HobbyKing.com New Products Flash Sale
Reply
Thread Tools
Old Apr 11, 2008, 10:31 PM
Registered User
Joined Apr 2008
11 Posts
How do you control yaw with 4 rotors

Thanks, ADI.

You mentioned tilting in the X-axis and Y-axis. The quatrorotor site you provided mentions controlling yaw (Z-axis). How is that done?

BTW, that site is fascinating.

Also, it appears, in that site, that a Kalman filter is not needed. Is that correct?

Thanks.
sthudium is offline Find More Posts by sthudium
Reply With Quote
Sign up now
to remove ads between posts
Old Apr 11, 2008, 11:15 PM
ADI
Registered User
New Zealand
Joined Nov 2004
51 Posts
Hi sthudium,

The website is my own personal website, so thanks for the compliment.
If you go back to my site there are links at the bottom of the page to other Quadrotor builder's sites.
Check out Crac's website, which gives a pictorial explanation of how yaw etc control works. Click on the 'Comment ca marche' (How it works) link.
It's in French unfortunately, but you'll get the idea from the pics.
Basically all you do to enable yaw in one direction or another, is alter the balance between CW and CCW rotor pairs using differential control. So if you decrease CW speed by 20% and increase CCW by 20% the craft will yaw CW.
Vertical lift remains the same because you have neither increased or decreased the 'total' power to the 4 motors.

Re Kalman Filter:
This is generally used in order to fuse the data from 2 or more sources, ie. gyroscope and accelerometer. A simpler 'complementary filter' algorithm can be used instead of Kalman filter.
If all you have is a gyroscope for each axis, then neither Kalman nor Complementary filters are needed.

Sorry for getting a bit 'off topic' here.

Cheers ADI
ADI is offline Find More Posts by ADI
Last edited by ADI; Apr 12, 2008 at 05:06 PM. Reason: .
Reply With Quote
Old Apr 12, 2008, 03:06 AM
Registered User
superchuckles's Avatar
somewhere on the end of the orion spur
Joined Jan 2008
23 Posts
Quote:
Originally Posted by Arthur P.
Hmmmmm... Just tested a modified TP25A type 2 for approx. 4 min at 10A on a motor I had picked up in Hong Kong quite some time ago (no clue what the kV is). At that amp with an 10x4.7APC SF prop it was producing approx. 600g thrust. Motor heated up to about 50C, ESC to 55-60C (measured with my little BMI ThermoGun infrared temp sensor). Interestingly, running the same motor at 15A (about 750g thrust) for about 2 minutes results in the ESC sticking around 45C but the poor little motor clearly approaching destruction with the motor temp shooting up to 60C. And to make this even crazier, running the motor at 6A for 2-3min results in an ESC peaking up to 80C (!!! ) while the motor seems happy at about 35C. What I don't undestand is why an unmodified ESC could be run at 15-20A, destroying the motor with the ESC only getting handwarm, while the modified ESC gets hot at 6 or 10A, but pushing it to 15A seem to be less of a problem for the ESC.

Of course there are static tests, so with the ESC fully in the propwash it should be a little better. Guess I can at least use them for now for some short testflights. But it would be nicer to have them stay nice and cool. So, if the only solution to that is replacing the FETS with another type identified by Quax: replace the TPC8114 (or TPC8111) with Si4425. Not to difficult if you use two solering irons, one on each side, to heat up both rows of pins at the same time, then just shove the old FET off."

Guess I'll have to go after them.

not sure if this may or may not have been partially responsible for the problem of heating but --

i've built several variable voltage power supplies that will supply 50 or more amps, while maintaing voltage to within 1/10th of a volt from 0 to 50 amps (or whatever the max it's made for). switching transistors under heavy load, tend to heat up most, when being switched on & then off & then back on somewhere (depends on trans specs) near 50%. what makes it seem wacky is, when they're delivering the heaviest load, they hardly get warm at all. the reason for this is simpler than one might think --- it is because it doesn't have to shut off power then re-apply from zero as much. an overly simplified explanation to be sure, but essentially correct.

now i know that for esc's, it's going to be switching all the time (duh - 3 phase, no getting around it) but..... perhaps your transistors (i forget what kind of FETs they were) you're using in one ver or another might not be optimized for use in the frequency range you're trying to use them at? if you can find a datasheet (i know, sometimes not always easy) from the manufacturer of your transistors, see what they show. perhaps towerpro got a deal from someone for a bunch of cheap ones that may or may not have met their specs perfectly, but were close enough for them, but when modified for what you guys are doing, the simply fall outside the manufacturers rated useage range?
just a guess - dunno.

only other thing i could think of might be rebounding impedence at specific frequencies from the windings in your motor (similar to what some people confuse "resistance" for in speakers -- resistance & impedence aren't as universally interchangeable as some might think, since frequency has direct impact on how hard it is to push electrons thru a wire). perhaps trying a different manufacturers motor of similar capacity but wound slightly different may result in a completely different result. i still think the FET's are the most likely culprits on the ones overheating at mid-range settings, but just trying to think of any/all situations that might be causing the problem (i'm of course figuring that no mistakes were made doing any of the wiring or programming - i doubt that would be the area 'cause it seems you have that part handled quite well).

btw - i can't think of the manufacturer off the top of my head, but i'll look it up tomorrow & see if i can get you a link to a programming socket for your SM chips. they do exist pre-made, so you don't have to build one from clay (though i know that's a possible solution). if - IF i were to make one myself by hand again.... -ok, here's how i made mine before i found the adapter - mask off all parts of the circuit board other than the chip & pins with blue painters masking tape - completely cover everything but the chip. next, lightly spray the chip with some PAM non stick spray (don't laugh). now, take some bondo & mix up about two tablespoons worth, along with the proper amount of hardner (not too much or it'll turn to a solid block before you can use it). mix quickly, then apply quickly, because even properly mixed proportions will harden very rapidly -- i woud take probably a regular tablespoon - fill it with bondo, make sure the spoon is filled to where it is just filled to the point where there's just a slight buldge over level. flip the spoon upside down & push down over the chip -- push down enough to get bondo just squishing out around the edges of the spoon & stop there. pushing down too much or trying to "roll" the bondo on can force some underneath the pins nearest the processor, which can cause the bondo to end up pitted & leave pieces under those pins. ok - after your bondo hardens, you can then drill holes thru the pin indentations all the way thru the bondo out to the other side & epoxy wires into place once you get them all how you like. you can also use bandsaw or grinder or file or even sandpaper to trim off excess bondo (before putting wires in of course makes it easier, though not absolutely necessary). the advantages of bondo are that you have a permanent and very hard piece of plastic that is an exact fit mold that you can use to program any/all chips of that type very quickly. you can also if you want to make a very neat & professional looking program interface, after done with making it fit the chip, make up a short harness & socket, etc --- then cut a small hole in a small cardboard box & stick the facing part of the socket adapter you've made thru it, & hole in the opposite side sticking the interface adapter thru it - then fill box with bondo from yet another hole in another side - making a solid block that's got your chip adapter on one side, interface adapter on the other - that you'll never have to worry about shorting wires or flexing & breaking them. after bondo hardens, you can also sand the block to any shape you want, smooth it all out - & paint it to make it all pretty. i used to make stuff like that all the time back in the day when i was trying to keep cost at or near zero using stuff i had around the house. anyhow, good luck & good work on this.
superchuckles is offline Find More Posts by superchuckles
Reply With Quote
Old Apr 12, 2008, 06:13 PM
Registered User
Joined Apr 2008
11 Posts
PID tuning

Thanks, again ADI. You are quite illuminating and get to the important issues.

Now, another question, please. How do you go about tuning a PID controller for a quad rotor?

Is this all explained in another thread?

Thanks!!
sthudium is offline Find More Posts by sthudium
Reply With Quote
Old Apr 12, 2008, 06:36 PM
Registered User
brashley's Avatar
Joined Dec 2006
517 Posts
Quote:
Originally Posted by sthudium
Thanks, again ADI. You are quite illuminating and get to the important issues.

Now, another question, please. How do you go about tuning a PID controller for a quad rotor?

Is this all explained in another thread?

Thanks!!
I second that!!!! I have been asking and all I get is ignored or told that it takes trial and error to understand.....

Richard
brashley is offline Find More Posts by brashley
Reply With Quote
Old Apr 12, 2008, 08:02 PM
ADI
Registered User
New Zealand
Joined Nov 2004
51 Posts
We're really getting off topic now, so here's a link to a really good explanation of PID control.
http://www.embedded.com/2000/0010/0010feat3.htm
There are a million other sites out there which deal with the topic also.
PID tuning via trial and error is not such a bad way of tuning, because it does help familiarise yourself with the dynamics of the machine and the effects that adjusting PID parameters has on it.

Cheers ADI
ADI is offline Find More Posts by ADI
Reply With Quote
Old Apr 13, 2008, 03:19 AM
Registered User
Hoofddorp, Netherlands
Joined Sep 2004
3,917 Posts
@superchuckles, Not sure investigating the reasons for the overheating of the TP25A/2-s is worth too much time. It certainly has to do with the FETs used on that series and it certainly looks like TP used those for a short period of time, most likely because they looked like a good deal. Bernard (Quax) has identified alternative FETs that don' overheat. And the latest 2 orders I placed resulted in TP25A/3-s which have different FETs again which don't get warm.

@adi, thanks for that discussion and link on PID control. Might be a bit lost in this thread though. It really deserves a thread of its own, may even just "Multikopter PID control theory and practice" here in the DIY electronics forum.
Arthur P. is offline Find More Posts by Arthur P.
Reply With Quote
Old Apr 13, 2008, 03:07 PM
Registered User
Hoofddorp, Netherlands
Joined Sep 2004
3,917 Posts
LIttle update on use of the TowerPro 25A/2s (with the hot FETs): I flew with some TowerPro 25A/2s on Turnigy 2217/20 motors on 3S1P Lipos with APC 12x3.8SF props today, RTF weight about 1.7kg. With those props at full throttle the motors on the testbench pull about 18A at 1/2-2/3 throttle they should be doing about 10A. Flighttime on a 3650mAh battery was about 8 to 9 minutes (forgot to start the timer on takeoff on both flights, and had to abort the first due to a motor suddenly seizing up due to a mechanical problem). Temperature was about 9C and the ESCs were fullyin the propwash. After the flight the motors were not warm at all and the ESCs barely handwarm. Used that way on a medium weight APMk, these ESCs don't seem to cause problems.

Of note also flew with TowerPro 25A/3 in a more or less similar configuration (GWS 10x6 3bladed props, rest the same), and those did not even get handwarm.
Arthur P. is offline Find More Posts by Arthur P.
Reply With Quote
Old May 12, 2008, 04:51 PM
Registered User
Hoofddorp, Netherlands
Joined Sep 2004
3,917 Posts
Converted 8 more TowerPro 25A/3's and 4 TowerPro 50A's over the last 3 days. Down to about 40 minutes for the 25As and about 30min for the 50As which are easier due to more space.

One tip with respect to soldering wires to the processor pins: you want some solder on the pins first. If you heat the pin and solder at the same time there's a good chance you'll regularly create bridges between pins. I-ve found that first heating the pin and then just shortly touching the pin and soldering iron tip with the solder tends to result in just enough solder and only on the heated pin.

Also pretin the wire a bit. That increases te chance of a long and strong bond with just a short application of the point of the soldering iron to wire pressed on pin.

And another tip: if you use wirewrap wire it is difficult to create bare ends on very short wires by stripping. But circling the tip of the soldering iron around the end of the wire which you want bare, you can melt away just enough isolation. That way it becomes possible to create very short shunts (1cm or so).
Arthur P. is offline Find More Posts by Arthur P.
Reply With Quote
Old Jul 11, 2008, 04:16 PM
Registered User
Joined Jun 2008
9 Posts
Hi!

Im wondering how a i2c command looks like.. in code that is.

Regards Niklas
niklascooke is offline Find More Posts by niklascooke
Reply With Quote
Old Jul 12, 2008, 10:57 AM
Registered User
brashley's Avatar
Joined Dec 2006
517 Posts
See http://svn.mikrokopter.de/mikrowebsv...%2Ftwimaster.c

Richard
brashley is offline Find More Posts by brashley
Reply With Quote
Old Jul 12, 2008, 01:54 PM
Registered User
Joined Jun 2008
9 Posts
What are the adresses of the updated Tower pro 25a when you have modified them?
Can you change them?
And ive heard that you should have two motors spinning cw and the other two spinning ccw but the air is blown upwards if i change the direction of the motor and it doesnt matter if i turn the prop upside down.. hmm ???
niklascooke is offline Find More Posts by niklascooke
Reply With Quote
Old Jul 12, 2008, 03:48 PM
Registered User
brashley's Avatar
Joined Dec 2006
517 Posts
You need counter rotating props (like the EPP1045 http://www.maxxprod.com/mpi/mpi-29a.html)

Richard
brashley is offline Find More Posts by brashley
Reply With Quote
Old Jul 13, 2008, 05:19 PM
Registered User
Joined Jun 2008
9 Posts
Hi again!

Whats the best brand that comes in both regular and counter rotating props?
I need props maybe in sizes 8x4

regards
niklascooke is offline Find More Posts by niklascooke
Reply With Quote
Old Jul 14, 2008, 07:12 AM
Registered User
brashley's Avatar
Joined Dec 2006
517 Posts
Not many options available. The above link has some 8x4.5 ones. You will have to take what you can find with counter rotating props, just scan the internet.
brashley is offline Find More Posts by brashley
Reply With Quote
Reply


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
2 Electro Fusion Motors & CC esc's from Coronas for sale mthomson Aircraft - Electric - Airplanes (FS/W) 7 Mar 18, 2004 06:13 AM
FS- FMA M5 dual conv Rx and 2 esc's- all NIB Etonic Aircraft - Electric - Airplanes (FS/W) 3 Nov 17, 2003 01:27 PM
How do I hook 2 esc's into 1 receiver ? Al P Power Systems 2 Aug 30, 2002 05:38 PM
EBAY Brushless Motor & 2 ESC's S Anastasio Aircraft - Electric - Airplanes (FS/W) 0 Aug 12, 2002 06:41 AM
WTB: 2 Hitec Feather Recievers and 2 ESC's Ryan Nau Aircraft - Electric - Airplanes (FS/W) 6 Jul 03, 2002 07:13 PM