|
-
POSTING RULES

-
Donate yearly (please).
-
Advertise in here!
-
Today's Posts
|
Insert Pics
|

09-16-2010, 02:20 PM
|
 |
VAF Moderator / Line Boy
|
|
Join Date: Jan 2005
Location: Dayton, NV
Posts: 12,256
|
|
Sawtooth Cruise (and other autopilot Musings?)
I was solo on this trip, Louise being otherwise occupied on the Gulf Islands National Seashore managing the oil spill cleanup efforts. She was going to join me in Reno (via SWA) at the end of the week, but I figured that since I had already freed up my schedule, I?d go out first. It?s wonderful having her support - she even suggested I go for a day of soaring down near Palomar, something I haven?t done for over a year. It?s been a while since I?ve done a long cross-country alone in the Val, so this trip gave me a chance to investigate a few improvements that have been made in the EFIS and autopilot software in recent months.
With smooth air, no weather and low winds, there wasn?t much to the task of flying, so I let the electronics take us westward. Two things piqued my interest along the way, one that I had noticed briefly before, and another that I hadn?t really thought about much. The first was an interesting pitch behavior that I am noticing now that the GPSS/GPSV gains between the GRT EFIS and the TT autopilot have been worked out. There was a strong tendency toward oscillations in pitch with the previous software, especially at aft CG?s, and that had precluded much evaluation work in that channel. But now that it is working well, I noticed a very subtle behavior which I nicknamed the ?Saw Tooth Cruise?.
Anyone that has done a good set of performance tests in Phase 1 should be familiar with the ?Saw Tooth Climb? ? a method of determining a lot of performance data quickly with a series of climbs and descents that look like a saw blade in profile. Well, the Saw Tooth Cruise is an odd behavior that is easy to miss if you?re not looking for it. Once you are settled down in cruise with GPSV holding altitude, it manifests itself as a very small, slow pitch in one direction, followed by a quick pitch back the same amount, but in the opposite direction. It is like a slow drift, followed by a quick correction ? and we are talking far less than a degree of pitch. In fact, if you have even mildly rough air, it would be completely masked by the chop. After watching this for awhile, I gave a blip of trim, and the behavior settled down. I then played with taking the trim a little farther in that same direction, and managed to get the saw teeth in the opposite direction ? whereas at first, the slow drift was a pitch down, followed by a quick pitch up, now the slow drift was upward in pitch. All the time, the airplane was trimmed close enough to neutral so that the autopilot was not calling for any trim action.
My theory, after watching this awhile, is that if the airplane isn?t perfectly in trim, it will tend to drift within the ?dead band? of the autopilot. Any measurement system has a threshold below which, it doesn?t know anything has happened. I think this is the slow drift. When it finally reaches a point where the system recognizes the deviation, it takes quick action to correct it. Very normal behavior, and a reason to get things in close to perfect trim before turning the flying over to ?George?- or sweetening it up once he?s on the job. That?s my theory ? I am ready for alternative ideas!
The other thing I played with on the trip was climb modes. The GRT EFIS will command the autopilot to climb, and you can have it work in one of two modes ? constant Vertical Speed mode, or Constant Indicated Airspeed mode. Generally speaking, I have always used the vertical velocity ? I can do the math for the distance I have to climb (or descend), and figure out the rate I need to hit my target altitude. Dial that in and watch it go. Jets frequently have an ?airspeed climb? mode, and I have tried to figure out how that might be useful in our applications. Certainly, if you have an airplane in a very long (high) climb in vertical speed mode, it will keep pitching up trying to maintain the rate, and unless it has stall protection (the GRT/TT does), it can get very exciting up high! So climbing on speed is a bit safer in that regard. But today, I found yet another good use for it ? engine cooling.
Climbing out of Casa Grande (south of Phoenix) with ground temps around 100 degrees, and an engine already warm from a long day of flying, there was no way I was going to get away with a 1500 fpm climb. Even 500 fpm was giving me an airspeed that yielded insufficient cooling, judging by the oil temps. But I knew that 130 KIAS seemed to be lowering the temps, so I switched the EFIS to IAS Climb mode, and took whatever rate I got. I had 100 miles to get to altitude for my mountain approach into Big Bear Lake, so I didn?t care how long it took. That worked really well ? the engine was happy, the airplane was stable ? and the pilot was happy because all he had to do was watch!
Smooth air on a long cross-country is rare, and it was good to take advantage of it to learn some more about this constantly-evolving software that we are getting in the Experimental market. I try to thoroughly investigate each new drop to see if they have any bad corners that I?d rather know about in advance. At this point, the GRT/TT combo is working really well ? it will be interesting to see what they come up with next!
Paul
__________________
Paul F. Dye
Editor at Large - KITPLANES Magazine
RV-8 - N188PD - "Valkyrie"
RV-6 (By Marriage) - N164MS - "Mikey"
RV-3B - N13PL - "Tsamsiyu"
A&P, EAA Tech Counselor/Flight Advisor
Dayton Valley Airpark (A34)
http://Ironflight.com
|

09-23-2010, 02:00 PM
|
 |
|
|
Join Date: Jan 2005
Location: Yorkshire, England
Posts: 2,052
|
|
What is interesting is that you notice the sawtooth characteristics stops with input of trim.
The autopilots on the Airbus use the trim as the pitch stabiliser in the cruise - i.e. in reverse of what our basic autopilots are doing.
What would be sweet would be an input from the autopilot to run the trimmer rather than feed a servo that moves the elevator.
I guess it wouldn't be too difficult but then you would get airliner smoothness in the cruise and a lot less work for the autopilot servo.
|

09-23-2010, 02:08 PM
|
 |
|
|
Join Date: Mar 2005
Location: Elkhart, Indiana
Posts: 1,186
|
|
That's funny, Paul. I noticed the same effect in roll on my flight last week down to Lambert's Cafe (yes, home of the "throwed rolls"). I only noticed it in roll because I, unlike you, trimmed my plane before engaging the alt hold.
However, as the gas was slurped from one tank, necessitating trim correction, the TruTrak AP would attempt to do it for me. I played with the trim to figure out if I could stop the oscillations by adjusting trim, and indeed I could.
I'm still fighting a slight heavy right wing anyway, so that prompted me to do something about it this weekend.
Next time, Paul, trim the stupid plane BEFORE you hit the AP button ... 
__________________
Don McNamara
Peoria, AZ
Builder: RV-8 "Smokey"
|

09-23-2010, 02:16 PM
|
 |
|
|
Join Date: Jan 2005
Location: Louisville, Ga
Posts: 7,840
|
|
Sawtooth cruise vs. dutch roll XC in an Airbus?
Last December, my wife and I flew in an Airbus mailing tube, out to California, and the darn thing had the most annoying slight Dutch roll the whole way....very annoying but it would roll a couple degrees right, hesitate, then roll a couple degrees left...for thousands of miles!
The proud front crew were standing, thanking everyone as they de-planed and I hesitated and told the Captain, "Sir, don't be offended but I'm a CFI and your airplane did Dutch rolls all the way out here. The autopilot needs adjusting, I think." To which he replied, "I didn't think anyone would notice."
Best,
__________________
Pierre Smith
RV-10, 510 TT
RV6A (Sojourner) 180 HP, Catto 3 Bl (502Hrs), gone...and already missed
Air Tractor AT 502B PT 6-15 Sold
Air Tractor 402 PT-6-20 Sold
EAA Flight Advisor/CFI/Tech Counselor
Louisville, Ga
It's never skill or craftsmanship that completes airplanes, it's the will to do so,
Patrick Kenny, EAA 275132
Dues gladly paid!
|

09-23-2010, 02:36 PM
|
|
|
|
Join Date: Jul 2005
Location: lexington, KY
Posts: 330
|
|
I wonder...
If Trutrak's Auto Trim would eliminate the "sawtooth bounce"--Lucas, are you there?
__________________
Thanks,
Jarvis
Purchased flying RV-7A 
RV-7A emp finished, wings next.
|

09-23-2010, 03:09 PM
|
|
Super Moderator
|
|
Join Date: May 2005
Location: Locust Grove, GA
Posts: 2,626
|
|
Been watching this for years. What I think happens is that the altitude needs to drift to a point that the s/w can send the minimum rate of climb to the a/p, such as 100 fpm. So, it drifts down or up a bit, then corrects back. It's been tightened up a lot between the Chelton and the Sorcerer and the AFS4500 and the Sorcerer over the years, and now we see maybe 20-40 feet drift.
The sad thing is that we probably never even noticed this stuff on our old round gauges, but with these digital tapes flashing every 10 feet it makes us think we need to do something about it.
Vic
__________________
 Vic Syracuse
Built RV-4, RV-6, 2-RV-10's, RV-7A, RV-8, Prescott Pusher, Kitfox Model II, Kitfox Speedster, Kitfox 7 Super Sport, Just Superstol, DAR, A&P/IA, EAA Tech Counselor/Flight Advisor, CFII-ASMEL/ASES
Kitplanes "Unairworthy" monthly feature
EAA Sport Aviation "Checkpoints" column
EAA Homebuilt Council Chair/member EAA BOD
Author "Pre-Buy Guide for Amateur-Built Aircraft"
www.Baselegaviation.com
|

09-23-2010, 05:44 PM
|
 |
VAF Moderator / Line Boy
|
|
Join Date: Jan 2005
Location: Dayton, NV
Posts: 12,256
|
|
Quote:
Originally Posted by N8RV
Next time, Paul, trim the stupid plane BEFORE you hit the AP button ... 
|
Ahhh....but this autopilot climbs to a preset altitude and levels off. You engage it at liftoff, and don't touch the stick until flair.....it's terrible when it spills your capucino....  
__________________
Paul F. Dye
Editor at Large - KITPLANES Magazine
RV-8 - N188PD - "Valkyrie"
RV-6 (By Marriage) - N164MS - "Mikey"
RV-3B - N13PL - "Tsamsiyu"
A&P, EAA Tech Counselor/Flight Advisor
Dayton Valley Airpark (A34)
http://Ironflight.com
|

09-23-2010, 06:52 PM
|
 |
been here awhile
|
|
Join Date: Dec 2004
Location: North Alabama
Posts: 4,301
|
|
Quote:
Originally Posted by Ironflight
Ahhh....but this autopilot climbs to a preset altitude and levels off. You engage it at liftoff, and don't touch the stick until flair.....it's terrible when it spills your capucino....  
|
It is amazing how quickly you get spoiled to a full-featured two-axis autopilot! The Trio ProPilot in my RV-6 works so nicely (this is not a "my autopilot is better than your autopilot" post) that I use the thing for casual flights around the area just to watch it work. Yep, engage the system shortly after take-off and don't touch the stick until short final.
Is the sawtooth behavior a function of the interface between the GRT equipment and the autopilot? In smooth air the Pro holds pitch so tightly I can't detect any movement in the Dynon altimeter (well, less than five feet, anyway) but it's running independent from any flight instruments. Maybe I just lucked out and got a favorable gains setup. Our experimental equipment has gotten really, really good!
Last edited by Sam Buchanan : 09-24-2010 at 07:35 AM.
|

09-24-2010, 09:28 AM
|
|
|
|
Join Date: Jan 2005
Location: northern Virginia (DC area)
Posts: 198
|
|
Sounds like a limit cycle
Quote:
Originally Posted by Ironflight
My theory, after watching this awhile, is that if the airplane isn?t perfectly in trim, it will tend to drift within the ?dead band? of the autopilot. Any measurement system has a threshold below which, it doesn?t know anything has happened. I think this is the slow drift. When it finally reaches a point where the system recognizes the deviation, it takes quick action to correct it.
|
It sounds like you have a classic example of a limit cycle going back and forth over a deadband boundary. I suspect the deadband is there to avoid persistent servo inputs when the elevator is close to its trim deflection and the aircraft is at its commanded flight condition.
I don't think it's an issue of being below the quantization threshold of the altitude, vertical speed, or pitch measurement. If it is a quantization effect, perhaps it's the elevator servo's deflection encoder or the stepper motor's step size (assuming you have stepper motor driven servos).
What was the scale of the oscillation? How much altitude, pitch, etc are we talking about?
__________________
Alex Roup
RV-12 (empennage) #120407
Northern VA
|

09-24-2010, 09:56 AM
|
 |
|
|
Join Date: May 2005
Location: Canby, Oregon
Posts: 1,786
|
|
I think you need to tune it some more.
I don't see this on my AFS 3500 / TT setup.
It might drift off +- 10 ft, but the corrections are very minor.
I do see sawtooth speed from the AP when going through MT wave air.
I wish that there was a mode to select constant speed and let alt drift. I found it more comfortable to hand fly the plane going over some of the Rockies.
Kent
__________________
Kent Byerley
RV9A N94KJ - IO320, CS, tipup
AFS 3500, TT AP, FLYING....
Canby, Or
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT -6. The time now is 01:36 AM.
|