HobbyKing.com New Products Flash Sale
Reply
Thread Tools
Old Oct 19, 2008, 03:32 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Data
Spektrum DX6i Elevon Weirdness: Worse Than You Think

Let's cut to the chase -- if you're looking for the Dx6i Elevon Fix, it's here:
http://www.rcgroups.com/forums/showp...2&postcount=35
Now back to your regularly scheduled long-winded explanation ->

By now it's common knowledge that the elevon hookup chart in the manual is incorrect. That's fairly minor, though it has confused quite a few people.

One thing I haven't seen mentioned on the boards? The differential setting does not work for elevons, only for Dual Ailerons (flaperons) even though the manual indicates it should. I find this particularly annoying.

Plenty of people have noticed that you don't get full travel from your servos when you set the elevon mixing, but not many seem to have noticed that the error is imbalanced -- radically so -- and will make the plane fly quite strangely.

To illustrate, I built a "plane" with a couple of servos, extended wire indicators, an AR6100, and a receiver battery. Here's the business view:
timocharis is offline Find More Posts by timocharis
Last edited by timocharis; Apr 11, 2009 at 06:06 PM. Reason: Add link to the fix
Reply With Quote
Sign up now
to remove ads between posts
Old Oct 19, 2008, 03:35 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Elevon Mix: Full Right Aileron

Here's what it looks like when you give the transmitter full right with the elevon mix set. The servos are cheapo Hex 5-grams, much like the units in the plane where I first noticed this problem.

Look to you like the left aileron is going down a bit less than the right aileron is going up?
timocharis is offline Find More Posts by timocharis
Reply With Quote
Old Oct 19, 2008, 03:43 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Elevon Mix: Full Down

Logically enough, I tried the four cardinal points and marked the positions of the deflection. Here's the "full down" position on the Elevon mix, and you can see the markings corresponding to the "needle" positions. Degree measurements were then taken with a standard goniometer.

After doing all four positions, here's what happens:

Full Down and Full Right Aileron give the same result on the left side: Down 12 degrees.

Full Up and Full Left Aileron give the same result on the left side: Up 26 degrees.

Whoa! What's with that?

On the right side, it's even weirder. Full Right Aileron gives 15 degrees up, but Full Up gives 22 degrees up!

Just to balance that out, Full Left Aileron gives 22 degrees down on the right side, but Full Down gives only 15 degrees. Up/Down and Left/Right are imbalanced by seven degrees.

Here's another way to look at it:

On the left side, everything goes up 14 degrees more than it goes down.

On the right side, Right/Left is rotated 7 degrees clockwise from Up/Down.

In no case is the operation symmetrical, so the plane will behave weirdly no matter what input it gets.
timocharis is offline Find More Posts by timocharis
Last edited by timocharis; Oct 19, 2008 at 03:58 PM.
Reply With Quote
Old Oct 19, 2008, 03:49 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Is It Just The Elevon Mix? Yup.

To isolate the problem, I blanked the model but otherwise changed nothing, then programed Aileron -> Elevator mix and Elevator -> Aileron mix to emulate an Elevon mix, bypassing the builtin Elevon routine.

The result was a consistent 37-degree deflection up/down and left/right. (I should note that actually that's too much throw, but you can calm it back down using dual rates). Here's one of the test photos so you can see how it looked:
timocharis is offline Find More Posts by timocharis
Last edited by timocharis; Oct 21, 2008 at 12:37 PM.
Reply With Quote
Old Oct 19, 2008, 03:52 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
That means on the left side, I get 25 degrees more down travel, but only about 11 degrees more up. On the right side, I get a minimum of 15 degrees more both up and down. And now my controls are completely symmetrical.

I think this establishes that there are serious problem with the Elevon Mix subroutine, at least in my DX6i. Would be very interested to see what results other folks might get.

However, using the "stock" elevon mix it would be nearly impossible to trim a plane out properly. Though I suppose it's possible somehow.
timocharis is offline Find More Posts by timocharis
Last edited by timocharis; Oct 21, 2008 at 12:36 PM.
Reply With Quote
Old Oct 19, 2008, 03:57 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Recap

1. The elevon mix is imbalanced.

2. Differential does not work.

3. Elevon hookup instructions are wrong.

It's surprising something this defective has come out in a production unit. Well, maybe I shouldn't be surprised, but I am.

There's only one reasonable conclusion: don't use the Elevon mix.

At this point, as far as I can tell, the Dual Aileron (Flaperon) setup does not suffer from any such problems and should work fine. The only really big downside to this program goofup is that we don't have any available mixes to use with an Elevon plane, since the meager allotment of two is completely consumed to fix the error.

Unless my unit is very anomalous, Spektrum deserves some healthy thumping over this screwup.


Dave
timocharis is offline Find More Posts by timocharis
Reply With Quote
Old Oct 21, 2008, 12:30 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Other Units Tested For Baseline

For comparison I set up the "plane" with a JR 6102 and a JR 7202 and a cheapo Corona single-conversion 4-channel receiver. In all cases there is no ESC -- just a receiver battery.

Both JR units behaved identically. The four cardinal points (up, down, left, right) give the ideal 30-degrees of travel to the appropriate servo in both directions. Going to the ordinals, the addition/subtraction is perfect in both units -- the "off" arm zeros out and the "on" arm goes to 60 degrees.

In other words, they work as expected with no glitches or weirdness.


Dave
timocharis is offline Find More Posts by timocharis
Reply With Quote
Old Oct 21, 2008, 01:30 PM
Inciting Riots
village_idiot's Avatar
Joined Dec 2006
9,687 Posts
Have you contacted Horizon about this yet? It looks like the email support is getting better, I only waited a week for the last email I sent them with a new product suggestion.
village_idiot is offline Find More Posts by village_idiot
Reply With Quote
Old Oct 21, 2008, 04:58 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Quote:
Originally Posted by village_idiot
Have you contacted Horizon about this yet?
Sort of. I sent an email to Spektrum (same thing, maybe) from their "contact us" web form. Perhaps that doesn't go to Horizon? (edt: yes it does, according to the page involved. It's addressed to Horizon with Attn: Spektrum as the 2nd line).

Here's the contents:

"I recently bought a DX6i. Nice unit mostly, but the Elevon mixing has several problems. The instructions are incorrect for plug-in, the differential adjustment does not work, the travel is constrained between 30-70% of normal and deflection is asymmetric by 7-14 degrees depending on input (up/down or left/right). Would be interested to know when you'll have a fix and how it will be managed.

Thanks, Dave"

No comment back yet, but it's only been a few days.


Dave
timocharis is offline Find More Posts by timocharis
Last edited by timocharis; Oct 21, 2008 at 08:27 PM.
Reply With Quote
Old Oct 21, 2008, 06:47 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Refined Data

I think I've isolated the problem. The angular error is mostly (or completely) not in the main elevon function, but rather in the trim subroutine. In order to match the "needles" up on my servo gauge,

I used some trim. I got curious about that because it was perhaps silly of me to assume the trim is working when the main program is not. So I took out the trim and normalized the needles mechanically (taping and gluing to the correct position). I also subbed out both servos and criss-crossed them to see if there was any error there (total of four servos tested).

Here's what I see now (and it's even weirder than I thought):

Travel is almost exactly 1/2 of normal both up/down and left/right. It's symmetrical across the left-right axis, but I'm seeing a persistent perhaps two degrees more "down" travel than "up" (a bit of built-in reverse differential). This is quite slight, and may be some kind of measurement error.

In the ordinal points (down+left, down+right, up+left, up+right) the needles go exactly where you'd expect: full "normal" travel on one side, and dead neutral on the other.

All mostly good so far, right? Well, pretty feeble throw compared to normal JR transmitters, but at least working sort of as you'd expect.
timocharis is offline Find More Posts by timocharis
Reply With Quote
Old Oct 21, 2008, 07:02 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
The Real Nasty Problem

Here's the weird part: If you trim toward up or down, the servos will move where you'd expect but the up/down movement endpoints will remain the same (left/right endpoints will act normally).

The same is true of the ailerons. If you put in, say, left trim ... the trim will be applied to the center point, but not to the left or right end point. However, in this case the elevator inputs will act as expected.

You should be able to easily see this without making a "gauge plane" like mine. Set your trims to neutral and note how far up and down your elevons move. Apply full up or down trim, then notice that if you go full up and/or full down they will go to the exact same place they did before you apply the trim.

To make things weirder, if you do the same correction using subtrim, there will be no issue -- the endpoints will move along with the center points.

At least they're consistent. This appears to also be the case when you use "Acro" rather than "Elevon," so the endpoints never follow the center point when you use the trim buttons.
timocharis is offline Find More Posts by timocharis
Reply With Quote
Old Oct 21, 2008, 07:05 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Is This Normal?

Not on any other transmitter I have. On all three, the endpoints follow the centerpoint when you adjust the trim using the digital input buttons.

Does anyone else have a transmitter that shows this behavior? It certainly can add up to some weird performance in the air.

So the elevon-specific problem is just the failure of Differential and the extraordinarily low throws. The rest of the problem is a flawed implementation of trim on the face buttons.
timocharis is offline Find More Posts by timocharis
Reply With Quote
Old Oct 21, 2008, 07:08 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Comments at this point

I'd love to see someone repeat this using digital servos. Or what the heck, lend me a couple. I'd like to eliminate (if possible) and servo "behavior." In particular, I'd like to see if the signals being sent at zero trim are in fact symmetrical, or if there is an up/down imbalance.

The comparative short throws and trim error are beyond any servo irregularities; that's a settled issue. And of course the incorrect plug instructions will remain incorrect regardless.
timocharis is offline Find More Posts by timocharis
Reply With Quote
Old Oct 21, 2008, 08:08 PM
Registered User
Daemon's Avatar
Lakewood, Colorado
Joined Aug 2002
28,896 Posts
Some radios have *absolute* programmed endpoints (like
Multiplex), and some let you push past them with mixes (like
most low end Hitecs). This latter includes the Elevon mix which
if allowed to drive both roll and pitch at 100%, would then
try to go to 200% when you apply both together. Obviously
they mitigate this by reducing both sets of endpoints by 50%
so that the two added together don't exceed 100%.

I suspect that if you want more total throw you can increase the
absolute endpoints to as much as 140%.

It sounds like the trim acts directly upon the signal from the gymbal pots,
but sub-trim is lower level. Imagine that the trim controls
were were screws that offset the pot position slightly. That's how it's
behaving, and is why if you tweak the trim, it doesn't change
the endpoints.

ian
Daemon is online now Find More Posts by Daemon
RCG Plus Member
Old Oct 21, 2008, 08:25 PM
Dave North
timocharis's Avatar
USA, CA, San Jose
Joined Apr 2004
5,254 Posts
Quote:
Originally Posted by Daemon
Obviously they mitigate this by reducing both sets of endpoints by 50% so that the two added together don't exceed 100%.
Ian, I suspect you're on to something there. It appears the root of the problem is they've set the full-throw endpoints to approximately 60% the normal value. This is slightly troublesome in other wing configurations, but obviously too little when in Elevon mode.
Quote:
I suspect that if you want more total throw you can increase the absolute endpoints to as much as 140%.
I think that would correspond to their "Travel Adjust" which goes to 125%. At that point we're probably at about 75% of normal. But that's getting closer
Quote:
It sounds like the trim acts directly upon the signal from the gymbal pots...
Makes sense, and may well be what's up. Nasty bug though, no?

And of course, Differential still fails. The failure mode is interesting though -- if you don't set Elevons (Plain Old Wing) and go to the Differential menu, trying to set anything gives a null result. However, if you turn on Elevon and try the same thing, the "List" menu is no longer highlighted ("active") but nothing else anywhere is highlighted! To get out of the mode, you have to shift one more time to highlight the "List" entry (escape) and then click.

In other words, it's trying to get to the control position that's supposed to be there, but it doesn't show up in the screen. However, changing values while it's there doesn't seem to do anything.

Seems weird to me!


Dave

ps: oh no. It's even more complicated. When I try the same thing on my improvised elevon mix (no special wing, ail -> elev, elev -> ail mixes) the trim error goes away. Argh! This is going to take a while to track down...
timocharis is offline Find More Posts by timocharis
Last edited by timocharis; Oct 21, 2008 at 08:40 PM.
Reply With Quote
Reply


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Trex XL tail pitch slider binds more than you think... GoRgOyL Mini Helis 10 Sep 22, 2005 06:14 PM
Just when you think it couldn't get any worse. Mr. Wiz Life, The Universe, and Politics 7 Oct 23, 2004 09:07 AM
Iraq, Better off than you think? LTChip Life, The Universe, and Politics 17 Sep 29, 2004 08:32 AM
Harder than you'd think... P-51 fan Humor 9 Nov 08, 2003 09:34 AM