atomic's thrown together Inverted V-tail (or should that be A tail)

[atomiclama]

Now kitted out with FPV magic

[kilby]

Now kitted out with FPV magic

Where';s the tablecloth?

[atomiclama]

The missus would not let me have the table cloth and I had to fight the dog for his towel . ::)

[atomiclama]

Just a short video to show how badly this thing flexes.

http://youtu.be/taH78RUmWUk

[atomiclama]

This week I shall mostly be fiddling with the custom mixer.

[bazzerh]

is it made from rubber tree wood? does look very bendy! i really like the idea of making a woodie, where';s a good place to get small pieces of ply now i wonder? skips?

[guest325]

where';s a good place to get small pieces of ply now i wonder? skips?

I get my birch faced ply from Penn models, it';s not too expensive either (unlike most of the rest of the stuff they sell).

[atomiclama]

My bits of wood come from various places, back of the garage, scrounged some off cuts from the local timber yard, friends at work that had some scrap, packing and yes skips.

[atomiclama]

A small problem that has been niggling at me for a while. At initial lift the tail would come up first then the FC would level as I took off. Once up in the air it was fine apart from a small drift forward. ACC calibrated when level so that should not have been the problem.

Anyway started looking at the mixer for non symmetrical layouts various V-tails, deadcat layouts and hexes.
Various threads and youtube vids gave me an idea what to do but most for multi wii not the clean flight mixer.

Measured my quad and started with the calculations


First to tweak is the throttle,

normalised the distance of the motors to the FC. I want the rear motors to produce less thrust relative to the front so this gives me.

# throttle, roll, pitch, yaw
cmix 1 0.880 0.000 1.000 -1.000
cmix 2 1.000 -1.000 -1.000 0.000
cmix 3 0.880 0.000 1.000 1.000
cmix 4 1.000 1.000 -1.000 0.000 

Well that worked out well, the quad now takes of level from the start, no drift forward. With the added bonus that the hover point has shifted up which gives me more room to control the height. Climbs and descents also seam more controlled again I think due to the PID having to do less adjustment.

Result.  ::)

[guest325]

I';m glad someone else agrees with me about the importance of getting the cog/cot right; yes the fc can cope with it (after a fashion!). It';s just not good practice to do that!

[atomiclama]

I';m being converted 

One other thing that has just twigged with me and will probably change the way I think about it.
The PID controller is just a trim. Not a full control on the outputs as I understood most PIDs to work

Code Select Expand

   
if (motorCount > 1) {
        for (i = 0; i < motorCount; i++) {
            motor[i] =
                rcCommand[THROTTLE] * currentMixer[i].throttle +
                axisPID[PITCH] * currentMixer[i].pitch +
                axisPID[ROLL] * currentMixer[i].roll +
                -mixerConfig->yaw_direction * axisPID[YAW] * currentMixer[i].yaw;
        }
    }


[guest325]

 Simply put, these values can be interpreted in terms of time: P depends on the present error, I on the accumulation of past errors, and D is a prediction of futureerrors, based on current rate of change

The PID controller is just a trim. Not a full control on the outputs as I understood most PIDs to work.

Not quite,  PID';s optimise the way the model responds.
The I and D parts are those which control oscillation and overshoot the P part determines how aggressive the model is; as you would expect the more aggressive the setup the higher I and D you need for STABLE flight, as you can guess if you don';t increase the I &D to the point where you get stability you get an unstable but controllable (acro) mode.

[atomiclama]

Ok you misinterpreted my highly detailed and thought out description of why I think the PID is a trim 

I';m no PID beginner, I have a basic understanding of how they work. Most of my experience comes from PID in motor control, where the output to the motor is purely the output of the PID controller. We do have modes of operation where the setpoint goes to the motor with a PID trim applied as well. This looks closer to how it is implemented in CF code bit above.

The mixer settings have a large part to do with the values that go to the motors and the PID is trimming those values. This is sort of backed up with what I';m seeing as the more I tweak the mixer the better the response I get as the less trim is being applied by the PID. The example being when I tweaked the thrust mixer the quad took off level, where it should have been able to take off level before purely down to the PID.

Anyway this is no criticism, just a penny dropping moment and shifts my attention away from PID and over to the mixer for awhile.

[atomiclama]

Well I should have thought about that before I started 

Read some where on my random travel through the internet that some people have had good experience with altering the motor rotation away from the standard layout. Why not I thought and jump straight in.

Just about manages to take off then sits there twitching about the yaw axis.

Ah when I give it some right hand yaw the left rear motor does the work and speeds up, but in the new layout it is now going CW so induces left YAW, like I said should have thought about that.

Anyway all back to normal now.

Back to trimming the mixer.

[teslahed]

Just a short video to show how badly this thing flexes.

It looks as if you only have a top plate on that thing. As soon as you add a bottom plate it should get a lot stiffer because of the way you are bracing the parts both top and bottom; the wood adds strength both in compression and in tension this way.

If using wood i suggest 3mm thick high quality plywood for the mainplates and for the arms you want to use high quality 1cm thick pine if you can find it. The cheapest stuff from B&Q is nowhere near the quality of the good stuff.

[atomiclama]

Yeah just a top plate, always meant to put a bottom plate on it but never got round to it, the cap from the SMPS is too big as well would have to make a hole for it.

Quality wood, there is no point. This is just thrown together to get me learning and flying crashing, without worrying about the cost of the frame. I would fly a lot more cautiously if every crash I had might cost me more than a couple of props to fix  ::)

[teslahed]

A stiff frame will fly better and a strong frame will crash better.

If you can spend another £5 or so and get good quality wood you may save yourself a lot of time, effort, and cost on rebuilding the thing after the crashes.

That';s my theory anyway.

Good quality wood can be got very cheap. 3mm laserply (designed for laser cutters) is high grade stuff and available cheap online if you look around.

[atomiclama]

A stiff frame will fly better and a strong frame will crash better.

I';ll take your word on that, you are a far better flyer than I.

£5 on some quality wood this frame cost me £0

It';s only had one major rebuild when it was just glued together and the bonding within the ply gave way.
Since then it';s had a few screws in it and been very sturdy.

[atomiclama]

Was going to take a few vids of this thing flying but the camera battery didn';t like the cold this morning.

So only got a few seconds, anyway here is a short hover just off the ground.
Nothing breath taking more for my own reference

http://youtu.be/NB-EgnkCyAk

[atomiclama]

During my random internet wanderings I came across a thread that mentioned Wii-Esc being modded to run oneshot. This reminded me that I had some HK BS Escs that I had programmed with Wii-Esc when I first started, I was impressed with them back then mainly they were quieter, I think due to the different switching mode and the fact that they had regen braking enabled.

So a bit of rummaging around my box and a few lunch times worth of soldering and I have my Emax swapped out for the HK BS.

One other thing about these is I enabled the extended range so the PWM needs to go from 16 - 2016. 16 was easy to setup but I';m going to need to recompile CF to get me 2016 as the max. I';ll get round to that at some point.

Anyway they fly smoother than before, I hope due to the extra range compared to the 800 I have read that normal ESC can handle. Also the descents are more responsive I think due to the regen braking. One thing though it seems a bit more sloppy on the default PIDs guess I need to up those a bit. So far very happy with the change and won';t be going back.

[atomiclama]

RIP my little friend 



one goal post too many  ::)

[kilby]

In our club you hit a list then you own it (I seem to be leading post owner because the fixed wing people are cowards)

[guest325]

RIP my little friend 
one goal post too many  ::)

Been there, done that one - but in my case a rugby post, got in the way on a RTL 

[atomiclama]

got in the way on a RTL 

Sorry I shouldn';t laugh but