737MAX Stab Trim architecture
 

STS/trim architecture
 

Salute!

I have an old FCOM, but no substitute for actual mechanical and electrical diagrams. And "code" of the black boxes wuld be even more valuable.

As you know, I am a fighter pilot that flew two very smart planes, then did system engineering for another 13 or 14 years in the weapon control and human interface aspects of military planes. I do not know anything about Councourse, as the public profile is scant, but from previous posts suspect technical savvy pilot or engineer of some sort.
+++++++++++++++++++++++++++++++++++++++++
- I found a pic of the AoA sensor on a 737, and the pilot side is below the cockpit and just behind the radome. Another contributor has verified my suspicion that removing the radome could influence the AoA "plumbing". That was an early post on the main thread. From our very own Dave;In the Viper, we used AoA for a few things. a) when gear up, AoA versus gee limit function, whereby 25 deg +/- was max AoA at 1 gee and 15 deg at 9 gees b) leading edge flap scheduling, which was a funcion of mach and AoA and c) gear down, AoA bias applied to the basic gee command and a warning horn above 15 deg AoA.
Three sources: two cones and one hemispherical probe that was used for many functions. It used middle value, not an average. As with the gee and rate sensors in our system, it used the most benign value when down to two sensors. Not sure of all the failure criteria, but I think we had to go down to one sensor and then system went to standby gains. My tech data is unclear about the criteria. But the plane flew well with no AoA inputs if you were gentle, and the rate sensors helped a lot in that situation.

1) I see some overlap in the STS and MCAS and even the "feel" implementation, and will have to look again at the FCOM. So we have three systems that might be using AoA, and the MCAS for sure. Tink STS uses only CA, is a poor man's speed stability system.

2) I will look back at the FCOM I have re: 'feel" implementation, which appears to use independent AoA and other air data and does not physically move the stab or elevators.
2) I do not like either MCAS or STS moving the stab the way it is described. I would use the Airbus implementation where elevator command is the driver, and use the 737 "feel" system to help pilots realize they needed trim ( like all the old planes did when we were taught to trim out the pressures we were holding). I don't like aerodynamically influencing the plane just to provide speed stability or the sense of speed stability unless you go whole hog fly by wire with a mechanical backup.

Gotta go, but will check back later.

Gums sends...

I was also wondering why they would use the trim instead of elevator feel to tailor the elevator forces, and FCeng made a post with a pretty good answer, in the STS thread. https://www.pprune.org/showthread.php?p=10303995

Salute!

I don't see eye to eye with FC about this issue My comments in red:

Gums sends,,,,

Salute!

@ concourse

Attached or embedded is the 737 pitch block diagram, but no way to tell what is pure electric or hydraulic or cables.

https://cimg6.ibsrv.net/gimg/pprune....a5956dbd40.jpg

Gums sends...

Gums: If you look at the dotted lines, they are all different. I hope the following edited fig makes sense.

https://cimg7.ibsrv.net/gimg/pprune....b1bd7b0759.png

If this accident turns out to be as people suspect right now will probably do more damage to the AI argument then good. If designers can't fix the GIGO problem computers have pilots will be here forever.

UHHH Most 737 stab trim motors are electric

http://www.eaton.com/ecm/idcplg?IdcService=GET_FILE&allowInterrupt=1&RevisionSelectio nMethod=LatestReleased&Rendition=Primary&&dDocName=PCT_20101 9

Which gives a pdf file DS400-1A_B737 Stab Trim Upgrade.pdf

All 737 models have electric stab trim motors.

@CONSO that's hydraulic for the elevator not for the trim.

Anyway without the text that normally accompanies that diagram it's not too good to understand the system.

Salute all!

Wow. Looks like a quorum of folks that wanna discuss the tragedy and "contribute". You never know when a regulator type or investigator or airline type digests our theories and such, but it might help to prevent another tragedy.
=================================================
You all know my background from the public profile, but too many "pruners" are "cute" with location and profile, including some here, so I always question some comments and assertions and opinions and..... The AF447 tech discussion was extremely technical and most contributors had extensive aviation and even "a/c type" background.
I was the lowly light puke that had flown the first fully FBW plane in service that had zero mechanical back up, had negative static stablity, and extreme body rates, AoA and gee values seen on almost every flight. But the heavy folks accepted me as well as Machinbird, and we got along fine.
I also worked with the human interfaces of the U.S. military platforms as a system engineer for about 14 years, so many of my opinions are based on "what would I do?", "How can I go back to a known condition?" , etc. And 99% of my control and display work was to prepare, target and launch a weapon. So I had to take into account many aspects of the process to ensure safety and reliability at the same time. With two no **** combat tours and only getting "shot down" once ( semi-crash landing with no power), and approx 4,000 hours in 5 warplanes plus some trainers, I was blessed to see many aero principles in action and all the "gotchya's". So with my location and bio out there on the Pprune profile and this missal, you all know where I come from.
++++++++++++++++++++++++++++++++++++++=
Present views/opinons on next post.

Gums presents...

They all need someone to blame when the GIGO offers up a lot of G.

@gums: salute! :)
I'll be sparse on comments as my core competency in pilot training, and in training systems for new models and new mods, is hampered by not knowing this system, and thus grasping the differences, well enough. I have discovered from those in the industry that training is viewed somewhat differently within the industry, and within each company, and within each corporate culture. Things that I had assumed are 'true' generically for pilot training, and crew training, may not be.

If you the pilot don't know what it(the bird) can and will do, what do you as a pilot do when it's doing that thing you don't know about?
(The test pilots I have worked with usually answer that with "get it back on the ground and don't go outside of known parameters on the way back" as often as not. )
Addressing that hole in the cheese is what draws me to the technical discussion.

Salute!

Concourse has raised a very important legal point for this discussion on the main thread, and I agree 100%, but my personal preference here is to keep the legal aspects "legal" and concentrate on understanding what happened and identifying whatever it takes to keep this from happening again or another incident that could have been prevented.

Some here may even be flying the type and could use some "education".
++++++++++++++++++++++++++++++++++++++++++++++=
Being a pilot on my accident boards, I mostly look for causal and contributing factors that were out of human control at the time or prevented the human from overcoming the problem. So my interest in sfwe and hdwe is a large part of my personal investigation. I have seen out-and-out pilot judgement/skill crashes, basic smoking holes, but also one serious loss that involved "little understood" aspects of our autopilot implementation. So this MCAS brings back some sad memories.
+++++++++++++++++++++++++++++++++++++++++++++++
The trim gearbox is of more interest to me now than my poor opinion of the decision to move large control surfaces to "help" the pilot and not tell the pilot That gearbox is the last mechanical connection in the chain, right? And the control surface it moves has more pitch authority than the elevators. So I start there and work back to the first input to the doofer, whether from a human input or Hal input or combination.

How did the STS logic fail to provide up trim cmds as speed increased when I only see down cmds to the gearbox. Maybe the AoA prevented that, but I did not see a direct input to STS from AoA. Seems like STS uses speed data from the ADIRU and not raw "q" from the probes. So if the ADIRU logic asserts airspeed is FUBAR due to AoA problems, then what does the STS do? Inquiring minds want to know before my next flight.

A poster not so far back thot we may have seen the result of a cascade of events that had deliterious effects not allowed for or imagined. i.e. a single point failure that by itself should have only been an irritant, but other systems that used the output of that "module" reacted as designed and the chain of events created control surface movement that was not required and, indeed, was unsafe and could not be mitigated by the crew. In other words, the crew did not recognize what the problem was and kept trimming because every time they trimmed, the plane seemed to return to "normal".

The crew actions, even without knowing about MCAS, are an important part of this tragedy, but if things didn't break, they would not have had the opportunity to diagnose a new malfunction. So second to last flight did not identify the real problem because the "runaway trim" procedure seemed to have allowed a successful flight and a writeup. The accident flight would have been the same, and maybe next hop would be the tragedy.

Gotta go, and forgive my sermons.

Gums....

A "malfunction that fixes itself." Red Flag.
What besides (maybe) an AoA probe 'broke' in this case? Still unclear to me.
Diagnosing malfunctions is, when a problem has persisted for flight after flight, something for a post maintenance check flight. (IMO) not a revenue flight. (Hence my question about MEL?)
But not an actual repair, it seems. So what is going on here: throwing parts in a problem, and hoping that fixes it? I'd hope the trouble shooting trees in the maintenance manuals would be a bit more expansive than that.
Maybe. depends on what actually broke.

Salute Wolf!

Here's my rationale for my opinion/theory/whatever regarding the AoA sensor and its influence on various aircraft systems.

The cascading failures or unexpected operations of several systems seems evident once you start with the AoA that appears up to 20 degrees different than the other AoA sensor. So let us see the flow chart/system diagram.[AoA data to the air data black boxes and the crew sees unreliable airspeed. Bad AoA flag might not help here if crew does not understand the MCAS, but relation with stick shaker would prolly be recognized by 99% of the crews ]

- At start of the data, one AoA shows about 10 - 12 degrees higher than the other. After liftoff the delta increases to maybe 20 degrees and tracks the other AoA sensor perfectly.
- the overall system fault monitoring system does not flag the AoA disagreement as such [after takeoff. It shows the crew unreliable airspeed once pitot speed is above "x" and AoA data is bad, like 20 degrees delta. ]
- At liftoff we see several trim inputs from the "system", which seems from STS, and they look like STS is doing its job. i.e. trim up to maintain the trimmed speed. But stick shaker is now active.
- flaps are now fully retracted, stick shaker continues because AoA from one vane is very high and the circus starts.
- Hal trims down, gums trims up a bit because we are climbing like normal, increasing speed and that crazy STS is wired backwards.
- Nose down trim geting really strong, but trim switch on yoke seems to be working, But after I let off for 5 seconds I see nose down trim again. This ain't STS, I think.

Somewhere in the scenario the "feel" system is acting on the control yoke, and I do not have the personal flight experience to comment on how it feels and not sure about how it is supposed to work in concert with the other sub-systems and their logic/specs when they go south.

Meanwhile, pilot and MCAS are commanding up trim and down trim for 5 or 10 second intervals until something big happens and we see lotta power commanded and a divergent control pressure for each yoke.

Gums...

I'm not sure. In effect the aircraft was repeatedly trimming against them, now I've read some of the archived discussions on Speed Trim/ STS both here and elsewhere and there are plenty of comments saying (paraphrased) "it keeps trimming against me", and plenty of replies saying "that's what it is designed to do". So this could easily be misidentified as STS simply being more enthusiastic than usual.

We may know now that this was MCAS, but that is knowledge the pilots didn't have. What they did have (on previous flight) was UAS, which might reasonably be expected to cause odd behaviour from the speed trim system, plus stick shaker. So UAS is what got written up and "fixed".

If the pilots had the AOA disagree (an optional item they apparently didn't have) then maybe that would get written up, if they had known about MCAS maybe they would have suggested that as a cause and AOA would have been identified from that.

We don't know if the probe broke or was mis-fitted or why the previous probe was replaced. All we know is the the recorded value was 20deg out. I think these are analog signals so could even be a wiring fault?

Agree on check flight - but easy with benefit of hindsight. MELs and AMMs possibly unlikely to be helpful when the system showing the symptoms is undocumented and the sensor that is (maybe) broken gives no disagree error because the values are never shown to the pilots.

I have acquired some chapters from NG AMM which I think cover what you are looking for, but it seems to have changed for the MAX, see this post in the R&N thread.

And of course the NG doesn't have MCAS. It does have (very lightly documented) a similar sounding function, which is not named, which gives ND trim on the STS trim output (which is also the autopilot trim output) from the FCC. So in the NG case, no wiring changes, only software changes in FCC. Max is probably the same except at the column switch module (column position cutouts) and the cutouts on the console. And the software.

Unfortunately MAX AMMs are not likely to be out in the wild yet as the aircraft is in service in only a few places. Would be very interesting to compare one though (chapter 22)...

Anyone who has ever worked in airline maintenance will have encountered that.

Before we hang the maintenance organisation out to dry, we really ought to wait until we know what was recorded in the tech log for those previous flights.

Can someone who works on current 737 models provide some information on the actual AOA sensors used? Some research on the subject shows that Boeing is using UTC made equipment on the 737.
I believe that they are based on a resolver principle https://en.wikipedia.org/wiki/Resolver_(electrical) , in which case there will be two AC signals coming out of each AOA sensor forming a sine and cosine signal pair that must be put together to readout the angle.
Since both right and left sensors appear to have been following the AOA, but the left sensor had a 20 degree offset, the point in the aircraft where the sine-cosine signal pair are combined before digitizing would be where to look for the source of the AOA error. Loss of either of the AC signals would create a stationary AOA reading.
I know that this is a bit technical for most people, but it might bear fruit as to the likely source of the AOA malfunction. .

Yeah, which is why I asked. I've been involved in aircraft maintenance (though some years ago). I am not out to hang the maintenance organization, David. With this being a comparatively new bird, I'd love to see what back and forth the maintenance team had with their Boeing rep over this fault/malfunction ... but that's likely never going to come out in the public domain. I've seen a wide variety of OEM "expertise" in the field: some great, some not so great.
@PEI: Thanks, except the second like gives me "page not found" (I read your #1650 and #1674 by scanning the page, I think that is what you were pointing to).
@Infrequent: Thanks.
@gums: copy all, nothing further, out.

I can't meet that first requirement but in the absence of any more authoritative answer, I think you are likely correct based on 737NG information published (which matches some NG AMM chapters I might have acquired) here. Search on that page for "SMYD Analog Int" and you'll find the SMYD diagram where in the top left we see "sin / cos / com" analog inputs from AOA sensor - bingo. Assuming it hasn't changed between NG and MAX.

However, although SMYD implements stick shaker, it doesn't do speed trim, that is the FCC (and they got both). The text beneath that diagram shows that SMYD outputs AOA to data acquisition, but not FCC/ADIRU. FCC gets AOA from ADIRU (where it is used in speed calculation), and as far as I can see ADIRU also gets it direct from the sensor.

Pretty sure (very rusty...) that decoding resolver output for position (not angular velocity) is just a matter of relative voltage magnitude of the sin/cos signals, so wouldn't be surprising to see that duplicated. That means you'd need to have two simultaneous conversion failures. I think, and if that bit is all still same as NG.

Partial loss, or rather additional impedance in one signal line might do what? I think it could lead to offset reading, not sure it would be a consistent offset though, been doing software-only for too long.

Broken sensor or misfitted sensor is probably the place to look first - not least because identifying a wiring issue now is going to be tricky...

One idea, does the tail plane trim motor have any thermal protection that could have tripped in any way rendering the motor useless at the end?

Salute!

- I still look for evidence of a mechanical failure, but looking at the traces..... I see no clear evidence of mechanical failure, although the overheat possibility put forward may be relevant. And then there's previous flight and maybe not enuf lube on the screw as we saw in Alaska crash. I am leaning more and more toward an electronic cause by Hal and his minions. They seem to be able to move the stab without pilot consent.
- Stab is way more powerful than the elevators. Even with full deflection of the elevator, once above a certain "q", the elevator will not be able to overcome full stab deflection.
- No big rate changes on the stab position. Same slope of movement as all previous down cmds.
- At "the event", engine data shows significant power demand for a few seconds. and then one large down command from Hal, and stab moves. Another large down cmd then comes from Hal.
- By now, we see large control forces by both folks and eventually 100 pounds by the pilot. Damn, but this is scary.

Gums posits....

Wouldn't the increase in airspeed and high opposite elevator input fit the Boeing warnings for the 727 and older 737s that you may need to relax elevator input to allow trim in the same direction as your input. The elevator airload overpowers the jackscrew leverage. They might have been pulling and trimming but the stab wasn't moving. I saw this in a 727 simulator a long time ago. Is that warning still in the newer 737 FCOMs?

Salute Jim!

Worth talking about the elevator authority and such.

High "q" will definitely make that big stab overpower the much smaller elevator.

Apparently, this bird ( 737) seems to act like most of us expected in the old days and even today until they let us fly a plane with negative or neutral static stability[see added below] . We trimmed to reduce control pressure/movement. That got you back t the original AoA/speed. So you could have a lot of elevator deflection but the trim operation would allow less and less control pressure until you reached the desired trim for speed/AoA. And then we had STS, which trimmed the stabilizer without pilot input. Huh???? This ain't no Airbus fly by wire FLCS. But rational was to tell the pilot ( I use pilot to assert whoever was in charge of moving controls), he needed to trim for the new speed/AoA. What airplane have they flown that did not need a new trim after changin speed/AoA?? Do they need to be "reminded"? SHeesh.

And then we get another tweak to the plane's flight control system.. MCAS. And its operation and such are not well known by the rank-and-file aircrew. How much interplay between MCAS and STS and the articicial pitch "feel"?

Gums sends...

ADDED: I wish to clarify my assertion about static stability or speed stability. Some folks think the FBW Airbus planes have neutral speed stability. So most of the time speed is controlled by power. The fact is that at least the AB330 that went into the ocean (AF447) showed very good stability and fairly benign stall characteristics. No violent wing rock or buffet. The stall "protection" was inhibited due to the low airspeed value feeding the flight control computers, so HAL did not keep the pilot from commanding back stick and the stabilizer from trimming to full nose up. From reading about the Concorde, which had a lotta fly by wire, even it had conventional speed stability.
The Boeing STS is a funny way of telling pilots they have to re-trim if speed changes. Duhhh? Oh well, I would rather have the plane try to get back to last trimmed speed and I can trim up or down to achieve the new desired climb angle or new airspeed//AoA. And remember that we' re really trimming to AoA. I do not unnerstan moving a large control surface to tell pilots they need to trim versus using the "feel" system that the plane already has and has done fine for decades.

Who needs a POGO?
 

Tucked into your control system pathways, you will also find a little gadget called a POGO. These give you the ability to maintain control although your flight control system is partially jammed. This item fits the category of "mechanical override systems" referred to in the previous posting. Rather than describe them myself, I figured that I would let a professional take that task. See Pogo Load Limiter
Now take a look at this simplified diagram of the elevator control on the 737NG (courtesy of Peter Lemme's Blog) Do you see the POGO?
Well, I said this is simplified. There has to be another set of Pogo-like mechanisms connecting the left and right control cables to the elevator input torque tube or else there would be no point to having the breakout mechanism discussed in my previous post.The two columns would be locked together through the elevator input torque tube. Consider the implications of what happens when the crew hauls back hard against the elevator feel and centering unit resistance!!
https://cimg0.ibsrv.net/gimg/pprune....a953e5fbec.png

737-300 Elevator Control

I wonder if the systems incorporated may prove to be installed to satisfy some certification clause, but not actually bringing any real benefit in the ability of the pilot to exercise control.

The helicopter industry had just this occur with the introduction of the pitch bias actuator (PBA) on the S-76 and Blackhawk.

The FAA requires that you must have Longitudinal Static Stability (LSS), where the stick position for any speed must be ahead of a lesser speed. Without the PBA the helos demonstrated negative stick gradient in certain areas of the envelope, which certification said was a no, no. The PBA provides longitudinal cyclic displacement proportional to airspeed. The DAFCS commands the PBA as a function of pitch attitude, pitch rate, and airspeed. Failure modes were,

Salute Megan!
You can attribute my quotes, but I recognized it right away.
From megan
I feel that Megan has nailed it above, but it would be nice for us to see the pitch authority at various AoA and stick force gradient that seems to be a factor in certification. Many here are real aero folks and mostly pilots. We can handle it.

Being from the military "cert" community, I flew a few designs that would prolly have not been certified by the U.S. or other country aviation bureaus. I even flew one that failed several pitch and roll gradient specs, but USAF produce a few hundred and we flew the suckers because we had good checkouts and were briefed at length on the "waivers". I contrast this with the hapless crew on that Lion Air jet.

Added..... I flew one type that behaved just like all the aerodynamic classes said it should. Right up to about a degree or so of the critical AoA, and then whahooo!!! It was the VooDoo, so youse can look it up, but the real explantion is hard to find. The thing gave plenty warning when subsonic unless you yanked back real quick. But we had a "pusher" that used AoA and pitch rate and stick force to "help" us. However, the stick force per gee and rate and such was perfectly within the military certification requiremnts until that tiny degree of AoA before the pitch-up. And then any more back stick /AoA exhibited what we called the stick "getting light", and if you persisted, then you witnessed the wahoo, heh heh.

Glad we can discuss stuff here. Guess the casual folks on the main forum think they must be aero engineers of test pilots to be here.

Gums sends...

Salute PEI ,

Yeppers, I saw the great posts and such and learned a lot, but the other thread did not get into the MCAS for about 60 + posts. It is a really great discussion about stability and how the STS works and why.

Since the MCAS seems to be of major concern for the Lion accident, and because my strong feeling is we pilots should know about a significant change to our flight controls, I gravitated to this thread..

One thing that mabe FCeng can explain is the interaction between STS and MCAS. Seems to this old pilot that at certain flight conditions they are opposing each other. In other words if speed/AoA is changing, then one system is trimming opposite the other. Or am I way off base?

Gums asks...

There would have to be a predetermined hierarchy of precedence of simultaneous/opposing inputs. And surely MCAS (i.e., the high alpha protection) would be above STS and its general speed stability. And this is nothing special, my jet has electric trim only with probably half a dozen inputs, each in its order in this type of hierarchy.

Me too. I'm on a few US-centric forums and the main topics there seem to be about industry complaints. But if you get too much into technical stuff (except maybe ATC procedures) you're too much of a nerd.