0 in your tx ?

In the second case I noticed that you have yaw ailerons stabilization enabled.

So if you try to roll or to pitch it could be that you move the plane on yaw axis also and this causes ailerons trying to stabilize the plane on yaw axis.

Best regards,

Ric

2011/12/11 William Premerlani <wprem...@gmail.com>

[William Premerlani]

Hi Phil,

I am doing some ground tests, and getting some results that I do not understand, so it may take me a while to sort this out. It may be that I have made some mistake in setting this up. But I do not have any experience with elevons, so it may simply be that I am not understanding what I am seeing.

However, I spotted something in your options.h file that I would like Ben to respond to, if he is listening, and has the time:

Ben, I notice that Phil has 4 input channels assigned (1, 3, 4, and 5), but has specified NUM_INPUTS 5. Is that correct, or should it be 4?

Finally, Phil, I do notice that you have several terms defined that mix into elevator or aileron, so it could be that what you are seeing is a result of your settings.

Best regards,

Bill

[Phil Giacalone]

Hi Bill,

I suspect you may be right about the ROLL_ELEV_MIX being the reason
for the asymmetry. Have to do some testing to find out...

Best regards,
Phil

On Dec 11, 11:29 am, William Premerlani <wpremerl...@gmail.com> wrote:
> Hi Phil,
>
> I will do a ground test with your options.h file and get back to you in
> about 60 minutes. In the meantime....
>
> With respect to roll response, I think there might be some roll-elevator
> mixing going on, the response that you describe is what you would want to
> happen in a turn. You need a bit of up elevator when the aircraft is
> rolled. This is controlled by the parameter ROLL_ELEV_MIX, which I see that
> you have set to 0.5, so I think that the controls are doing what you want
> with respect to roll response. If you want to check that, set both
> RUDDER_ELEV_MIX and ROLL_ELEV_MIX to zero, and see what happens.
>
> With respect to pitch response, I can not think of any reason why it would
> do that. But I note that you said that is not as pronounced.
>
> Best regards,
> Bill
>

[Phil Giacalone]

Hi all,

I'm having an unrelated problem and need some advice. My brand new
UDB4 is being stubborn during programming. The PICkit3 refuses to
connect is very high percentage of the time with this error:

Target Detected
Target Device ID (00000000) does not match expected Device
ID (07ff0000).

So I just keep trying and trying by turning the UDB off/on, disconnect/
reconnect PICkit, until it finally works. Frustrating. Any advice?

Best regards,
Phil

[William Premerlani]

Hi Phil,

I confirm with ground testing with your settings that there is asymmetric response to roll. That is what is expected from ROLL_ELEV_MIX, that is exactly what it is for, to produce some up elevator during a turn.

The only thing you need to figure out is what you want it set for.

Best regards,

Bill

[Phil Giacalone]

Hi Ric,

These tests are done in stabilized mode, letting the UDB control the
elevon movements. I'm trying my best to test in pure roll and pure
pitch. There may be some slight yaw movement, but it should be small.
That's another thing I can try, though -- turn off yaw aileron
stabilization.

At the moment, I can't get the board to be recognized by the PICkit3,
so I'm stuck.

Best regards,
Phil

On Dec 11, 11:40 am, Riccardo Kuebler <kueb...@ticino.com> wrote:
> Hi Phil,
>
> I don't understand if you mean when you move the sticks or when you check
> for UDB stabilization.
>
> So in the first case you could check some few other things :
> - do you have same throws programmed on your transmitter ? I mean both up
> and down the same amount and each channel the same
> - and also, do you have any aileron differential set on a value other than
> 0 in your tx ?
>
> In the second case I noticed that you have yaw ailerons stabilization
> enabled.
> So if you try to roll or to pitch it could be that you move the plane on
> yaw axis also and this causes ailerons trying to stabilize the plane on yaw
> axis.
>
> Best regards,
>
> Ric
>

> 2011/12/11 William Premerlani <wpremerl...@gmail.com>

>
>
>
>
>
>
>
> > Hi Phil,
>
> > I will do a ground test with your options.h file and get back to you in
> > about 60 minutes. In the meantime....
>
> > With respect to roll response, I think there might be some roll-elevator
> > mixing going on, the response that you describe is what you would want to
> > happen in a turn. You need a bit of up elevator when the aircraft is
> > rolled. This is controlled by the parameter ROLL_ELEV_MIX, which I see that
> > you have set to 0.5, so I think that the controls are doing what you want
> > with respect to roll response. If you want to check that, set both
> > RUDDER_ELEV_MIX and ROLL_ELEV_MIX to zero, and see what happens.
>
> > With respect to pitch response, I can not think of any reason why it would
> > do that. But I note that you said that is not as pronounced.
>
> > Best regards,
> > Bill
>

[William Premerlani]

Hi Phil,

Don't you just hate it when the alligators bother you when you are trying to drain the swamp?

It sounds like a loose connection between your PICkit3 and your UDB4. Are you using one of those Olimex cables? They are notorious for loose connections.

I make up my own cables for connecting between PICkit and UDB.

Also, you can connect the PICkit directly to the UDB without a cable.

Best regards,

Bill

[ben levitt]

On Sun, Dec 11, 2011 at 11:57 AM, William Premerlani <wprem...@gmail.com> wrote:

Ben, I notice that Phil has 4 input channels assigned (1, 3, 4, and 5), but has specified NUM_INPUTS 5. Is that correct, or should it be 4?

 

Yup, this is right.  NUM_INPUTS should be the same as the highest assigned input channel number.  And leaving one or more unassigned channels within that range is fine.  Similarly, I actually set up my tiny flying wing to use only output channels 6,7,8 (on a UDB4) since they were physically closer to the servos, and allowed me to avoid extension cables.  :)

Ben

[Riccardo Kuebler]

Hi Phil,

yes, reading again your first post I got it :), but in this case I don't understand.

Roll elevator mix should work only on radio or navigation input.

While banking the plane you should only get ailerons counteracting to regain stabilization, at least this is what I observed until now, am I wrong (Bill ?) ?

Ric

2011/12/11 William Premerlani <wprem...@gmail.com>

Hi Phil,

[ben levitt]

Another possibility is that you may need to reboot your computer.  I've needed to do that a few times, but I think it's because I'm running a virtual windows machine on my mac...

Ben

[William Premerlani]

Everyone,

There are two parameters in the options.h file that cause mixing into the elevator. They are:

RUDDER_ELEV_MIX

ROLL_ELEV_MIX

They are active in stabilized mode and waypoint mode. In manual mode, they are turned off. (In manual mode, all controls are turned off, everything is a pass-through).

ROLL_ELEV_MIX multiplies the square of the roll, and adds it into the elevator, in stabilized and waypoint mode. In the case of elevons, this causes an upward deflection of the elevons.

RUDDER_ELEV_MIX multiplies the actual rudder deflection times the roll, and addes it into the elevator, in stabilized and waypoint mode.

You can use either, neither, or both, depending on your philosophy of mixing, and the type of aircraft.

For elevons, only ROLL_ELEV_MIX has any impact, because there is not any rudder.

Best regards,

Bill

[Phil Giacalone]

Hi all,

First of all, thanks everyone for being there to help me out on such
short notice. I have just a few more hours of daylight and will be
trying a T3 run very soon.

The Olimex cable was the problem. Thank goodness that problem is
behind me.

I turned off ROLL_ELEV_MIX (= 0.0) and now the UDB control surface
movements ARE symmetrically in response to both roll and pitch
changes.

Now that I think I understand what's happening, I guess the question
is whether to actually fly with ROLL_ELEV_MIX = 0.0. My gut says no,
since the T3 race should involve some fairly steep turns, so elevator
mixing will hold altitude better. So better to set it back to
ROLL_ELEV_MIX = 0.50???

I'm interested in your thoughts, as always :-)

Best regards,
Phil

On Dec 11, 12:47 pm, William Premerlani <wpremerl...@gmail.com> wrote:
> Everyone,
>
> There are two parameters in the options.h file that cause mixing into the
> elevator. They are:
>
> RUDDER_ELEV_MIX
> ROLL_ELEV_MIX
>
> They are active in stabilized mode and waypoint mode. In manual mode, they
> are turned off. (In manual mode, all controls are turned off, everything is
> a pass-through).
>
> ROLL_ELEV_MIX multiplies the square of the roll, and adds it into the
> elevator, in stabilized and waypoint mode. In the case of elevons, this
> causes an upward deflection of the elevons.
>
> RUDDER_ELEV_MIX multiplies the actual rudder deflection times the roll, and
> addes it into the elevator, in stabilized and waypoint mode.
>
> You can use either, neither, or both, depending on your philosophy of
> mixing, and the type of aircraft.
>
> For elevons, only ROLL_ELEV_MIX has any impact, because there is not any
> rudder.
>
> Best regards,
> Bill
>

> On Sun, Dec 11, 2011 at 3:37 PM, Riccardo Kuebler <kueb...@ticino.com>wrote:
>
>
>
>
>
>
>
> > Hi Phil,
>
> > yes, reading again your first post I got it :), but in this case I don't
> > understand.
>
> > Roll elevator mix should work only on radio or navigation input.
> > While banking the plane you should only get ailerons counteracting to
> > regain stabilization, at least this is what I observed until now, am I
> > wrong (Bill ?) ?
>
> > Ric
>

> > 2011/12/11 William Premerlani <wpremerl...@gmail.com>

>
> >> Hi Phil,
>
> >> Don't you just hate it when the alligators bother you when you are trying
> >> to drain the swamp?
>
> >> It sounds like a loose connection between your PICkit3 and your UDB4. Are
> >> you using one of those Olimex cables? They are notorious for loose
> >> connections.
>
> >> I make up my own cables for connecting between PICkit and UDB.
>
> >> Also, you can connect the PICkit directly to the UDB without a cable.
>
> >> Best regards,
> >> Bill
>

[Riccardo Kuebler]

Phil,

yes, you need elevator to turn !

Good luck !

Ric

2011/12/11 Phil Giacalone <scienc...@gmail.com>

[William Premerlani]

Hi Ric,

You definitely need a positive value of ROLL_ELEV_MIX, otherwise you will lose altitude in the turns. I would try a value of 0.5 to start with. If that is too high, you will find your aircraft will pitch up in the turns, and you will then want to reduce it. If your plane still loses altitude in turns, then you should raise it.

Its actually not too critical for T3, because altitude control will compensate for loss/gain of altitude in the turns.

Best regards and good luck,

Bill

[Phil Giacalone]

Ok, got it. Will go with 0.5 to start.

Thanks again, everyone. I'm headed to the field. Weather conditions
are surprisingly good. Stay tuned.

Best regards,
Phil

On Dec 11, 1:09 pm, William Premerlani <wpremerl...@gmail.com> wrote:
> Hi Ric,
>
> You definitely need a positive value of ROLL_ELEV_MIX, otherwise you will
> lose altitude in the turns. I would try a value of 0.5 to start with. If
> that is too high, you will find your aircraft will pitch up in the turns,
> and you will then want to reduce it. If your plane still loses altitude in
> turns, then you should raise it.
>
> Its actually not too critical for T3, because altitude control will
> compensate for loss/gain of altitude in the turns.
>
> Best regards and good luck,
> Bill
>

[Phil Giacalone]

Hi all,

Well, I gave it the old college try but couldn't get a good run. It
started raining before I could get the gains right.

The first flight I had a nasty turn 2 -- to much steering and not
enough stability. I adjusted the gains and flew again but the turns
were way too wide. Then the rains came so I had to call it a day.

Best regards,
Phil

[William Premerlani]

Hi Phil,

Thanks for trying. I am very glad that, in any case, you flew safely. I was a little worried that you were pushing a bit too hard.

I would expect that it will take a few flights for you to tune the gains. So, with the contest over, the pressure is off now, and you can get things adjusted at your leisure.

Best regards,

Bill

[Phil Giacalone]

Thanks again for being so responsive today, everyone! It's been said
before, but you are an amazing group of people.

Ric: congrats on that final inverted T3 run. This hobby can be so
humbling at times. Realizing how difficult things can be makes your
accomplishments even more amazing.

Best regards,
Phil

On Dec 11, 4:15 pm, William Premerlani <wpremerl...@gmail.com> wrote:
> Hi Phil,
>
> Thanks for trying. I am very glad that, in any case, you flew safely. I was
> a little worried that you were pushing a bit too hard.
>
> I would expect that it will take a few flights for you to tune the gains.
> So, with the contest over, the pressure is off now, and you can get things
> adjusted at your leisure.
>
> Best regards,
> Bill
>

[Coby]

Great effort Phil!
I am still waiting to get a UDB in something other than a client's
prototype. Two of our current projects use elevons. I'll try to pay
more attention to our next test. In general most of our elevon issues
have been flaky initialization (not steady at ten second mark), or bad
boards. I have had a 2 out of 10 failure rate on this project in UDB
3's from Sparkfun – we are guessing it's cold solder related, but
have not tried to reflow them as we have not tweaked a toaster oven
yet;-). We also had to reset our Spektrum DX8 model memory last
night. Not sure if it's something we did our it just got corrupted.
We were having all kinds of elevon direction and travel issues; it was
working fine previously. So we re-flashed the UDB; reset the model on
the DX8; set everything up from scratch; and it's working fine now.
We have also noted that it can make a difference in which servo - left
or right - you make the aileron/elevator in an elevon set-up. I never
can remember what works for Spektrum so have to try it out on the
bench until I get it to work right. Essentially, we see the radio
wanting to drive the servos in one direction and the IMU in the other,
and no amount of servo reversing on either device will fix it. They
just flip back and forth and never give you the direction and response
you want. So we first try to make sure the radio is happy with the
elevon setup and then adjust the UDB as required. I have also
received a PicKit3 recently that was giving me a similar error as
yours. We make our own cables and have a few Olimex cables around.
Did not make a difference. I could use any of the cables on our known
good PicKit3 and get it to work. Also tried MapLabX on a PC and could
get it to compile and load but the UDB did not seem to like the load.
I have no idea what I am doing with MapLabX, so it's probably me. All
that to say there are a bunch of variables in this game, so hope you
stick with it and have some fun.

R,
C

> ...
>
> read more »

[William Premerlani]

Hi Coby,

Here is a sure-fire way to get elevons set up with MatrixPilot.

1. First, get your Tx set up correctly with MatrixPilot in manual mode, or without the UDB connected.

2. Set the values of AILERON_CHANNEL_REVERSED and ELEVATOR_CHANNEL_REVERSED to get the elevons to respond correctly to pitch in stabilized mode. The response of each elevon to pitch is controlled separately with these two parameters. When you are done with this step, both elevons will respond in the same, correct direction to pitch.

3. Set the value of ELEVON_VTAIL_SURFACES_REVERSED to get the elevons to respond correctly to roll.

The "problem" with elevons is that it would seem off hand that you need 2 bits of information to set polarity, but there is actually 3 bits of information, because it is possible to swap the left and right servo.

Best regards,

Bill

[Coby]

Thanks Bill.

Yes, that's generally how we try to attack our elevon setup as well
(now anyway). Although I usually use the H/W switches on the board to
reverse if needed. Our first attempt a few months back provided many
days of lessons learned. The great news – it was so frustrating I
actually took time to document, in writing no less, something we were
doing. Also got better at programmable mixes before we figured out it
was not required at all if one follows your steps.

R,
C

[Phil Giacalone]

Thanks for the kind words and encouragement, Coby. No worries about
sticking with it... This hobby has me hooked :)
Best regards,Phil

> ...
>
> read more »