Feedback on MultiWii_dev20110714 + R212

Point65 · Post by **Point65** » Thu Jul 21, 2011 4:08 pm

I tested the 0714 + R212 code on my mini-tri (WMP+BMA020) which has strong motors and light weight today. It works great in
acro and stable mode. Compared to 1.7 it's more stable with the deafult values (P=4.0) than with 1.7 where I hade to reduce
Roll & Pitich P down to 2.0. Some fine tuning needed for FFF and fast decent where I still have some wobble.
I'll wait for 1.8 before I do more testing.

Link to the mini-tri: http://www.rcgroups.com/forums/showpost.php?p=18116094&postcount=13832 The WMP is now
back to original insteda of the "hacked" ITG3205. Props are now 70x60 and I'll also swap the 6A ESC's for 10A ESC's to avoid
overheating with the 3-cell battery and aggressive flying.

I did a short test on my big tri with FFIMU V1 and needed to go up to Roll & Pitich P to 9.0 before it was flyable. I'll install a
original WMP+BMA020 to compare later with 1.8.
/Bo

spagoziak · Post by **spagoziak** » Thu Jul 21, 2011 7:33 pm

I've found that auto level is pretty weak in dev 0714... when I tip the quad to a sharp angle and release the sticks to center, the quad doesn't hurry to regain level. And this is with 10.x on P--the absolute maximum working value for version 1.7. Did changes in the dev code contribute to how autolevel is calculated?

spagoziak · Post by **spagoziak** » Fri Jul 22, 2011 5:40 am

spagoziak wrote:I've found that auto level is pretty weak in dev 0714... when I tip the quad to a sharp angle and release the sticks to center, the quad doesn't hurry to regain level. And this is with 10.x on P--the absolute maximum working value for version 1.7. Did changes in the dev code contribute to how autolevel is calculated?

Getting better! I added 2.0 to P in Level and it seems to be more assertive.

How have changes since 1.7 affected this? Should my understanding of autolevel's PI terms be revised?

Point65 · Post by **Point65** » Fri Jul 22, 2011 10:46 am

spagoziak wrote:
spagoziak wrote:I've found that auto level is pretty weak in dev 0714... when I tip the quad to a sharp angle and release the sticks to center, the quad doesn't hurry to regain level. And this is with 10.x on P--the absolute maximum working value for version 1.7. Did changes in the dev code contribute to how autolevel is calculated?

Getting better! I added 2.0 to P in Level and it seems to be more assertive.

How have changes since 1.7 affected this? Should my understanding of autolevel's PI terms be revised?

It may be the introduction of MAX_CORRECTION, I'm using the default of 500. I'm not sure what the correct value
is to emulate 1.7.
/Bo

spagoziak · Post by **spagoziak** » Fri Jul 22, 2011 4:16 pm

Alex posted on RCG about Max_correction when he introduced 0714; the default was 100 then... where'd you find 500? He said that 1000 would disable it, effectively returning it to 1.7's feel.

Hamburger · Post by **Hamburger** » Fri Jul 22, 2011 6:29 pm

Max_correction is no longer in the config.h of the current dev. It was a temporary fix to a hidden bug, iirc.

captaingeek · Post by **captaingeek** » Fri Jul 22, 2011 7:31 pm

is it just me or is the gui config much smoother in this version?

Alexinparis · Post by **Alexinparis** » Fri Jul 22, 2011 8:24 pm

Hamburger wrote:Max_correction is no longer in the config.h of the current dev. It was a temporary fix to a hidden bug, iirc.

Hi,
this is exactly the good reason

The GUI is smoother just because the GUI update time is higher, there is nothing revolutionary behind

It's possible to make it even smoother, but the CPU time of your PC will increase maybe too much.

dramida · Post by **dramida** » Sat Jul 23, 2011 8:00 pm

one fpv and one altitude hold flight with dev0714

http://www.youtube.com/watch?v=iOY9JuYKWns

http://www.youtube.com/watch?v=RMMMP0wz1WY

dramida · Post by **dramida** » Sat Jul 23, 2011 8:02 pm

and those are the settings of the flights:

captaingeek · Post by **captaingeek** » Sat Jul 23, 2011 10:33 pm

cant seem to get mine in the air with this version. doesn't seem like its outputting enough power.

spagoziak · Post by **spagoziak** » Sun Jul 24, 2011 6:52 am

dramida wrote:and those are the settings of the flights:

Dramida, it seems you have some insight on how to tune auto level. Can you impart what you've come to understand about the PID values for the baro?

crashlander · Post by **crashlander** » Sun Jul 24, 2011 7:48 am

captaingeek wrote:cant seem to get mine in the air with this version. doesn't seem like its outputting enough power.

Same for me! Low MAX_CORRECTION (100) backflip, high MAX_CORRECTION (500) not enough lift. R214 without MAX_CORRECTION behaves same as older with seeing of 500.
Andrej

spagoziak · Post by **spagoziak** » Sun Jul 24, 2011 8:06 am

Alex has removed max_c in the latest revision... it was there to fix a bug, but another approach was used and max_c was no longer necessary. All gone!

Alexinparis · Post by **Alexinparis** » Sun Jul 24, 2011 10:09 am

R214 should behave very close to a 1.7 with a higher gyro resolution.
I think the main bug was removed.
The lack of power is generally due to another hardware problem: one weak ESC of one faulty motor could cause a loss of overall power.

crashlander · Post by **crashlander** » Sun Jul 24, 2011 10:37 am

Alexinparis wrote:R214 should behave very close to a 1.7 with a higher gyro resolution.
I think the main bug was removed.
The lack of power is generally due to another hardware problem: one weak ESC of one faulty motor could cause a loss of overall power.

HW is not an issue in my case, because if I return to 0607 everything returns to normal (stable fly, strong pitch control).

crashlander · Post by **crashlander** » Sun Jul 24, 2011 10:41 am

spagoziak wrote:Alex has removed max_c in the latest revision... it was there to fix a bug, but another approach was used and max_c was no longer necessary. All gone!

Yes I'm aware of that, but behavior on r214 (latest) is the same as previous one with high max_c (low motor power).

Alexinparis · Post by **Alexinparis** » Sun Jul 24, 2011 11:21 am

crashlander wrote:
Alexinparis wrote:R214 should behave very close to a 1.7 with a higher gyro resolution.
I think the main bug was removed.
The lack of power is generally due to another hardware problem: one weak ESC of one faulty motor could cause a loss of overall power.

HW is not an issue in my case, because if I return to 0607 everything returns to normal (stable fly, strong pitch control).

So, this is not a good news for the code.
Something might remain wrong somewhere

I have almost the same setup as you: Quad X, promini, original WMP, TGY Plush 10A (HK SS 18A for me), and R214 is the best soft I've seen so far on this setup. strange...

crashlander · Post by **crashlander** » Sun Jul 24, 2011 2:14 pm

My tests with r214 were done with poorly balanced quad. With 0607 it only wobbles on one axe but it is quite flyable and stable.

spagoziak · Post by **spagoziak** » Sun Jul 24, 2011 3:28 pm

Interesting indeed! I'm using the same setup for every bit of code--I only have one quad, so that makes things easy

I have more than enough power, though I've only tested indoors in dead air. But the trust/stick position is familiar and reminiscent of the other versions I've flown. I'll try to test it outside today, there's another stiff breeze. I'll report how the PDF handles too, if I get a chance to lift off.

captaingeek · Post by **captaingeek** » Sun Jul 24, 2011 5:10 pm

im not convinced that its not a problem with the esc's. Im testing a new paris v4 board with an atmel inertia 1. i use a seperate BEC to power the board. I've cut the 5v lines from all the esc's. maybe they need that to operate.

in order to calibrate, i power the board via usb, activate the controller and goto full power. rear left motor shows 1850 out in the gui. i connect this output individually to each esc in order to calibrate max throttle one by one.

captaingeek · Post by **captaingeek** » Sun Jul 24, 2011 5:51 pm

crashlander wrote:My tests with r214 were done with poorly balanced quad. With 0607 it only wobbles on one axe but it is quite flyable and stable.

what do you mean by 0607?

crashlander · Post by **crashlander** » Sun Jul 24, 2011 8:58 pm

I meant dev version released as zip with 20110607 in name (probably somewhere around r197).

dramida · Post by **dramida** » Sun Jul 24, 2011 11:35 pm

this is how it flyes in fpv mode

http://www.youtube.com/watch?v=7ASX0Awtkes

captaingeek · Post by **captaingeek** » Mon Jul 25, 2011 1:17 am

let the battery run down to far nice flight! wish i was there i have all the fpv gear i need just not quite dialed in yet.

ziss_dm · Post by **ziss_dm** » Mon Jul 25, 2011 1:27 am

Hi Alex,

I have just noticed:

Code: Select all

rcCommand[THROTTLE] = MINTHROTTLE + (int32_t)(MAXTHROTTLE-MINTHROTTLE)* (rcData[THROTTLE]-MINCHECK)/(2000-MINCHECK);

I think, it is better to use MAXCHECK instead 2000:

Code: Select all

rcCommand[THROTTLE] = MINTHROTTLE + (int32_t)(MAXTHROTTLE-MINTHROTTLE)* (rcData[THROTTLE]-MINCHECK)/(MAXCHECK-MINCHECK);

regards,
ziss_dm

crashlander · Post by **crashlander** » Mon Jul 25, 2011 4:43 pm

Alexinparis wrote:
crashlander wrote:
Alexinparis wrote:R214 should behave very close to a 1.7 with a higher gyro resolution.
I think the main bug was removed.
The lack of power is generally due to another hardware problem: one weak ESC of one faulty motor could cause a loss of overall power.

HW is not an issue in my case, because if I return to 0607 everything returns to normal (stable fly, strong pitch control).

So, this is not a good news for the code.
Something might remain wrong somewhere
I have almost the same setup as you: Quad X, promini, original WMP, TGY Plush 10A (HK SS 18A for me), and R214 is the best soft I've seen so far on this setup. strange...

To continue...

Today I tested with new 9DOF from Atmel with R214 and the symptoms of not enough power to motors are still here.
If I hold quad on the ground firmly and adding throttle i can see (in GUI) almost full output to ESC and can hear fast spinning of engines, if I repeat same test while holding quad in my hand I can feel no lift (and see weaker response in GUI).

crashlander · Post by **crashlander** » Mon Jul 25, 2011 6:15 pm

I'm attaching screenshot of gyro readings from GUI. First half is taken while holding in hand and last part is while holding firmly on the ground. With my current setup that is probably the best balance I can obtain (cheep HK outruners, cheep HK props with prop savers) but it works with older versions.

: Gyro in hand and on the floor

ziss_dm · Post by **ziss_dm** » Mon Jul 25, 2011 10:35 pm

Hi crashlander,
Can you attach full screenshot of GUI?

regards,
ziss_dm

Alexinparis · Post by **Alexinparis** » Mon Jul 25, 2011 11:38 pm

Hi,

If the throttle stick is well adjusted, it should go from 1000 up to 2000 in the GUI.
Then 2 situations:
- under MINCHECK: throttle = MINTHROTTLE (armed) or 1000 (unarmed)
- between MINCHECK and 2000: the thottle is linearly adjusted between MINTHROTTLE and MAXTHROTTLE

no error here.
with MAXCHECK, the upper value would be greater than MAXTHROTTLE.

ziss_dm wrote:Hi Alex,

I have just noticed:
Code: Select all
rcCommand[THROTTLE] = MINTHROTTLE + (int32_t)(MAXTHROTTLE-MINTHROTTLE)* (rcData[THROTTLE]-MINCHECK)/(2000-MINCHECK);
I think, it is better to use MAXCHECK instead 2000:
Code: Select all
rcCommand[THROTTLE] = MINTHROTTLE + (int32_t)(MAXTHROTTLE-MINTHROTTLE)* (rcData[THROTTLE]-MINCHECK)/(MAXCHECK-MINCHECK);
regards,
ziss_dm

captaingeek · Post by **captaingeek** » Tue Jul 26, 2011 4:03 am

fyi - with eosbandits code changes it flew great with the inertia 1. level mode works great. mag mode works. flew pretty well with all default. and this is a quad with 18" arms thats a bit of a pig. thanks alex!

crashlander · Post by **crashlander** » Tue Jul 26, 2011 6:06 am

ziss_dm wrote:Hi crashlander,
Can you attach full screenshot of GUI?

regards,
ziss_dm

Hello.
That is it:

: r214_gyro

ziss_dm · Post by **ziss_dm** » Tue Jul 26, 2011 7:38 am

Hi crashlander,

Sorry for stupid question, but sumptomps fit:
Are you sure you not trying to takeoff with Alt.Hold (Baro) turned on?

regards,
ziss_dm

crashlander · Post by **crashlander** » Tue Jul 26, 2011 12:45 pm

Screenshot is taken during (too)low motor power and shows baro, alt. hold off.
And symptoms were there also with (just) original WM+, without baro present and also in older versions (r212) when deprecated max_c parameter was set to 500 (also with just WM+)!
The actual symptoms are: it is initially possible to take off but during the flight I must add full throttle just to stay airborne and sometimes it is not enough, during flight quad is very unstable!

ziss_dm · Post by **ziss_dm** » Tue Jul 26, 2011 1:27 pm

Hi crashlander,

Can you drop all PIDs to the 0 and do hand test again?

regards,
ziss_dm

crashlander · Post by **crashlander** » Tue Jul 26, 2011 2:30 pm

By all you mean ALL (gyro, acc., baro...)?!

crashlander · Post by **crashlander** » Tue Jul 26, 2011 2:50 pm

ziss_dm wrote:Hi crashlander,

Can you drop all PIDs to the 0 and do hand test again?

regards,
ziss_dm

ALL 0!
Engines are roaring! (strong lift)

: PID 0

Another observation (full throttle):
With PID's properly set and holding quad firmly on the ground (soft carpet) I can see engine commands jumping between 1800 and 2000 (more time closer to 2000),
when I'm holding quad in my arm those values are moving between 1300 (MINTHROTTLE) and 2000 (MAXTHROTTLE) (more time closer to 1300.

ziss_dm · Post by **ziss_dm** » Tue Jul 26, 2011 3:02 pm

Hi crashlander,
And if you add roll, pitch, yaw P?

regards,
ziss_dm

captaingeek · Post by **captaingeek** » Tue Jul 26, 2011 3:19 pm

crashlander wrote:
ziss_dm wrote:Another observation (full throttle):
With PID's properly set and holding quad firmly on the ground (soft carpet) I can see engine commands jumping between 1800 and 2000 (more time closer to 2000),
when I'm holding quad in my arm those values are moving between 1300 (MINTHROTTLE) and 2000 (MAXTHROTTLE) (more time closer to 1300.

I noticed this same behaviour until i made the code changes as suggested by eos.

crashlander · Post by **crashlander** » Tue Jul 26, 2011 4:19 pm

STOP what you are thinking!

It's my fault! And it is called TOOMUCH gyro dampening!
I had 8mm if soft sponge and some lead under my gyro to dampen vibrations! It worked with previous versions of software and original WM+ (it worked very good also with clone NK). But it is absolutely useless with new SW and Atmel 9DOF!
After attaching my new gyro more firmly (only very thin sponge) I can get full motor power (in GUI and in real).

Sory for all the confusion!

BTW: The proof that my previous dampening worked are those videos, all made with that setup.
http://www.youtube.com/watch?v=Y9yxrp8O ... BC&index=0

OCaerial · Post by **OCaerial** » Tue Jul 26, 2011 5:59 pm

been trying out the dev_0714 and its been working very nice for me except one problem.
When I switch the "BARO" on the copter hovers around for about 30 sec average, then woosh, it shoots up to the sky until I turn of the BARO. Most of the time it only does this once when I first start flying?
Anyone have any idea how to fix this?

I have a video example here: http://vimeo.com/26922516

FC: Flyduino & FREEFLIGHT IMU 1.22L

captaingeek · Post by **captaingeek** » Thu Aug 04, 2011 4:50 am

atmel i1 and current dev:

http://www.youtube.com/watch?v=MsbslvYMR8o

mag seems to work. havn't touched new settings other then that. stable mode takes quite a bit of software trim (20 clicks left aileron and 15 back) and tends to dwel a bit during flight testing.