Autopilot question
Join Date: Feb 2006
Location: Frozen North
Posts: 53
Likes: 0
Received 0 Likes
on
0 Posts
An outbreak of harmony??
Maybe I'm dreaming.
It would appear that HeliComparitor and Nick Lappos have contributed to the same thread and not only not fallen out but agreed with each other!!
OA
It would appear that HeliComparitor and Nick Lappos have contributed to the same thread and not only not fallen out but agreed with each other!!
OA
HC
The 330G autopilot was single channel with big switches, just to jog your memory. The 330C had (probably still has) press buttons like the 332 but still single channel.. The nice thing about it was that it had was a yaw trim, a recessed wheel that would trim out the yaw in the cruise correcting the annoying habit some Pumas have of flying half-a-ball out. If you were really bored you could trim it out in the hover and do spot turns with it.
Then there is the full SRIM 4 axis autopilot, that's something you can play about with for hours.
The 330G autopilot was single channel with big switches, just to jog your memory. The 330C had (probably still has) press buttons like the 332 but still single channel.. The nice thing about it was that it had was a yaw trim, a recessed wheel that would trim out the yaw in the cruise correcting the annoying habit some Pumas have of flying half-a-ball out. If you were really bored you could trim it out in the hover and do spot turns with it.
Then there is the full SRIM 4 axis autopilot, that's something you can play about with for hours.
HC, I'd say your description of the FD/AP relationship is reversed: the FD tells the AP what to do, or the pilot, not the AP tells the FD what to do.
ECF do not use FDs so their functionality is incorporated into the AP (in a more appropriate way, I think), however with a/c that have separate FDs, they are independent and can function autonomously from the AFCS system. Take a basic S-76 A model, for instance, with a phase 2 SCAS: no coupling at all, but all the FD info for the pilot.
ECF do not use FDs so their functionality is incorporated into the AP (in a more appropriate way, I think), however with a/c that have separate FDs, they are independent and can function autonomously from the AFCS system. Take a basic S-76 A model, for instance, with a phase 2 SCAS: no coupling at all, but all the FD info for the pilot.
Join Date: Sep 2004
Location: Australia
Age: 52
Posts: 698
Likes: 0
Received 0 Likes
on
0 Posts
Well, I have a CPL (H) and I never knew ANY of that ****. 
So there you go, Tonic Please - it sounds like heaps of us have benefited from your "stupid question".
Keep it up, I could definitely do with learning more!
(and Yes, I DO google, but with the 1,657,875,998 pages that you get in response, I can certainly see why he came in here and asked.)
So there you go, Tonic Please - it sounds like heaps of us have benefited from your "stupid question".
Keep it up, I could definitely do with learning more!
(and Yes, I DO google, but with the 1,657,875,998 pages that you get in response, I can certainly see why he came in here and asked.)
OK then, for a 'proper' autopilot question - is anyone familiar with the term 'attitude clamp' regarding stab/autopilot systems?
There are no limits
Yes, Sir. I refer the Honourable Gentleman to my previous post.
Off to make another cup of credibility. (Which aisle is that in Sainsbury's?)
Off to make another cup of credibility. (Which aisle is that in Sainsbury's?)
212 my experience of FDs is limited to the AS332L2, which was as I described. No doubt there are other ways to skin the cat and it could be as you describe - it depends on your terminology I suppose. Something has to move the command bars and whether you want to call that bit the FD or the AP probably depends on the manufacturer.
With my L2 heritage I call the FD just the bit the pilot sees on the AI, but it could be that the FD is the clever bit that works out how to position the command bars as well, and the AP is the dumb bit that can couple that info to the flight controls. Personally I would not call that bit an AP because its not doing anything clever, but as you say of course another manufacturer can use different conventions, and in the S76 case you cite it sounds as though the AP does the SAS bit and the FD does the ASE & "upper mode" bit?
And yes WL, it looks like I have to humbly withdraw part of my comment on your post
HC
With my L2 heritage I call the FD just the bit the pilot sees on the AI, but it could be that the FD is the clever bit that works out how to position the command bars as well, and the AP is the dumb bit that can couple that info to the flight controls. Personally I would not call that bit an AP because its not doing anything clever, but as you say of course another manufacturer can use different conventions, and in the S76 case you cite it sounds as though the AP does the SAS bit and the FD does the ASE & "upper mode" bit?
And yes WL, it looks like I have to humbly withdraw part of my comment on your post
HC
crab - no, but it sounds like the ASE-type thing I was referring to - ie you fly around with the stick trim release pressed, manoeuvre to a pitch and roll attitude, and then let go. The autopilot will then retain that pitch and roll. Is that what you are talking about (in your usual superior military style)?
HC
HC
To clarify - on most autopilot systems there is a position transmitter or force sensing link to tell the autopilot not to oppose human pilot input. However, on the Sea King 3A there appears to be an element in the system which is referred to as an attitude clamp that seems to require a specific level of demand before it allows the pilot to override the autopilot.
This is most evident in the hover (not with cyclic modes coupled) where a small cyclic input doesn't actually change the disc attitude - so a larger input is required which, when the autopilot finally concedes control, results in a larger movement than was required. For the most part this is easy to work round but for precision work like deck winching can be offputting, especially when you are working hard - night for example.
This attitude clamp may be normal fare for autopilot systems but our documentation doesn't cover it - any expert knowledge out there?
This is most evident in the hover (not with cyclic modes coupled) where a small cyclic input doesn't actually change the disc attitude - so a larger input is required which, when the autopilot finally concedes control, results in a larger movement than was required. For the most part this is easy to work round but for precision work like deck winching can be offputting, especially when you are working hard - night for example.
This attitude clamp may be normal fare for autopilot systems but our documentation doesn't cover it - any expert knowledge out there?
Thread Starter
Join Date: Jul 2002
Location: Hungary
Age: 39
Posts: 684
Likes: 0
Received 0 Likes
on
0 Posts
Well, I'm glad I have got a debate going. It's all very interesting. Where would I have found this on Google?
Oh, I know.. It would have been a link to PPRUNE!!
Join Date: Apr 2003
Location: USA
Age: 75
Posts: 3,012
Likes: 0
Received 0 Likes
on
0 Posts
Crab,
I helped develop 2 generations of Sikorsky flight controls, and have intimate knowledge of these things.
The older SAS systems (where your specific comment resides) had no stick canceler system, so that the stability system actually "fought" all disturbances when the trim is engaged and the stick is pushed against trim. The CG trim wheel centers the inner loop actuators to help retain optimum control, and to minimize hardovers.
The S-76 SAS resists your stick motions, too, and it is not at all unpleasant, since the net feel is a soft, but responsive cyclic. In effect, the stick is less sensitive than normal, since the sas desensitizes it. The system was designed to require you to press the trim release button when you move the cyclic, this removes the attitude stability, at least, and reduces the "resistance" to a small rate damping.
One early experimental version of the S76 SAS II used stick movement switches to remove the attitude hold in the given axis when you moved the stick against trim, and when you returned the stick, the system would lock on the new attitude quickly. This command stick steering was so pleasant to fly that we tried to get the FAA to buy it, but the classic (and nearly worthless) stick gradients that the FAR/JAR require stood in the way, since this system had no gradients of any kind. We junked it a few weeks later. Interestingly, this command stick steering concept is precisely what the fly-by-wire sticks now do!
Later systems use stick motion transducers that tell the autopilot that you have commanded a rate, and so the autopilot allows that rate (it "cancels" the resistance to the detected attitude change, thus "stick canceler")
More modern systems have a full model of the helo inside their little heads, and they compute what a given stick motion should do and also what it should not do. The autopilot then gives a pure output that models the ideal helicopter. An example - all helos are cross coupled, so that a pure forward stick always produces a roll. The model following control system knows this, and automatically puts in the correct roll stick when you make a pure forward stick input. Result - you think the helo has pure sweet controls. These "model following" laws are now state of the art.
I helped develop 2 generations of Sikorsky flight controls, and have intimate knowledge of these things.
The older SAS systems (where your specific comment resides) had no stick canceler system, so that the stability system actually "fought" all disturbances when the trim is engaged and the stick is pushed against trim. The CG trim wheel centers the inner loop actuators to help retain optimum control, and to minimize hardovers.
The S-76 SAS resists your stick motions, too, and it is not at all unpleasant, since the net feel is a soft, but responsive cyclic. In effect, the stick is less sensitive than normal, since the sas desensitizes it. The system was designed to require you to press the trim release button when you move the cyclic, this removes the attitude stability, at least, and reduces the "resistance" to a small rate damping.
One early experimental version of the S76 SAS II used stick movement switches to remove the attitude hold in the given axis when you moved the stick against trim, and when you returned the stick, the system would lock on the new attitude quickly. This command stick steering was so pleasant to fly that we tried to get the FAA to buy it, but the classic (and nearly worthless) stick gradients that the FAR/JAR require stood in the way, since this system had no gradients of any kind. We junked it a few weeks later. Interestingly, this command stick steering concept is precisely what the fly-by-wire sticks now do!
Later systems use stick motion transducers that tell the autopilot that you have commanded a rate, and so the autopilot allows that rate (it "cancels" the resistance to the detected attitude change, thus "stick canceler")
More modern systems have a full model of the helo inside their little heads, and they compute what a given stick motion should do and also what it should not do. The autopilot then gives a pure output that models the ideal helicopter. An example - all helos are cross coupled, so that a pure forward stick always produces a roll. The model following control system knows this, and automatically puts in the correct roll stick when you make a pure forward stick input. Result - you think the helo has pure sweet controls. These "model following" laws are now state of the art.
Join Date: Nov 2004
Location: Cambridgeshire, UK
Posts: 1,334
Likes: 0
Received 0 Likes
on
0 Posts
Nick, do you know any papers on this written by Sikorsky? I'd like to learn a bit more about the design of these systems. In particular i'm curious how much of the model following system relies on rate gyros to correct either the model or helicopter, and how much hub flapback torque controls stick dihedral position.
Last edited by Graviman; 27th Sep 2007 at 14:19.
Join Date: Apr 2003
Location: USA
Age: 75
Posts: 3,012
Likes: 0
Received 0 Likes
on
0 Posts
Grav,
The American Helicopter Society has an annual forum where the various designers strut their stuff, with excellent technical papers and experimental work.
I suggest that you go to your nearest tech library resource and search the Forum publications for the last 15 years or so. The standard texts have a bit on this stuff (Prouty, Padfield) but I don't know that they peel too far back into the details.
The rotor characteristics are part of the "plant model" that is in the model following system, that is, the REAL behavior is cataloged by extensive simulator and flight test ("system identification" in the parlance of the controls engineer). When the model following system crunches out what the controls must do to make the machine obey the pilots request, it must subtract everything that the "plant" does that is wrong, so an accurate understanding of all the characteristics of the helo, main rotor, tail rotor, tail surfaces, engine characteristics are all part of the plant model. With this type of feed forward, the controls don't wait for a wrong motion to occur, they stomp out the improper response before it starts. We do the same thing daily, when we learn to adjust windage on a shot, or when we put a little more on the ball to account for some variable.
wikipedia has a nice set of write-ups, I might add this stuff to them. Note that controls are controls, thermostats have a lot in common with autopilots! I often use thermostats are a way to discuss how feedback controls work.
http://en.wikipedia.org/wiki/Model_predictive_control
Here is a web page by Joe Horn, who is a good controls designer (we worked together a bit on Comanche). He describes a model following control system, but unfortunately, he is a bit too advanced for where this discussion is. You will find most of the papers in the AHS data base similarly esoteric!
http://www.personal.psu.edu/faculty/j/f/jfh19/dmc.html
Here is the whole paper:
http://www.personal.psu.edu/dob104/papers/mfcdmc.pdf
and another. Fig 5 is a bit better for our purposes, the boxes to the right of the "ideal" (which is the model that the controls try to follow) are the subtractive plant model elements:
http://www.personal.psu.edu/users/d/...airwakemfc.pdf
The American Helicopter Society has an annual forum where the various designers strut their stuff, with excellent technical papers and experimental work.
I suggest that you go to your nearest tech library resource and search the Forum publications for the last 15 years or so. The standard texts have a bit on this stuff (Prouty, Padfield) but I don't know that they peel too far back into the details.
The rotor characteristics are part of the "plant model" that is in the model following system, that is, the REAL behavior is cataloged by extensive simulator and flight test ("system identification" in the parlance of the controls engineer). When the model following system crunches out what the controls must do to make the machine obey the pilots request, it must subtract everything that the "plant" does that is wrong, so an accurate understanding of all the characteristics of the helo, main rotor, tail rotor, tail surfaces, engine characteristics are all part of the plant model. With this type of feed forward, the controls don't wait for a wrong motion to occur, they stomp out the improper response before it starts. We do the same thing daily, when we learn to adjust windage on a shot, or when we put a little more on the ball to account for some variable.
wikipedia has a nice set of write-ups, I might add this stuff to them. Note that controls are controls, thermostats have a lot in common with autopilots! I often use thermostats are a way to discuss how feedback controls work.
http://en.wikipedia.org/wiki/Model_predictive_control
Here is a web page by Joe Horn, who is a good controls designer (we worked together a bit on Comanche). He describes a model following control system, but unfortunately, he is a bit too advanced for where this discussion is. You will find most of the papers in the AHS data base similarly esoteric!
http://www.personal.psu.edu/faculty/j/f/jfh19/dmc.html
Here is the whole paper:
http://www.personal.psu.edu/dob104/papers/mfcdmc.pdf
and another. Fig 5 is a bit better for our purposes, the boxes to the right of the "ideal" (which is the model that the controls try to follow) are the subtractive plant model elements:
http://www.personal.psu.edu/users/d/...airwakemfc.pdf
Last edited by NickLappos; 27th Sep 2007 at 15:50.
Nick - the older SAS system is what the Mk3 Sea King has, a simplex system with limited authority requiring pitch and roll trim wheels to keep everything as central as possible. There is a control position transmitter for the cyclic to stop the system opposing pilot inputs.
On the 3A, the ASE computer is allowed acces to the pitch and roll beeper trims on the hyd pack, thus removing the need for trim wheels as it can always beep the cyclic back into authority.
There are FSLs in the control runs to stop the ASE fighting the pilot and the dead zone that these give has been dramatically reduced in the last few years by reducing the tolerances of the microswitches in them.
The problem is that just making the FSLs by moving the cyclic isn't always enough to change the attitude - I believe the 'attitude clamps' are a further element which has to be overcome by sufficient demand (cyclic movement) before the ASE computer gives up control (briefly).
The handling qualities change dramatically when the auto-trim element is bypassed by selecting manual trim - this removes the ASE computer from the equation and reverts to a more Mk 3 mode.
On the 3A, the ASE computer is allowed acces to the pitch and roll beeper trims on the hyd pack, thus removing the need for trim wheels as it can always beep the cyclic back into authority.
There are FSLs in the control runs to stop the ASE fighting the pilot and the dead zone that these give has been dramatically reduced in the last few years by reducing the tolerances of the microswitches in them.
The problem is that just making the FSLs by moving the cyclic isn't always enough to change the attitude - I believe the 'attitude clamps' are a further element which has to be overcome by sufficient demand (cyclic movement) before the ASE computer gives up control (briefly).
The handling qualities change dramatically when the auto-trim element is bypassed by selecting manual trim - this removes the ASE computer from the equation and reverts to a more Mk 3 mode.
Join Date: Nov 2004
Location: Cambridgeshire, UK
Posts: 1,334
Likes: 0
Received 0 Likes
on
0 Posts
Thanks for the links, Nick. That info is spot on what i'm after. It's clear helicopter control systems are about as state-of-the-art as FBW gets.
Wikipedia is getting better on engineering articles like helicopter design. The trouble is that like most topics in engineering there is often a great deal to understand before you can document it. Also there are many good texts like Prouty, Newman, and Coyle which invariable become the reference. I'm suprised you don't document your experience - that would be an interesting read.
Wikipedia is getting better on engineering articles like helicopter design. The trouble is that like most topics in engineering there is often a great deal to understand before you can document it. Also there are many good texts like Prouty, Newman, and Coyle which invariable become the reference. I'm suprised you don't document your experience - that would be an interesting read.
Join Date: Apr 2003
Location: USA
Age: 75
Posts: 3,012
Likes: 0
Received 0 Likes
on
0 Posts
Crab,
That Sea King system is a home-grown British one from back when men were men, and so I don't have much info on it. The same folks who built that also did the tail rotor with an extra blade!
What you mention is a classic issue with older autopilots - either George is flying or you are. The systems I helped work on were always "fly through" autopilots where the pilot can gracefully enter the control loop at any time.
That Sea King system is a home-grown British one from back when men were men, and so I don't have much info on it. The same folks who built that also did the tail rotor with an extra blade!
What you mention is a classic issue with older autopilots - either George is flying or you are. The systems I helped work on were always "fly through" autopilots where the pilot can gracefully enter the control loop at any time.
Thanks Nick, I was just wondering if the use of these 'attitude clamps' was widespread in autopilots but it seems it is a Westlands special
