Attitude indicator – How does it work? | General Aviation

The Aviation Guru » General Aviation » Attitude indicator – How does it work?

Attitude indicator – How does it work?

Question:

The position it should try to erect to would be to indicate zero bank in this constant turn. Does it do this? Anyone tried sitting in a 30 degree turn for ten minutes to find out?

But the correction is in an absolute sense, relative to the earth (an inertial frame). So if the plane’s say, going north and turning to the right, the errection mechanism thinks that gravity is over to the west. But after 180 of turn,it will try to correct the other way, to the east — so it all tends ot balance out if you go in a circle. The true test is to do the 30 degree bank in a very high speed aircraft, so that after several minutes the plane hasn’t really turned that much — in which case you’d see the gyro trying to show wings level. Light aircraft just turn to fast to see the effect.

Response:

Roll-erection cut-out, set at 7 degrees roll in most AIs. RTFM.

Trouble is, most used aircraft tend to be sold without the FMs for instruments like the AI. So I’d guess few pilots actually have access to such a document. Will King etc. send them out on request? Julian Scarfe

Response:

True. Other errors: 1) application of `G" loading during a turn will cause the turn indicator to record an error. ie. If you pull up HARD when doing a rate 1 turn, the indicator will show a higher rate of turn.. and vice versa

Is this really an error? If you are in a turn and pull up hard, two things will happen: 1) you will climb, and 2) your turn will tighten and, therefore, the turn rate will increase. I believe the rate of turn shown by the instrument will still be accurate. 2) The VSI is a great TREND instrument.. very useful for catching slight climbs / descents b4 the altimeter shows anything. However, it lags pretty badly. ie when you level off, it will initially show a climb even though the altimeter has stopped.

It will show a descent if, prior to leveling off, you were descending. 3) DG also has a turning error associated with it. For a constant rate turn, it tends to speed up passing through E/W and slow down passing N/S (or the other way round.. but there is an error)

I believe you are mistaken here. I know of no such error. This sounds like compass error, sorta. See below. 4) Compass also has turning and acceleration errors (Could someone plse elaborate to them? I can’t recall what the exact errors are

In the northern hemisphere, turns in the northern half of the compass will cause the indicated heading to lag the turn. In the southern half of the compass, the compass indication will lead the turn. On headings of east or west, the compass will indicate correctly. Accelerations to the east (or deccelerations when facing west) will cause the compass to indicate a turn to the left. Opposite accelerating to the west. No acceleration effects north or south. Bill Levenson PP-ASEL-IA

Response:

indicate zero bank in this constant turn. Does it do this? Yes, but it doesn’t do this as rapidly as it does on startup. Apparently the erection mechanism is only fully functional when the gyros aren’t spinning rapidly.

No, No. The gyro should *never* erect to indicate zero bank when in a turn. In a 360 degree turn, the first part of the turn will cause the gyro to erect in one direction and the last part of the turn will cancel this out and cause it to erect in the opposite direction (remember, we’re talking inertially here, that is, the gyro is tilting relative to it’s proper orientation relative to the horizon). There is an error that is greatest for a 180 degree turn. And the erection mechanism *is* fully functional when the gyro’s are spun up. Anyone tried sitting in a 30 degree turn for ten minutes to find out? A friend of mine did. He reported that the AI had a distinct tilt after he resumed level flight. The instrument in question was about 15 years old.

In a sustained turn, the error should be greatest when turning through the heading that is 180 degrees from where you started the turn. Ideally, this is true no matter how many times you go around and the amplitude of the error is the same no matter how many times or how long you sustain the turn. The error will also be zero at the heading that you started the turn from no matter how many times you go around. Bill Levenson PP-ASEL-IA

Response:

I don’t think I can ever remember climbing into a "rent-a-wreck" and not seeing the HI tilted at a crazy angle – albeit still with brown generally down and blue generally up. If the center of gravity is below the mounting gimbals, given the quality of the bearings required to prevent significant precession errors due to friction, why doesn’t the gyro right itself after it has spun down? Aw rats! You’ve got me. OK, we need ten volunteers to rip apart their AIs please… The c of g isn’t necessarily below the mounting gimbals, in fact I’d guess it’s at a very similar level. The auto-erecting depends only on small free swinging plates which allow air to blow out from a hole which they uncover when they swing to point down. There are four of these, one for each cardinal direction if you like, and each can only swing one side of straight down. If the AI is not erect according to gravity, one or at most two of these little plates will swing, exposing a hole through which air blows, applying a couple to the AI which helps erect it to the position where all four plates cover their respective holes. The original question still remains though. Dave and I had a chat about it last Saturday whilst failing to go flying due to bad weather. In a prolonged constant turn, with apparent gravity through the floor of the plane (we’re talking about a coordinated turn here), one of the aforementioned plates should swing open, thus allowing the air to blow and created the couple which starts to re-erect the AI. The position it should try to erect to would be to indicate zero bank in this constant turn. Does it do this? Anyone tried sitting in a 30 degree turn for ten minutes to find out? — Harlequin Ltd. 44 1223 872519 (fax) Barrington Hall Barrington Cambridge CB2 5RG England

Response:

- Hide quoted text — Show quoted text – DM Any instrument technical gurus out there? I’m not an instrument guru, but I may be able to field some of the many good questions that Dave asks. Can I change the order a little? DM I read somewhere that it is impossible to distiguish DM between the force of accelleration and the force of gravity without any DM external reference. That is true enough — you can’t tell the difference between a uniform gravitational field and an acceleration. The classic example is that it’s impossible to distinguish the following situations: a) being in a lift (elevator) car in interstellar space with no gravity b) being in a lift car in a lift shaft on Earth just after the cable has been cut.

One small nit-pick….. in a real situation in b) you could tell that you were in the vicinity of the earth because the force at the bottom of the car would be slightly higher than at the top, so a free falling collection of bits would gradually be separated in the vertical direction. This would not occur in free space (if this were even possible !) The situation in b) is a tidal gravity force. Unfortunately you cannot get a "uniform" gravitational field ! Only a "small" point… but just to keep the record straight ! Mike

Response:

indicate zero bank in this constant turn. Does it do this?

Yes, but it doesn’t do this as rapidly as it does on startup. Apparently the erection mechanism is only fully functional when the gyros aren’t spinning rapidly. Anyone tried sitting in a 30 degree turn for ten minutes to find out?

A friend of mine did. He reported that the AI had a distinct tilt after he resumed level flight. The instrument in question was about 15 years old. | | Convenience stores are named that because they’re George Patterson – | open all night. It saves the robbers the trouble | of breaking in. |

Response:

Well, some really good replies so far, but still no "absolute" on whether a sustained turn will cause the AI to start falsely "erect".

I will give you my humble absolute: An AI will "falsely erect" in a sustained coordinated turn of any degree. (It will take longer in a steeper turn.) Does it matter? Not really- my main wings level instrument is the TC/inclinometer anyway, not the AI. However, my CFII, knowing this (since I told him last weekend that I never use the AI except for a few moments during transitions) decided that having my TC was a bad idea and he failed it for the whole flight. Bummer. At least it has a little red flag indicating OTS! I think that’s a great idea. Spend a whole day flying w/o the TC, then a whole day flying w/o the AI. You’ll learn the strengths and weakness of each really quickly, and when you have them both you’ll be that much better. I just wish my AI had an inclinometer on it now. (BTW, my "whole day" is about three hours under the hood.) Cheers, Doug — _ Doug / | Fields http://www.311wc.com / N311WC

Response:

: Any instrument technical gurus out there? : I would like to know how a modern AI compensates for the actual : vs the gyroscopic horizen. Any limitation in the instrument would be : very usefull indeed. Does it use any form of pendulum to adjust its gyros : to the aircraft attitude? Pendulous vanes. You need it even in an nose dragger to initially errect a non-caging gyro. : Is there some sort of slow-acting pendulum in a modern AI? If so, : is it possible to "fool" the instrument by a sustained co-ordinated turn? Yes it is. Also, they are subject to accelleration effects as well. -Ron

Response:

[a valiant attempt at describing gyro operations without pictures deleted... nice job] The gyro is calibrated so that the top of the rotor housing is leaned (slanted) forward about 2 1/2 degrees. It is also bottom heavy by a couple of ounces. This aids in the erection as well as the vanes. If my memory serves me right, the slant is to allow for some acceleration/precession errors.

Hmmm. I’ve never heard of this before. Is it possible you are mixing up the directional gyro with the artificial horizon? I have found reference to balance weights and "tilt" to compensate the directional gyro for earth rotation. The gyro in the DG is, of course, fixed in space. Once you set the DG to point to North, it stays pointed that way. But the earth rotates under it such that at some time later, the DG (let’s assume it’s perfect will no longer be pointed North. This effect is dependent on the latitude at which you are located. At the equator, this effect is zero if you are pointing to true north because true north remains in the same inertial direction relative to the equator as the earth rotates. The effect is greatest very near the pole. If you point the DG to true north when you are very near the pole, the error will build at roughly 3.75 degrees every 15 minutes. To compensate for earth’s rotation, a "balance weight" is added to the DG and the DG is tilted a few degrees. With the tilt, the balance weight applies a restoring moment which acts in the plane of the tilt. Since gyroscopes respond 90 degrees out of phase from the applied torque, the result is a precession of the gyro. The text I have that describes this states that the gyro’s in the U.S. are designed to be accurate between 25 and 48 degrees latitude. This seems like a small effect to me but that’s the only thing I could come up with that would explain a "tilt" and a "balance weight" as you describe It seems to me that if such a mechanism were employed on an AI, it would cause a nasty wobbling effect due to precession. One calibration test was to turn the indicator (while stabilezed) 180 degrees from front to back as if an airplane could make a flat turn instantaneously. Because of the slant, the indicator will be off by 5 degrees. The gyro was required to erect itself withing certain amount of time.

Or perhaps the tilt is some mechanism for compesating for the errors induced by a 180 coordinated turn? Just guessing. I still don’t see how a balance weight would work. Also tested was the erection speed by tilting the gyros 30 degrees and timing it to 0 degrees. This was down for roll left, roll right, pitch up, and pitch down. In all cases, it would have to erect itself to 0 (visually) withing a certain amount of time.

Can you tell us what the time constant for this was (i.e. how long did it take for the error to reduce by a factor of 2)? Well I wish I could draw some pictures, but that’s all I have.

Nice job! I’m also not 100% sure my memory is serving me right, so take everthing with a grain of salt. It was alot of fun fixing gyros, it was almost an art form (the fact that they were so delicate). It required alot of feel and hearing to repair them correctly and took a while to master. What a limited skill to have.

Sound great! -ds

Bill Levenson PP-ASEL-IA

Response:

A friend insists that an attitude indicator is what a drill sergeant points at his platoon so that he knows who to march up to and shout "You’ve got a bad attitude, mister!" Probably an old joke, but I liked it!

Response:

Roll-erection cut-out, set at 7 degrees roll in most AIs. RTFM.

Response:

- Hide quoted text — Show quoted text – Well, some really good replies so far, but still no "absolute" on whether a sustained turn will cause the AI to start falsely "erect". My guess based on the thread so far is that it will. A pendulum will be fooled by centripetal force (it will behave the same as the ball in the turn & slip), and if this is the *only* "earth" reference that the AI has …. It’s not quite the same. Don’t forget that as the aircraft turns, the centrifugal force also rotates in the inertial frame, and it’s the inertial frame that the gyro knows about. So unless the correction is *very* rapid, the effect of that force will average out to zero. I was fully intending to test this out this weekend by holding a 30deg banked turn for, say, 10 minutes, but the weather was not on my side during the times I had to do it. If there is anyone who can afford to burn 10 minutes or so of flying to check this out, I think that it would be a worthwhile exercise. I’ll wager you your ten minutes fuel that there will be no error due to the centrifugal force! If you’re still dubious, try imagining what the gyro’s doing. Remember that in the turn, the aircraft (and AI mount) banks while the gyro stays vertical. So, say your error were introduced, and the gyro axis tilts 5 degrees towards the North. On the other side of the turn, the correction mechanism would tilt it 5 degrees South… It *must* cancel out over the turn. So the error can’t build up, it can only be (almost) instantaneous. And if you’re going to make the AI correct itself to the apparent vertical almost instantaneously, there’s no point in spending all that money on a gyro, just buy a plumb line (or look at the slip ball!). Julian Scarfe

Hello again Julian! :-) In my IFR training, the precession error we expect due to turns was for a 180 deg coordinated turn. When rolling out level on the reciprocal heading you expect were to expect an error of a few degrees in pitch and bank (I forget which way). Hmmm, didn’t I say this before? :-) However, in a 360 deg turn, you didn’t expect any error, because, just as you said, the errors of the second half of the turn cancel out the errors of the first half. Bruce Bateman

Response:

a sustained shallow turn will cause the AH / AI to read erroneously (acquire a false vertical). I don’t know the physics behind it, but this is a basic instrument error that was taugh to me in flight school. Other errors: 1) application of `G" loading during a turn will cause the turn indicator to record an error. ie. If you pull up HARD when doing a rate 1 turn, the indicator will show a higher rate of turn.. and vice versa 2) The VSI is a great TREND instrument.. very useful for catching slight climbs / descents b4 the altimeter shows anything. However, it lags pretty badly. ie when you level off, it will initially show a climb even though the altimeter has stopped. 3) DG also has a turning error associated with it. For a constant rate turn, it tends to speed up passing through E/W and slow down passing N/S (or the other way round.. but there is an error) 4) Compass also has turning and acceleration errors (Could someone plse elaborate to them? I can’t recall what the exact errors are I have actually witnessed the AH/AI error, as well as 1-3 in flight. Was deliberate attempt to demonstrate the errors and prove them. Anyone care to add to the list? Song

Response:

Well, some really good replies so far, but still no "absolute" on whether a sustained turn will cause the AI to start falsely "erect". My guess based on the thread so far is that it will. A pendulum will be fooled by centripetal force (it will behave the same as the ball in the turn & slip), and if this is the *only* "earth" reference that the AI has …. The question wil then be how long before a significant false reading results. An e-mail respodent mentioned a rate of 3 deg per minute. If so, this would be significant. I would figure that in order to take turning forces into account, the AI would need to be very "clever", because ISTM that the rate of turn would need to be referenced to *speed* to figure bank angle (and an AI has no speed input). I was fully intending to test this out this weekend by holding a 30deg banked turn for, say, 10 minutes, but the weather was not on my side during the times I had to do it. If there is anyone who can afford to burn 10 minutes or so of flying to check this out, I think that it would be a worthwhile exercise. The consequences of such an instrument error are quite big. In a sustained turn in IMC, such instrument error would tend to make the pilot increase bank. If the turn indicator is not taken into account during the pilot’s instrument scan, a spiral dive could result – not a trivial error then! This is the main reason that I don’t believe that such an error could exist – surely it would be a *major* caution taught in an instrument course if it were. I can appreciate that normally sustained turns would not be done IFR, and a racecourse hold would maybe give enough time during the straight legs to get the AI correctly earth-referenced. I have never felt the need *yet* to do a sustained turn in IMC, but I can imagine real situations where I might want to. E.G., holding a position whilst taking cross-bearings on several VOR’s or ADF’s etc, or making an ad-hoc hold along a route whilst IMC, but not under IFR rules (as is a normal condition with a UK IMC rating). I will try again next weekend as I can’t fly during the week Dave Mould Ordinary PPL

Response:

DM Any instrument technical gurus out there? I’m not an instrument guru, but I may be able to field some of the many good questions that Dave asks. Can I change the order a little? DM I read somewhere that it is impossible to distiguish DM between the force of accelleration and the force of gravity without any DM external reference. That is true enough — you can’t tell the difference between a uniform gravitational field and an acceleration. The classic example is that it’s impossible to distinguish the following situations: a) being in a lift (elevator) car in interstellar space with no gravity b) being in a lift car in a lift shaft on Earth just after the cable has been cut. Both a and b look like freely falling inertial frames. If you release something it obeys Newton’s first law (uniform velocity or no motion at all). [Another way of looking at it is that gravity *is* a manifestation of the failure to use an appropriate freely falling frame!] However, you *can* tell the difference between a rotating frame and a non-rotating frame (pace Ernst Mach, who explored some very interesting alternatives). Give something a velocity in a rotating frame and watch it curve (coriolis force)! That’s in essence *why* a gyro works. It remembers which direction was which in a rotating frame. But there’s no perfect gyro, so gyros need to be referenced to something on a regular basis to avoid errors due to precession. With the DG that’s easy: you set it against the compass from time to time, but the AI is different. What you want to know is which way is "down", "down" being the direction in which gravity is acting, and *any* AI worth having must have a mechanism for sensing "down". DM If I fly several hours on a N/S heading, the earth has rotated many DM degrees. I would expect a free-floating gyroscope to precess DM and indicate an increasing amount of bank. But the modern AI does not. DM How is this achieved? Even an old gyro does not, because it’s not (as Dave has deduced) free gyro. The name (almost confined to books on aircraft!) given is "Earth gyro". DM I would like to know how a modern AI compensates for the actual vs the DM gyroscopic horizen. Any limitation in the instrument would be very DM usefull indeed. Does it use any form of pendulum to adjust its gyros DM to the aircraft attitude? …So how does a gyro detect down so that it can reorient itself? Well how would *you* do it? Drop a plumb line, right? Older vacuum-driven gyros worked on a very similar principle, using hanging "pendulous vanes" to open up or cut off a a current of air that applied an appropriate correcting force on the gyro. A don’t know if modern vacuum-driven gyros still use exactly that principle. An electric gyro can use mercury switches to the same effect. But like any such self correction mechanism (just like an autopilot), the designer has a choice about how fast the device tries to make the correction. Too slow and the instrument may become inaccurate before the correction can compensate. Too fast and any temporary or transient influences on the sensing mechanism could cause it to correct itself constantly. This has some interesting consequences. DM For instance: If I start up in a DM taildragger, I am obviously in a very nose-up attitude. Modern AIs DM have no manual erection knob, but nevertheless seem to indicate the DM correct attitude. The old AI’s need to be manually erected after DM establishing straight & level. How do modern AI’s do this DM automagically? When you say "old AI’s need to be manually erected after establishing straight & level", that’s true, but it’s not a *calibration* that you’re performing, it’s getting the thing close so that the automagic mechanism has an easier job. A bit like "helping" the autopilot by putting the aircraft in a close to level-flight attitude before pressing the altitude hold. It’s not essential, but it helps to sort things out quicker. I believe many modern gyros have a "fast erect" button for this. That substantially increases the rate at which the automagic seeks the apparent vertical. So you can press it when you’re stationary on the ground and you know that there are no inertial forces messing up what it thinks is vertical. DM Is there some sort of slow-acting pendulum in a modern AI? I think I covered that… DM If so, is it DM possible to "fool" the instrument by a sustained co-ordinated turn? My DM suspicion is that it might begin to "erect" itself to align with the DM new apparent "gravitational" direction. You might have thought so, but remember that the gyro *does* know what’s a rotating frame and what’s not. So in a sustained co-ordinated turn, the gyro "knows" that the apparent vertical is revolving in space, and the average correction that it makes is towards the average Earth vertical. If you could sustain a *linear* acceleration for long enough, you could, in theory, fool it. However, that depends on the timescale over which the correction is applied being much longer than the period of the turn. If you were to use a fast erect mechanism in coordinated turn, you might get it confused enough for long enough to think that the vertical had moved, which is why that’s only for unaccelerated conditions. DM If I do aerobatics, including loops and rolls, my modern AI still DM appears to read correctly afterwards. Are there any limitations? Is DM it possible to "topple" a modern AI? Well, the need for gimbals on which to mount the gyro restricts the orientations somewhat. My hat off to the person who designed the gyro that continues to read through the vertical into inverted flight! As I understand it the sort of AI used in most modern fighters is vastly more expensive for just that reason. (Maybe it’s all done with laser gyros now?). DM If left uncaged, an old-fashioned DM AI also fitted in my aircraft is performing rapid gyrations after the DM same manoevours untill I manually erect it. Is there then an essential DM design difference between the two? I doubt it. Perhaps the stops are in a different orientation? DM If I accidentally entered a spin in DM IMC, but recovered OK (whew!) can I still use my AI afterwards, or DM should I fly on limited panel in case the AI is fouled up? (Particularly) good question. That depends very much on the construction of the AI. There’s no reason in principle why the gyro shouldn’t cope, but the practicalities of engineering the thing might intervene to cause problems. DM How about an extended "hold"? No problem. As described above, as long as the erection mechanism is slow enough it should average out the centrifugal force. Julian Scarfe

Response:

Aw rats! You’ve got me. OK, we need ten volunteers to rip apart their AIs please… Julian Scarfe

Response:

: Yes, there’s a sort of pendulum in there. I’ve not seen one disassembled, : but there’s a description in the Jepessen manual. So far I have read a bunch of good information on this subject in this thread, especially Julian’s article. I used to repair AI (or artificial horizons – AH), from Edo’s to Kings. Primarily for general aviation airplanes. They all worked essestially the same, even electric ones. And to the best of my knowledge, they’ve haven’t change much in operation for decades. But what has advanced is the electronics (autopilot, flight directors, etc.) and the bearings. If anything fails in an AH, it will be the bearings. But I’ll try to describe what I remember. The gyro is a spinning mass (rotor) about 2 inches in diameter and about 2 inches tall and spins about a vertical axis (10-20,000 RPM?). It has groves along the perimeter that air (suction) is forced over to spin the rotor. It is contained in a housing that has four vents in the bottom: fore, aft, left, and right. The housing contains bearings so that it pivots fore and aft about a horizontal axis that runs from left to right. Housing pivots translate into pitch indication. Air is forced into the housing through one of it’s pivot points and exits the vents. There are two pendulums (vanes) that cover the vents, one for the fore and aft, the other for the left and right. As one vent is covered, the opposite is opened. The vanes are set up so when the gyro is erected, there is an equal amount of opening on both sides. There is a u-shape part (name?) the supports the rotor housing. The rotor housing is limited by spring stops from spinning all the way around. This u-shaped part is hollow on one side for air to flow into the rotor housing. The u-shape has it’s own axis that it spins that is horizontal and fore and aft via a hollow shaft in the back part of the U. This shaft is also hallow for air to flow. The U-shaped part is allowed to rotate freely 360 degrees within the housing of the instrument casing and translate to roll in the indicator. It is through this shaft that air originates as it is open to the outside of the indicator (in the back). The instruments’s chamber represents where air is let out through the vents and then eventually through another hole in the back of the unit. On most (if not all) AI’s, air is sucked (vacuum) from here and not forced through the shaft. The gyro is calibrated so that the top of the rotor housing is leaned (slanted) forward about 2 1/2 degrees. It is also bottom heavy by a couple of ounces. This aids in the erection as well as the vanes. If my memory serves me right, the slant is to allow for some acceleration/precession errors. One calibration test was to turn the indicator (while stabilezed) 180 degrees from front to back as if an airplane could make a flat turn instantaneously. Because of the slant, the indicator will be off by 5 degrees. The gyro was required to erect itself withing certain amount of time. Also tested was the erection speed by tilting the gyros 30 degrees and timing it to 0 degrees. This was down for roll left, roll right, pitch up, and pitch down. In all cases, it would have to erect itself to 0 (visually) withing a certain amount of time. Well I wish I could draw some pictures, but that’s all I have. I’m also not 100% sure my memory is serving me right, so take everthing with a grain of salt. It was alot of fun fixing gyros, it was almost an art form (the fact that they were so delicate). It required alot of feel and hearing to repair them correctly and took a while to master. What a limited skill to have. -ds

Response:

BB I was taught in my instrument training that when rolling out of a BB 180deg standard rate co-ordinated turn, that I should expect to see BB an error in the HI of several degrees in both pitch and bank. I BB though this was supposed to be due this exact effect – the gyro trying BB to correct itself to an "incorrect" down. An excellent point. In the light of Bruce’s comment, the notes for the PPL/IR that I just consulted, and a comment from Ken Sykes, the more I think about it, the more I think that I have underplayed the effects of acceleration error in my original article. To look at acceleration error in more detail, take the classic textbook example of an aircraft accelerating along the runway: its antiquated AI shows a nose-up right-wing down attitude, from two independent sources: 1) The erection mechanism feels the inertial force of the acceleration (acting towards the rear of the aircraft), which it can’t differentiate from gravity and tries to "right" the AI, so it tips to indicate nose-up. 2) The erection chamber is a big lump of iron on the bottom of the gyro. That means that the CG of the gyro is below the pivot on the gimbal. As a result, the inertial force of the acceleration causes a couple on the gyro, and, since gyros display this perverse habit of *precession*, the couple results in a right-wing down indication for an AI rotating couterclockwise when viewed from the top (which apparently is the de facto standard for vacuum AIs). Have you ever seen either of these happen on a modern AI? Have you seen the error that Bruce describes when rolling out of a standard rate turn? I’m not sure that I have, and I will make some speculative deductions about the development of AIs from that. Please bear in mind that I’m not in the aircraft instruments business and I’d welcome any corrections. a) From the absence of (1) I infer that the normal erection mechanism in a modern gyro is probably designed to work more slowly than in an old brick-sized AH. After all, if gyro mechanisms are getting better, the precession because of mechanical instrument imperfections is likely to require less rapid correction. No technology can eliminate part (1) of acceleration error (unless it cheats by time-averaging the acceleration, and even then there will be some error) because of the equivalence principle (elevators in space ‘n all that) I described before. b) If you think about (2) for a minute, you might see a logical flaw. Why on Earth do we *allow* the CG of the gyro to be below the pivot? Why not put a balance weight on the top so that CG is exactly in line with the pivot? Ken Sykes described to me by email a cunning way of eliminating this effect altogether using a pair of counterrotating gyros. The only deduction I can make is that the designers *wanted* it this way so that the gyro is self-righting when it’s not spinning. After all, when you’ve started the engine, before the gyro has got up to speed, wouldn’t it be better if a straightforward pendulum effect put it in almost the correct attitude to start with? Evidently modern gyros have a better mechanism for doing this (the fast-erect mechanism?), even on vacuum driven gyros and can be perfectly balanced so that an acceleration fore or aft need not cause a bank indication on the AI. So I’d like to correct what I said to in reply to Dave: DM If left uncaged, an old-fashioned DM AI also fitted in my aircraft is performing rapid gyrations after the DM same manoevours untill I manually erect it. Is there then an essential DM design difference between the two? JS Not really. Perhaps the stops are in a different orientation? I think there *is* good reason to suspect that the older gyro will be much less reliable after accelerations than modern ones, for the reasons I describe above. There may well be an essential design difference between the two. I hope that clarifies things a little. Julian Scarfe

Response:

I would like to know how a modern AI compensates for the actual vs the gyroscopic horizen. Any limitation in the instrument would be very usefull indeed. Does it use any form of pendulum to adjust its gyros to the aircraft attitude?

Yes, there’s a sort of pendulum in there. I’ve not seen one disassembled, but there’s a description in the Jepessen manual. The old AI’s need to be manually erected after establishing straight & level. How do modern AI’s do this automagically?

They have a quick erection air bleed system. My understanding is that this only operates for a short time after the unit starts up. Some units have a "non-locking caging knob" which can be used to engage this system at any time. These are popular for use in aerobatic aircraft. Is there some sort of slow-acting pendulum in a modern AI? If so, is it possible to "fool" the instrument by a sustained co-ordinated turn? My suspicion is that it might begin to "erect" itself to align with the new apparent "gravitational" direction.

You are correct. If you stay in a coordinated turn long enough, the thing will gradually erect itself. ….impossible to distinguish between the force of accelleration and the force of gravity without any external reference. But the AI is (apparently) able to achieve this??

Not really. The gyroscope is simply resistent to any change in attitude, so it stays nice and level while the aircraft moves around it. If I do aerobatics, including loops and rolls, my modern AI still appears to read correctly afterwards. Are there any limitations? Is it possible to "topple" a modern AI?

According to Sigma-Tek, their basic models are accurate through 360 degrees. If I accidentally entered a spin in IMC, but recovered OK (whew!) can I still use my AI afterwards, or should I fly on limited panel in case the AI is fouled up?

I would simply double check the AI against the turn coordinator or T&B and use it if they agree. I don’t have much instrument training, and I’ve no spin training, however, so that opinion isn’t worth much. How about an extended "hold"?

The holding procedure which I was taught involves standard-rate procedure turns separated by two minute legs of straight & level flight. Since it requires at least five 360 degree turns before the unit starts to tilt, holding turns are no problem. | The Swordfish relies on her "peggy" George Patterson – | The modified Taurus ain’t sound | So the Swordfish flies off on her missions | And the Albacore stays on the ground.

Response:

Ahhh Julian, you continue to impress me with your astute analysis. But….. :-) – Hide quoted text — Show quoted text – ….<snip… To look at acceleration error in more detail, take the classic textbook example of an aircraft accelerating along the runway: its antiquated AI shows a nose-up right-wing down attitude, from two independent sources: 1) The erection mechanism feels the inertial force of the acceleration (acting towards the rear of the aircraft), which it can’t differentiate from gravity and tries to "right" the AI, so it tips to indicate nose-up. 2) The erection chamber is a big lump of iron on the bottom of the gyro. That means that the CG of the gyro is below the pivot on the gimbal. As a result, the inertial force of the acceleration causes a couple on the gyro, and, since gyros display this perverse habit of *precession*, the couple results in a right-wing down indication for an AI rotating couterclockwise when viewed from the top (which apparently is the de facto standard for vacuum AIs). Have you ever seen either of these happen on a modern AI? Have you seen the error that Bruce describes when rolling out of a standard rate

I’ve never noticed, but then, I’m still at a point where if I end up within a couple of degress of ANYTHING when I finish a maneuver (take-off roll, 180deg turn, etc) I’m pretty happy. :-) a) From the absence of (1) I infer that the normal erection mechanism in a modern gyro is probably designed to work more slowly than in an old brick-sized AH. After all, if gyro mechanisms are getting better, the precession because of mechanical instrument imperfections is likely to require less rapid correction. No technology can eliminate part (1) of acceleration error (unless it cheats by time-averaging the acceleration, and even then there will be some error) because of the equivalence principle (elevators in space ‘n all that) I described before.

The only comment I’d add here is that in all the descriptions I’ve ever read about vacuum gyros, they’ve all indicated the use of "pendulous vanes" as an erecting mechanism. I suspect that improvements have come in the form of lighter materials, more accurate tolerances, better bearings, etc. – Hide quoted text — Show quoted text -b) If you think about (2) for a minute, you might see a logical flaw. Why on Earth do we *allow* the CG of the gyro to be below the pivot? Why not put a balance weight on the top so that CG is exactly in line with the pivot? Ken Sykes described to me by email a cunning way of eliminating this effect altogether using a pair of counterrotating gyros. The only deduction I can make is that the designers *wanted* it this way so that the gyro is self-righting when it’s not spinning. After all, when you’ve started the engine, before the gyro has got up to speed, wouldn’t it be better if a straightforward pendulum effect put it in almost the correct attitude to start with? Evidently modern gyros have a better mechanism for doing this (the fast-erect mechanism?), even on vacuum driven gyros and can be perfectly balanced so that an acceleration fore or aft need not cause a bank indication on the AI.

Back in your court! :-) Bruce Bateman

Response:

Julian Scarfe covered most everything very well, but may I add that 2 counter-rotating gyros on the gimbaled platform will cancel any acceler- ation precessions. The Earths rotational precession is 15deg/hr times the sine of the latitude. Therefore, the vertical sensor corrector need work at no more than 15deg/hr. Actually that’s a pretty slow rate, 1/2 the rate of a clock’s hourhand. Tallyho ! Alpha Kilo

Response:

Excellent (as usual) analysis, Julian. I have only one comment….. – Hide quoted text — Show quoted text – DM Any instrument technical gurus out there? I’m not an instrument guru, but I may be able to field some of the many good questions that Dave asks. Can I change the order a little? ….<snip……. DM If so, is it DM possible to "fool" the instrument by a sustained co-ordinated turn? My DM suspicion is that it might begin to "erect" itself to align with the DM new apparent "gravitational" direction. You might have thought so, but remember that the gyro *does* know what’s a rotating frame and what’s not. So in a sustained co-ordinated turn, the gyro "knows" that the apparent vertical is revolving in space, and the average correction that it makes is towards the average Earth vertical. If you could sustain a *linear* acceleration for long enough, you could, in theory, fool it. However, that depends on the timescale over which the correction is applied being much longer than the period of the turn. If you were to use a fast erect mechanism in coordinated turn, you might get it confused enough for long enough to think that the vertical had moved, which is why that’s only for unaccelerated conditions. ….<snip……

I was taught in my instrument training that when rolling out of a 180deg standard rate co-ordinated turn, that I should expect to see an error in the HI of several degrees in both pitch and bank. I though this was supposed to be due this exact effect – the gyro trying to correct itself to an "incorrect" down. ???? Bruce Bateman

Response:

Any instrument technical gurus out there? I would like to know how a modern AI compensates for the actual vs the gyroscopic horizen. Any limitation in the instrument would be very usefull indeed. Does it use any form of pendulum to adjust its gyros to the aircraft attitude? For instance: If I start up in a taildragger, I am obviously in a very nose-up attitude. Modern AIs have no manual erection knob, but nevertheless seem to indicate the correct attitude. The old AI’s need to be manually erected after establishing straight & level. How do modern AI’s do this automagically? If I fly several hours on a N/S heading, the earth has rotated many degrees. I would expect a free-floating gyroscope to precess and indicate an increasing amount of bank. But the modern AI does not. How is this achieved? Is there some sort of slow-acting pendulum in a modern AI? If so, is it possible to "fool" the instrument by a sustained co-ordinated turn? My suspicion is that it might begin to "erect" itself to align with the new apparent "gravitational" direction. I have not conducted any extensive tests, ‘cos I’m hoping that someone already knows all the answers I read somewhere that it is impossible to distiguish between the force of accelleration and the force of gravity without any external reference. But the AI is (apparently) able to achieve this?? If I do aerobatics, including loops and rolls, my modern AI still appears to read correctly afterwards. Are there any limitations? Is it possible to "topple" a modern AI? If left uncaged, an old-fashioned AI also fitted in my aircraft is performing rapid gyrations after the same manoevours untill I manually erect it. Is there then an essential design difference between the two? If I accidentally entered a spin in IMC, but recovered OK (whew!) can I still use my AI afterwards, or should I fly on limited panel in case the AI is fouled up? How about an extended "hold"? I have questioned CFIs and qualified mechanics on the above, but have received contradictory advice, and nothing that sounds authorotative. Dave Mould Ordinary PPL

Response:

I think it has to do with the limits of motion of the gyro cage. The older gyros had stops that limited the travel of the gimbels. If you hit the stops, it forced the gyro to a new position in space. This caused the gyro to tumble and then it had to be erected or caged and restarted in the new position. The new gyros do not have these stops. The gimbals are completely free and the gyro is "non tumbling" and does not require the caging and manual erection to straighted it out after it tumbles. Many of the new non tumbling gyros can get pretty confused after some aerobatic maneuvers though. It can take them awhile to damp out and stop their oscillation after you roll them over. John – Hide quoted text — Show quoted text – Any instrument technical gurus out there? I would like to know how a modern AI compensates for the actual vs the gyroscopic horizen. Any limitation in the instrument would be very usefull indeed. Does it use any form of pendulum to adjust its gyros to the aircraft attitude? For instance: If I start up in a taildragger, I am obviously in a very nose-up attitude. Modern AIs have no manual erection knob, but nevertheless seem to indicate the correct attitude. The old AI’s need to be manually erected after establishing straight & level. How do modern AI’s do this automagically? If I fly several hours on a N/S heading, the earth has rotated many degrees. I would expect a free-floating gyroscope to precess and indicate an increasing amount of bank. But the modern AI does not. How is this achieved? Is there some sort of slow-acting pendulum in a modern AI? If so, is it possible to "fool" the instrument by a sustained co-ordinated turn? My suspicion is that it might begin to "erect" itself to align with the new apparent "gravitational" direction. I have not conducted any extensive tests, ‘cos I’m hoping that someone already knows all the answers I read somewhere that it is impossible to distiguish between the force of accelleration and the force of gravity without any external reference. But the AI is (apparently) able to achieve this?? If I do aerobatics, including loops and rolls, my modern AI still appears to read correctly afterwards. Are there any limitations? Is it possible to "topple" a modern AI? If left uncaged, an old-fashioned AI also fitted in my aircraft is performing rapid gyrations after the same manoevours untill I manually erect it. Is there then an essential design difference between the two? If I accidentally entered a spin in IMC, but recovered OK (whew!) can I still use my AI afterwards, or should I fly on limited panel in case the AI is fouled up? How about an extended "hold"? I have questioned CFIs and qualified mechanics on the above, but have received contradictory advice, and nothing that sounds authorotative. Dave Mould Ordinary PPL