Espritmodel.com Telemetry Radio
Reply
Thread Tools
Old Oct 05, 2009, 08:31 PM
Registered User
Great White North
Joined Jun 2008
628 Posts
Quote:
Originally Posted by npole
So far i've had too fast turns, or in some cases not turning at all in one direction... so i'm trying to find a configuration that would works *always*
Npole - I totally agree and that is what I am striving for as well. I must admit I have had a few instances where my plane was correctly circling over home in RTH mode, but then suddenly started flying away. I was not recording so I could not review the video after to try and see what happened.

It would be good to record your tests so that you can review later in detail when you are not busy getting control back of the plane. this may show you that for example, the satellite count went low and therefore the heading was no longer accurate which would screw up RTH.

regarding the plane turning too fast to one side - have you tried increasing the gain on the CPD4 so that it has "more" control. for me I have the Copilot II on my easy star at 100% which is moderate control and have the RTH turn gain to 10% and limit at 20. Also Initially I had my air speed set too low. Are you using the speed sensor or GPS speed? with GPS speed alone you can stall flying with the wind because your ground speed will be high, but air speed low. if you are turning with the wind the turn can result in a spiral down, as you quickly loose air speed.

In the video posted by new360 here his left hand turns are ok, but he can't make a right hand RTH turn. he has his integral gain set very low, and so if he turns into the wind with a low integral gain he may not be able to initiate a turn. Could it be that wind is affecting your turns?

Not sure if any of this applies to you, but it would be interesting to see a video of your experience (sorry I didn't check if you already posted one)

Al
alros_100 is offline Find More Posts by alros_100
Reply With Quote
Sign up now
to remove ads between posts
Old Oct 05, 2009, 08:59 PM
Registered User
Joined Apr 2009
717 Posts
The thing is that im "playing" with values because i want to check how the affects the RTH to find then the best match.
I record 100% of my flight, even when im not doing anything "special", i just dont find the time to encode/upload em all..

After further review of the records, i've found two things that were probably (almost sure) affecting the RTH. The first one was the "wizard", i mentioned that i'm using the EPA (endpoints) while doing the wizard, so i'm sure that a full stick will end into the wanted value, what i did wrong is that i've set the EPA only on the left side (since during the wizard it asks only to move the stick to left), but i totally forgot that moving your EPA only to a side would change the stick center value by "miles", so when i've set the "level flight" stick it wasn't the real one i were using in fly, so when i moved the EPA back to the default the center learned by the OSD PRO was a way different than the one i were really using. That ended into having a massive left turning, but a zero right turn, infact looking at the values when it were supposed to turn right i've notice that the RTH was really applying the correct value, but that value was around my level flight stick (and not for a right turn).
A "reset flight level" during the flight would have adjusted it, but i didn't tought it atm, so i didn't tried it.

Now this opens another question (to bill or any other that may know): when the RTH wizard learns the value, it does learn the difference by the center (that seems logic...) or it does learn the absolute values (that would be wrong but you never know...). I'm asking this because i've imagined a scenario: (fake numbers) i have a center stick at 0, and left/right set by the wizard are -200 (and then +200); now when the RTH engages it outputs +/-200 max values (indeed); now i land and i'll adjust my radio trims to: center stick at 100, left at -100, right at +300; i do *NOT* do the wizard again, because i just moved the center (trim); so i just use the option to reset the "flight level"; in this scenario the RTH will correct uses the new center (at 100) and the previous learned values (+/-200) so it applies the correct values: -100/+300?

I know it sounds logic.. but i wanted to be sure before doing further tests. This may be the problem experienced by Neo360 as well, if he moved the airelons trim a way much during the flight and he never reset the RTH, he wouldn't have enough stick left on a side, that would explain why his plane were turning only on a side.
npole is offline Find More Posts by npole
Reply With Quote
Old Oct 05, 2009, 11:20 PM
Registered User
Great White North
Joined Jun 2008
628 Posts
Quote:
Originally Posted by npole
I know it sounds logic.. but i wanted to be sure before doing further tests. This may be the problem experienced by Neo360 as well, if he moved the airelons trim a way much during the flight and he never reset the RTH, he wouldn't have enough stick left on a side, that would explain why his plane were turning only on a side.
Relooking at neo360's video after reading your post, I think you may be right. his centre rudder is -28, when he banks right manually the rudder value increased to 100 or so, when he banks left it reaches -100 or so. in RTH when it banks left you see values for rudder of -75 to -100 or so, but when it appears the RTH is trying to bank right, I don't see any rudder values above 10 or 20 or so. If it was just wind like I had thought we should have seen higher rudder values, associated with the plane trying to fight the wind. But I still wonder about his very low turn integral gain and how it plays in this scenario.

I guess the question is does the simulator RTH show the true rudder movements and values? That should be a way to see if RTH works to turn right and left by reading the rudder values

Al
alros_100 is offline Find More Posts by alros_100
Reply With Quote
Old Oct 06, 2009, 03:05 AM
Registered User
mr..speed's Avatar
near Stgt/Germany
Joined Feb 2003
1,147 Posts
Quote:
Originally Posted by vpersiani
OK got RTH and simulator running... using Aux1/2 inputs.

Still a couple of weird things:
-sometimes I get "Error 5" when trying to update OSD parameters.
-on my laptop, everytime I connect the OSD he wants to upgrade firware saying "unknown firmware version". And even after updating, apparently it still can't read firmware version.
On desktop, no such issue (both PCs running XP).

Anyway, looking forward to some nice weather this weekend hopefully...
Laptops are sometimes strange on USB, maybe a lower voltage, other chip set, had the same problems but with my new laptop it works. Laptops are designed for low power consumption.
mr..speed is offline Find More Posts by mr..speed
Last edited by mr..speed; Oct 06, 2009 at 03:21 AM.
Reply With Quote
Old Oct 06, 2009, 03:11 AM
Registered User
mr..speed's Avatar
near Stgt/Germany
Joined Feb 2003
1,147 Posts
Quote:
Originally Posted by alros_100
Npole -
It would be good to record your tests so that you can review later in detail when you are not busy getting control back of the plane. this may show you that for example, the satellite count went low and therefore the heading was no longer accurate which would screw up RTH.
Al
As I understood, there will be a msg if satellite count goes down and RTH gives up.
mr..speed is offline Find More Posts by mr..speed
Reply With Quote
Old Oct 06, 2009, 03:20 AM
Registered User
mr..speed's Avatar
near Stgt/Germany
Joined Feb 2003
1,147 Posts
Bill, is the RTH algorithm a secret or how about a logic diagram of the flow for us ?
Like in the manual for the CO2.
regards Volker
mr..speed is offline Find More Posts by mr..speed
Reply With Quote
Old Oct 06, 2009, 06:18 AM
kreu01
Joined Oct 2008
547 Posts
Quote:
Originally Posted by bentel
Bill, is the RTH algorithm a secret or how about a logic diagram of the flow for us ?
Like in the manual for the CO2.
regards Volker
Look here

http://en.wikipedia.org/wiki/PID_controller
kreu01 is offline Find More Posts by kreu01
Reply With Quote
Old Oct 06, 2009, 10:52 AM
KF7JBN
Jhondra's Avatar
United States, ID, Nampa
Joined Nov 2008
845 Posts
Data Logger filter

Bill,

Seems like unless I filter the data from the eLogger on download, it wont refilter. Problem is, I have some data points that I want to filter out but I've already cleared the logger memory. I just have a saved file. Would be nice if you could apply filter settings at "load" time from files too, or just a "refilter" button to apply the filter to the data? Sorry, bit off topic.

-Jhondra
Jhondra is offline Find More Posts by Jhondra
Reply With Quote
Old Oct 06, 2009, 12:50 PM
Registered User
mr..speed's Avatar
near Stgt/Germany
Joined Feb 2003
1,147 Posts
Quote:
Originally Posted by kreu01
In general I understand PID, but this does not declare what I see at my deflection. At least not for me.
Look at my posted video and declare me the motor on/off every 2 sec. very at the end.
http://www.vimeo.com/6770625

at the beginning you can see my settings, finished at 3:12
cruise speed 40
cruise altitude 120
altitude error 40
pitch proportinal gain 18
pitch derivate gain 30
turn prop. limit 12
turn prop. gain 10
turn integral gain 24
turn deriv. gain 23

5:10 the flight starts
9:02 smaller sequence
9:32 on/off
10:27 ==>> on/off longest and best sequence

and then please tell me what to change.
Thanks
mr..speed is offline Find More Posts by mr..speed
Last edited by mr..speed; Oct 06, 2009 at 04:47 PM.
Reply With Quote
Old Oct 06, 2009, 01:18 PM
KF7JBN
Jhondra's Avatar
United States, ID, Nampa
Joined Nov 2008
845 Posts
Quote:
Originally Posted by bentel
In general I understand PID, but this does not declare what I see at my deflection. At least not for me.
Look at my posted video and declare me the motor on/off every 2 sec. very at the end.

at the beginning you can see my settings.
5:10 the flight starts
9:02 smaller sequence
9:32 on/off
10:27 ==>> on/off longest and best sequence

and then please tell me what to change.
Thanks
I still have to prove it out in practice on my own setup, but I'm convinced, given the nature of the slow GPS update response, that people are crippling their RTH by having too high of a derivative gain. I would start this as low as possible (probably never more than HALF the proportional gain). With the derivative gain too high, it is adding in WAY to much response to flakey/too late measurement deltas. Feed back systems are critically sensitive to the return "measurement" path, and if that is crappy (where am I heading now? How 'bout now? Now?) or slow, you HAVE to tune the derivative gain way down...


edit:here is stuff from the wiki PID page that supports what I'm saying above, and should be a generally good way for everyone to test out the RTH and setup the gains (Ki is integral, Kd is derivative, Kp is proportional):

Quote:
If the system must remain online, one tuning method is to first set Ki and Kd values to zero. Increase the Kp until the output of the loop oscillates, then the Kp should be left set to be approximately half of that value for a "quarter amplitude decay" type response. Then increase Ki until any offset is correct in sufficient time for the process. However, too much Ki will cause instability. Finally, increase Kd, if required, until the loop is acceptably quick to reach its reference after a load disturbance. However, too much Kd will cause excessive response and overshoot. A fast PID loop tuning usually overshoots slightly to reach the setpoint more quickly; however, some systems cannot accept overshoot, in which case an "over-damped" closed-loop system is required, which will require a Kp setting significantly less than half that of the Kp setting causing oscillation.
note: I would do all of the above entering RTH heading between 90' and 120' heading from home and enter at just above the nominal elevation and speed and get the heading response dialed in, then try different variations. Once THAT is done, then you need to work on the elevation p/d and speed. and follow up with how it circles your position once it flies over with small tweeks to the derivative after that.

I also am not sure how to control my throttle response a little more "gently" short of limiting the throttle stick limit for climb in the wizard. I don't see any P, I and D gain settings for the speed to feed back into the throttle (your video shows how under damped the throttle response is). Would sure be nice to be able to enter these values in as well...hint, hint, wink, wink.
Jhondra is offline Find More Posts by Jhondra
Last edited by Jhondra; Oct 06, 2009 at 02:12 PM.
Reply With Quote
Old Oct 06, 2009, 01:50 PM
Registered User
Joined Apr 2009
717 Posts
I can confirm now that the "reset level flight" really reset the center maintaining the servo relative positions. I've did it on purpose setting my trims a way off, then i've reset it, and the RTH were working correctly (it DID NOT, when i've tried to engage it with the trims moved *after* the wizard without a reset). Actually i'm pretty confident about it, i just need to adjust to not overshoot the home (but i don't care much, i need it to regain the signal, not to use it as an autopilot), i've also tried it with some wind and it works as expected. Finally i can try my very long range FPV as soon the LRS hit my mailbox, with a certain safety margin.
npole is offline Find More Posts by npole
Reply With Quote
Old Oct 06, 2009, 03:31 PM
I love FPV!
Cralis's Avatar
Eastern side of Australia
Joined Jan 2007
4,650 Posts
Feature request!!

For those of us who have both the GPS and the AirSpeed add on - could you not add a small 'Wind sock' on the screen? I know it can't work out which side the winds coming from - but could you not show the headwind or tailwind speed? That would rock!
Cralis is offline Find More Posts by Cralis
Reply With Quote
Old Oct 06, 2009, 03:49 PM
Registered User
mr..speed's Avatar
near Stgt/Germany
Joined Feb 2003
1,147 Posts
Quote:
Originally Posted by Jhondra
I still have to prove it out in practice on my own setup, but I'm convinced, given the nature of the slow GPS update response, that people are crippling their RTH by having too high of a derivative gain. I would start this as low as possible (probably never more than HALF the proportional gain). With the derivative gain too high, it is adding in WAY to much response to flakey/too late measurement deltas. Feed back systems are critically sensitive to the return "measurement" path, and if that is crappy (where am I heading now? How 'bout now? Now?) or slow, you HAVE to tune the derivative gain way down...


edit:here is stuff from the wiki PID page that supports what I'm saying above, and should be a generally good way for everyone to test out the RTH and setup the gains (Ki is integral, Kd is derivative, Kp is proportional):



note: I would do all of the above entering RTH heading between 90' and 120' heading from home and enter at just above the nominal elevation and speed and get the heading response dialed in, then try different variations. Once THAT is done, then you need to work on the elevation p/d and speed. and follow up with how it circles your position once it flies over with small tweeks to the derivative after that.

I also am not sure how to control my throttle response a little more "gently" short of limiting the throttle stick limit for climb in the wizard. I don't see any P, I and D gain settings for the speed to feed back into the throttle (your video shows how under damped the throttle response is). Would sure be nice to be able to enter these values in as well...hint, hint, wink, wink.
Hello Jhondra,
thanks for your analyses.
I extracted this from the video

cruise speed 40
cruise altitude 120
altitude error 40
pitch proportinal gain 18
pitch derivate gain 30
turn prop. limit 12
turn prop. gain 10
turn integral gain 24
turn deriv. gain 23

what should I change in numbers

and the motor starts and if it comes to 40 it stops at once,
BUT DID YOU LOOK on the altitude I set 120 +- 40 and it stays at around 40,all in km/h and meters, looks like it doesn't care the altitude.
I think speed is a problem too, maybe I should increase the value, when it reaches the speed limit, it doesn't care about the altitude any more.
mr..speed is offline Find More Posts by mr..speed
Last edited by mr..speed; Oct 06, 2009 at 04:08 PM.
Reply With Quote
Old Oct 06, 2009, 04:25 PM
Registered User
Joined Apr 2009
717 Posts
Quote:
Originally Posted by Cralis
Feature request!!

For those of us who have both the GPS and the AirSpeed add on - could you not add a small 'Wind sock' on the screen? I know it can't work out which side the winds coming from - but could you not show the headwind or tailwind speed? That would rock!
Do you mean if the difference between the AS and GS is negative/positive then paint an arrow according...? Ye it would be cool, but i would consider it as a fancy addition (it takes half second to figure if the GS is lower or greater than AS looking at their values), i would prioritize some much important stuff (the list is so looooooooong) before investing time into this. :=)
npole is offline Find More Posts by npole
Reply With Quote
Old Oct 06, 2009, 04:40 PM
I love FPV!
Cralis's Avatar
Eastern side of Australia
Joined Jan 2007
4,650 Posts
Yes, but maybe grow the arrow depending on the head/tail wind. Also, I'd prefer to show only true airspeed, and hide the gps one. If I showed both, I prefer not to do maths while flying. Actually, I prefer not to do maths - period! Hehe. I think it would be handy.
Cralis is offline Find More Posts by Cralis
Reply With Quote
Reply


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
For Sale BRAND New CP Pro/with extra battery stgtech2 Aircraft - Electric - Helis (FS/W) 0 Sep 06, 2006 09:35 PM
New eolo pro glitches with cc60 aschen Electric Heli Talk 24 May 20, 2004 08:12 AM
FS New Piccolo Fun with bearing upgrade and HA tail motor Bud Morrison Aircraft - Electric - Airplanes (FS/W) 1 Mar 21, 2004 06:32 PM
For Sale: NEW Pic Pro rotor head with stiffening mod Eco8gator Aircraft - Electric - Airplanes (FS/W) 2 Jun 02, 2003 09:16 PM