Go Back  PPRuNe Forums > Flight Deck Forums > Tech Log
Reload this Page >

B-737 Speed Trim System

Tech Log The very best in practical technical discussion on the web

B-737 Speed Trim System

Old 28th Nov 2018, 04:45
  #61 (permalink)  
 
Join Date: Feb 2009
Location: Seattle
Posts: 379
Likes: 0
Received 0 Likes on 0 Posts
Mana - When Boeing developed the C*U control law that is the augmented pitch axis of the 777 a very deliberate design decision was made to provide positive speed stability that would require pilot action to "trim" column forces that build up when airspeed changes. C*U speed stability is implemented by managing within the control law a reference speed and requiring steady column force to fly at speeds away from that reference. Pitch trim on the 777 works to slew the reference speed up and down. The 777 does not drive the stabilizer directly to create speed stability, but it does augment the response characteristics in such a way that column force builds up that must be trimmed off. Without trim, the 777 seeks to return to the reference speed. 777 pilots must trim against the speed stability provided by C*U whenever they change speed regardless of the unaugmented speed stability characteristics of the bare airplane itself. In this way, the pilot task with regard to speed changes is not that different between 777 and 737.
FCeng84 is offline  
Old 28th Nov 2018, 06:15
  #62 (permalink)  
Thread Starter
 
Join Date: Aug 2000
Posts: 1,499
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by FCeng84
Mana - When Boeing developed the C*U control law that is the augmented pitch axis of the 777 a very deliberate design decision was made to provide positive speed stability that would require pilot action to "trim" column forces that build up when airspeed changes. C*U speed stability is implemented by managing within the control law a reference speed and requiring steady column force to fly at speeds away from that reference. Pitch trim on the 777 works to slew the reference speed up and down. The 777 does not drive the stabilizer directly to create speed stability, but it does augment the response characteristics in such a way that column force builds up that must be trimmed off. Without trim, the 777 seeks to return to the reference speed. 777 pilots must trim against the speed stability provided by C*U whenever they change speed regardless of the unaugmented speed stability characteristics of the bare airplane itself. In this way, the pilot task with regard to speed changes is not that different between 777 and 737.
Thanks! So there is a difference between US and European certification in this area? I don’t fly Airbus but I believe Airbus aircraft are kept in trim (pitch).
If I understand the 777 system correctly, you do not need at lot of trim if a short trim activation resets the speed reference?

I know what I would prefer my aircraft to do.
ManaAdaSystem is offline  
Old 28th Nov 2018, 06:29
  #63 (permalink)  
 
Join Date: Mar 2001
Location: I wouldn't know.
Posts: 4,497
Likes: 0
Received 0 Likes on 0 Posts
I don’t think that it is a major difference between European and US certification rules as such, just a difference in control design. After all there are no changing stick forces in an airbus, the only force is a spring setup in the stick itself to return it to neutral, completely independent of control law and speed. Trim is always automatic except in direct law, and most people forget at first to trim manually if they happen to find themselves in direct law, which is the last fallback level available.

That said, the airbus FBW design is older than the Boeing one, and I think that Boeing learned from what happened with the Bus.

By the way, it is absolutely astonishing how fast one de-learns trimming on switching from Boeing to Airbus. Just my personal observation. Simply because it is never needed in normal line operation.
Denti is offline  
Old 28th Nov 2018, 18:39
  #64 (permalink)  
 
Join Date: Sep 2008
Location: Scotland
Posts: 890
Received 6 Likes on 2 Posts
The 777 flight control system is excellent. It flies like a large, more stable 737. Trimming is natural and intuitive and exactly like flying a non-few aircraft. It also keeps you in practice for the two reversionary modes, where manual trim is required. It also means the handling characteristics in reversionary modes are very similar to normal mode and require less capacity in a non-normal situation. C* U is excellent - Boeing certainly as Denti says looked hard and learned lessons from the Airbus approach.
Jwscud is offline  
Old 28th Nov 2018, 20:01
  #65 (permalink)  
 
Join Date: Jun 2009
Location: florida
Age: 81
Posts: 1,609
Received 52 Likes on 15 Posts
Salute Denti!

A great point about
how fast one de-learns trimming on switching
The Boeing control laws are different than the 'bus, but do not seem to have as many reversion sequences. I would have to fly one to see how it 'feels". A lotta difference between trimming for speed/AoA versus a gee command that is biased by pitch attitude and bank angle.

As denti says, the move from a "conventional" plane to the new, improved one is not always a big deal. We old curmudgeons that resisted the Viper limiters abandoned the prejudice real fast. We also had no problem NOT TRIMMING for speed/AoA. It was more trimming for attitude, and the 'bus does that without a "coolie" hat button or whatever. If you are holding stick fore or aft, the the FLCS tries to reduce that pressure/displacement/force. AF447 showed one problem with that philosophy, in that we had a fully deflected stab after "x" seconds since the pilot held back stick forever and the FLCS tried to reduce the back stick displacement.

Gums...

Last edited by gums; 29th Nov 2018 at 00:03. Reason: typos
gums is offline  
Old 29th Nov 2018, 17:44
  #66 (permalink)  
 
Join Date: Aug 2009
Location: Texas
Age: 64
Posts: 7,126
Received 315 Likes on 203 Posts
Intermim Report's Out.
http://knkt.dephub.go.id/knkt/ntsc_a...y%20Report.pdf

As mentioned elsewhere, the MCAS system didn't get a mention.
As also mentioned elsewhere, multiple previous sectors were flown using manual trim wheel, apparently, to handle the defect that never got quite fixed.
The CVR still not recovered.
I encourage your attention to pages 14 and 16 of the report: the difference in the parameters on the accident flight and the previous flight.
Lonewolf_50 is online now  
Old 29th Nov 2018, 20:59
  #67 (permalink)  
 
Join Date: Feb 2017
Location: Isla Grande
Posts: 996
Received 0 Likes on 0 Posts
MCAS is mentioned in the BOEING letter.

Chapter 5.12, p62
gearlever is offline  
Old 29th Nov 2018, 21:32
  #68 (permalink)  
 
Join Date: Jul 2014
Location: Germany
Posts: 344
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by Lonewolf_50
As also mentioned elsewhere, multiple previous sectors were flown using manual trim wheel, apparently, to handle the defect that never got quite fixed.
Could you point to the page where "multiple" previous sectors were flown using the trim wheel?
Looking at FR24 data the typical altitude excursions associated with the MCAS trim and stickshaker are only present on the accident and previous flight DPS to Jakarta.

Reading pages 7, 8 and 9 it appears to me the plane intermittently didn't show any speed or altitude on the captains side.
If i understand correctly this could happen if the left AoA signal was missing as static pressure which is needed for speed and altitude gets corrected by AoA.
So intermittently failing AoA sensor might have been the problem.

Now in Denpasar the AoA sensor was replaced. This is the most likely point for the introduction of the 20 degree offset.
A stuck sensor that just transmits one value might happen as a failure. But tracking a 20 degree offset is very unlikely to occur without some change being made.

Possible reasons for the 20 degree offset i can imagine are:
Wrong part number
Faulty manufacturing
Very creative attachment to the plane
AoA vane bent after installation and test


To quote the report:
For troubleshooting due to repetitive problem perform replaced angle of attack sensor in accordance with Aircraft Maintenance Manual (AMM) Task 34-21-05-000-001 and task 34-21-05-400-801 carried out. Installation test and heater system test result good.
Some people said the AoA sensor needs to be tested with a special jig to orient it while someone in the cockpit is checking the output value.
Now either that test was not performed or they made a mistake here.
wiedehopf is offline  
Old 29th Nov 2018, 23:06
  #69 (permalink)  
 
Join Date: May 2016
Location: Sunshine Coast
Posts: 1,146
Received 181 Likes on 90 Posts
Originally Posted by wiedehopf
Some people said the AoA sensor needs to be tested with a special jig to orient it while someone in the cockpit is checking the output value.
Now either that test was not performed or they made a mistake here.
Or it's not a raw data (sensor) problem rather it's a processed data (ADIRU) problem. The Left ADIRU seems to pop up quite a bit in the AFML Resolution Descriptions. I've got a fiver on a fault in the Left ADIRU featuring as a contributing factor.
MickG0105 is offline  
Old 30th Nov 2018, 05:33
  #70 (permalink)  
 
Join Date: Feb 2009
Location: Seattle
Posts: 379
Likes: 0
Received 0 Likes on 0 Posts
The difference between the C* pitch control approach taken by Airbus and the C*U approach taken by Boeing is not so much a difference in cert authority requirements as it is a difference in how these two groups chose to address speed stability requirements. Boeing with C*U chose to continue to meet the existing requirements for stick force per knot as speed changes. Airbus chose to seek a special condition to allow their design not to require any force for speed changes within the normal envelope. Airbus was able to convince cert authorities that because their system provides protections for both overspeed and underspeed conditions there was no need to provide pitch controller force cues for speed changes that do not exceed upper or lower limits. As a result, the Airbus C* system does not require the pilot input to manage pitch trim where the Boeing C*U system does. The Airbus system presents the pilots with less workload. The Boeing system is more conventional when compared with non FBW airplanes. As noted earlier, the Boeing system promotes the pilot maintaining pitch trim skills that come in very handy if failures result in dropping to a degraded mode. It may be easy for pilots to unlearn their pitch trim skills when transitioning to a system that does not require pitch trim. Hopefully those same crews are not caught out too far when a failure drops them to a mode that requires manual control of the stabilizer to keep the airplane in pitch trim.
FCeng84 is offline  
Old 30th Nov 2018, 10:55
  #71 (permalink)  
Thread Starter
 
Join Date: Aug 2000
Posts: 1,499
Likes: 0
Received 0 Likes on 0 Posts
With that logic all FBW aircraft should operate in degraded mode at all times. That would keep the pilots ready always.

ManaAdaSystem is offline  
Old 30th Nov 2018, 11:07
  #72 (permalink)  
 
Join Date: Jan 2008
Location: uk
Posts: 857
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by wiedehopf
Some people said the AoA sensor needs to be tested with a special jig to orient it while someone in the cockpit is checking the output value.
Now either that test was not performed or they made a mistake here.
On the classic the last step of the install procedure 27-32-11/401 is to perform system test 27-32-00/501 which includes testing the stall warning system end to end by using the calibrator,
turning the vane and checking at what point stick shaker triggers.

I don't have the relevant bits of NG or MAX AMM to see if it is the same, but it seems likely it will be similar. We know from the log that they recorded doing an "installation test", but we don't know exactly what. However, if as with the classic, the test is specified as the last step of the install procedure, do you actually need to note the test number given that you have recorded completing the install procedure? Not sure.

There is further oddity in the traces too - the 20deg offset is actually not consistent. At the beginning of the previous flight, while on the ground, it looks to be about 10, falling and briefly going -ve, then rising to a constant 20 once the airspeed goes up. At the end of that flight, when airspeed drops, the offset appears to increase a little, and that higher offset is the same at start of next flight, until airspeed comes up and then it goes back to 20.

So we may be looking at a sensor fault that is dependent on airspeed and therefore wouldn't have been found on a ground test anyway. On the other hand the vanes are not alive without airspeed and may settle at different points anyway. But that drop at the start of the previous flight may be significant, probably at the end of taxi, the right AOA bounces a bit too, but the left much more. Something came loose? Image to show what I am rambling on about:



Note that if you are thinking sticky vane (I was), the test procedure on the classic contains the following:
slowly rotate sensor vane between stops using light finger pressure. Check that vane rotates without binding or variations in torque
infrequentflyer789 is offline  
Old 30th Nov 2018, 11:30
  #73 (permalink)  
 
Join Date: Jan 2008
Location: uk
Posts: 857
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by MickG0105
Or it's not a raw data (sensor) problem rather it's a processed data (ADIRU) problem. The Left ADIRU seems to pop up quite a bit in the AFML Resolution Descriptions. I've got a fiver on a fault in the Left ADIRU featuring as a contributing factor.
Yeah, but, stick shaker goes off too, and assuming SMYD (which provides stick shaker) hasn't changed from the NG, it gets raw sin/cos signal for AOA which is what the sensor outputs, not the ADIRU.

Also, on the NG at least, the data acquisition point for AOA is downstream of SMYD, so we are seeing, in the traces, what SMYD saw/calculated, not ADIRU output.

Unless the MAX air data architecture is significantly different, but I don't think it will be.
infrequentflyer789 is offline  
Old 30th Nov 2018, 11:41
  #74 (permalink)  
 
Join Date: Feb 2017
Location: Isla Grande
Posts: 996
Received 0 Likes on 0 Posts
Concerning MCAS on the accident flight.

In the tech log no report of SS on previous flight...!?

Anyhow "the DFDR showed the stick shaker activated during the rotation and remained active throughout the flight."

So the first activation of SS was after replacement of LH AoA sensor.
Is it possible to install a RH AoA sensor in the LH position?

Part number confusion...?

Just asking.
gearlever is offline  
Old 30th Nov 2018, 16:04
  #75 (permalink)  
 
Join Date: Sep 2008
Location: Scotland
Posts: 890
Received 6 Likes on 2 Posts
Originally Posted by ManaAdaSystem
With that logic all FBW aircraft should operate in degraded mode at all times. That would keep the pilots ready always.

Not at all, it is simply a design feature that degradation is graceful and intuitive.
Jwscud is offline  
Old 3rd Dec 2018, 15:17
  #76 (permalink)  
 
Join Date: Jan 2008
Location: uk
Posts: 857
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by infrequentflyer789
There is further oddity in the traces too - the 20deg offset is actually not consistent. [...]
Well, since I wrote that Peter Lemme over at satcom.guru has done an entire piece about failure modes in AOA sensors, and he picked up on that bit of the trace too, as well as much more. Worth a read: https://www.satcom.guru/2018/12/angl...ure-modes.html

In summary: he doesn't have a failure mode that fits all the trace data either.
infrequentflyer789 is offline  
Old 4th Dec 2018, 01:04
  #77 (permalink)  
 
Join Date: Jul 2009
Location: Not far from a big Lake
Age: 81
Posts: 1,454
Likes: 0
Received 0 Likes on 0 Posts
Infrequentflyer, I think we can easily explain the WTF point on the graph you presented.
Siince the aircraft is not yet airborne, that is probably brief jet blast impingement on the nose of the aircraft.

Additionally, Mr. Lemme has called attention to the difference in signal between left and right AOA while taxiing. The left AOA was noisy and the right AOA was rather smooth in that segment of the FDR data.
IMO that would be typical of a sneak circuit that is being excited by vibration transmitted by the landing gear into the fuselage.
Sneak circuits commonly originate at bent connector pins, conductors that are chafed by adjacent structure, or wires that become pinched when other components are installed, but there are also many other possibilities.
Machinbird is offline  
Old 4th Dec 2018, 10:47
  #78 (permalink)  
 
Join Date: Jan 2008
Location: uk
Posts: 857
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by Machinbird
Infrequentflyer, I think we can easily explain the WTF point on the graph you presented.
Siince the aircraft is not yet airborne, that is probably brief jet blast impingement on the nose of the aircraft.
Thanks, hadn't thought of that. Could the asymmetry between sides then be due to the aircraft turning at the time?

Additionally, Mr. Lemme has called attention to the difference in signal between left and right AOA while taxiing. The left AOA was noisy and the right AOA was rather smooth in that segment of the FDR data.
IMO that would be typical of a sneak circuit that is being excited by vibration transmitted by the landing gear into the fuselage.
Yep I've been thinking that way too, Mr Lemme doesn't seem to have picked up on this possibility - but that's maybe because his article was about sensor failure modes (depends at what point he picked his title I guess).

An intermittent wiring or connection fault, at or near the sensor, could account for the issues that triggered the sensor replacement and the worsening of the problem after it was replaced. If it is vibration triggered then ground testing might fail to find it (and the AOA sensor might test out fine). Be interesting to know if that AOA sensor had been replaced before or was factory fitted.
infrequentflyer789 is offline  
Old 4th Dec 2018, 15:11
  #79 (permalink)  
 
Join Date: Dec 2002
Location: UK
Posts: 2,451
Likes: 0
Received 9 Likes on 5 Posts
From the descriptions in this thread and external inks, #97, and particularly https://www.satcom.guru/2018/11/stabilizer-trim.html, the STS is a ‘crutch’ on late series 737 to meet low speed stability requirements.
MCAS appears to be a similar ‘crutch’ but addressing more specific nose up issues when approaching or at stall (25.203), and when turning where the trim would be more nose up, and where there may be a greater pitching moment associated with the new engines -737 MAX.
However, it is difficult to understand the effects of input failures (e.g. AoA) amongst the complex computations and interactions in these trim systems.

(Forgive the ‘Non PC’ quip, but a failure in heavily ‘crutched’ systems, literally leaves you with no leg to stand on.) - ‘close coupled’ systems.

The association of the Elevator Feel and Centering Unit with trim has been described, and the reasons for separating the (independent?) pneumatic ‘muscle’ from the electronic logic in the trim system - mechanical, elevator feedback.
Is it possible that a false AoA will trigger a change in the trimmed - elevator neutral shift unit due to a false change in the feel / centering unit ?
If so, then a consequence might be an erroneous trim datum, which may also be misplaced by MCAS and / or electric pitch trim, so that no stable trim condition can be achieved. A pilot will continuously ‘chase’ a stable pitch condition - there is no trimmed position for speed (trying to fly in-trim with the elevator offset).
Furthermore with similar mechanism, would trim / AoA offset cause the feel unit to position at it’s maximum extent, and so doing limit the available amount of stabilizer / elevator for control ? (and / or stick force limit).


Aerodynamic slat operation also appears as a connection within the various ‘trim’ computations, relating to AoA input.
Is slat extension possible based on a false AoA, but at much higher air speeds those normally expected ? Possibly with significant adverse pitching moments competing with pitch control and trim.
Or even with the ‘handed’, separate-side AoA inputs and computations, could there be asymmetric slat deployment.
(The slat thoughts above come from the apparent roll control issues in the Lion accident.)



safetypee is offline  
Old 4th Dec 2018, 21:11
  #80 (permalink)  
 
Join Date: Jan 2008
Location: uk
Posts: 857
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by safetypee
From the descriptions in this thread and external inks, #97, and particularly https://www.satcom.guru/2018/11/stabilizer-trim.html, the STS is a ‘crutch’ on late series 737 to meet low speed stability requirements.
MCAS appears to be a similar ‘crutch’ but addressing more specific nose up issues when approaching or at stall (25.203), and when turning where the trim would be more nose up, and where there may be a greater pitching moment associated with the new engines -737 MAX.

There was/is a similar mechanism added to the NG as part of speed trim function, my guess is MCAS evolved from it and replaced it. The NG AMM says:

Near stall, the speed trim function trims the stabilizer to a nose down condition to allow for trim above the stickshaker AOA and idle thrust. The trim continues until the stabilizer gets to its limits or the aft column cutout position is exceeded
However it looks like the activation conditions for MCAS may be wider (more of the envelope) and the trimming is more aggressive and ignores the aft column cutout, there are probably other differences too. It isn't clear (to me) that the NG function can kick in from just one AOA.

Originally Posted by safetypee
However, it is difficult to understand the effects of input failures (e.g. AoA) amongst the complex computations and interactions in these trim systems.

(Forgive the ‘Non PC’ quip, but a failure in heavily ‘crutched’ systems, literally leaves you with no leg to stand on.) - ‘close coupled’ systems.
I agree but I do think there is a wider more general issue (shown nicely in AF447) with close coupling / interdependence of airdata from "independent" probes that are in reality all of the same type and sat in the same outside environment. The redundancy is illusory, and the interdependence means that when we do get failures they cascade - bad AOA fails altitude and speed, bad speed fails altitude and AOA, etc. Tying everything together and correcting for everything to put "perfect" data in front of pilots is great when it works - when it doesn't, the pilots are left trying to disentangle multiple failures, with less data they can trust, and may end up focusing on the wrong thing.

Essentially the automation paradox again.

Is it possible that a false AoA will trigger a change in the trimmed - elevator neutral shift unit due to a false change in the feel / centering unit ?
[...]
Is slat extension possible based on a false AoA, but at much higher air speeds those normally expected ?
Been wondering about those two as well - can't be sure if it's possible or not on the NG from the AMM, MAX could do anything since this is an area we know they've changed.
infrequentflyer789 is offline  

Thread Tools
Search this Thread

Contact Us - Archive - Advertising - Cookie Policy - Privacy Statement - Terms of Service

Copyright © 2024 MH Sub I, LLC dba Internet Brands. All rights reserved. Use of this site indicates your consent to the Terms of Use.