Help Diagnosing Unintended/Uncommanded Acceleration Issue

Since it wasn't a crash - it easily *could* have been though - but rather an "incident," separate people at NTSB were assigned to clear the report

Since pilot error is usually the cause of most incidents, they naturally blamed the pilots the first time around.

Once it became apparent similar issues were experienced by different flight crews, on the same aircraft, the database immediately flagged this.

With further investigation, the sidestick was replaced, and Airbus introduced hardware changes (A resister to introduce a degree of damping to the signal) and software changes (Desensitize control inputs with Flaps 2 or Flaps 3 setting). This is supposed to be applied to all Airbus A320's of that series

Since I have background with fly-by-wire and industrial process control (The same company, Honeywell, makes systems for both, with similar design), I don't automatically reject claims of odd behavior. I've seen enough odd problems caused by "simple" connector corrosion to personally determine that we're putting a bit too much faith in these systems

Eg: these systems are not as infallible as the engineers and fanboys would like us to believe. Wild acceleration, such as Hemi V8 style acceleration from a Prius, is highly improbable

But uncommanded acceleration, yes, I have no doubt it *can* happen, and as the fleet ages, *will* happen. Statistically, it *must* happen

 

What got my attention is that for a "common" database, the first report did not get updated with the latest findings, but was totally uncoordinated with the second report. Or as you clearly state, since it was not a crash, it got insufficent attention.

The counterpart to this is how to pursue automobile issues of the same nature, like the one covered in this thread. If a proven Airbus problem gets "brushed off" as indicated in the first report, then it's got to be immensely harder to get real (but rare) auto issues identified and taken seriously. chogen2 listed all the similar statistics, but raw reports statistics does not tell us which ones were "real problems" vs. operator error. Obviously, most of those are indeed operator induced issues, but what percent are real? 0.01% or 5%?

My starting suggestion would be a comprehensive "flight recorder" provided to the dealers that could be attached with the owners permission. Comments?

 

alright. so there are two throttle position sensors. one is on the accelerator pedal and the other is on the throttle body. their readings are matched up. if one mismatches with the other, there will be a code thrown and it sends you into limp-home mode.

DH cannot think of any situation that would cause this problem and not throw a code.

so it's not hardware. if it's software, as others here seem to be saying, then why is this only happening in this one vehicle?

 

The TPS and pedal agreement is just to ensure relevant control input to output. Eg: if the throttle blade were to stick due to heavy deposits in the bore and on the blade, that would be similar to the industrial valve problem I discussed, eg "stiction"

But if there is an internal failure in the pedal assembly, as Bob documented with that Gen I pedal assembly, there is no way for the computer to determine if it's a spurious signal or a genuine control input

Since there are situations where the driver may have to floor it, the car is programmed to allow that sort of control input.

The only way to determine if a control input is "valid" is to have 2-3 different sensors on the accelerator pedal assembly, with separate loop wiring back to the computer. Software "voting" would then be used to determine if the control input is valid

That approach has obvious drawbacks. Even in the case of the Airbus incident with NorthWest Airlines that I discussed, the deposits in the sidestick potentiometer track were a common failure point. Eg a single mode failure that was - by design - impossible to avoid

In that case, the aircraft responded to the spurious control input from the sidestick. If the sidestick would have failed in a full-deflection mode, a *very* dangerous event, there is an override button on the other sidestick. THe other pilot could have hit the override button, and assumed control again

As a result of that Airbus incident, changes were made to the sidestick (Better sealing to keep trash out of the potentiometer tracks), the wiring (A resister was added to "dampen" spikes), and the software (Aircraft sensitivity in Flaps 2, Flaps 3, and Flaps Full settings was detuned)

This problem may be rare for now, but as the fleet ages I'm sure we'll see more of these issues.

IN case any lurkers are ready to pounce on this incident as a uniquely Prius problem, it isn't. My FJ Cruiser also has throttle-by-wire. Almost every new car and pickup has throttle-by-wire.

I would imagine if the accelerator pedal assembly failed full input on a 2009 Dodge Ram hemi, there would be one truly spectacular burnout at a red light

 

Now that I don't have to fiddle with figuring out the FARS database, I have time to look at this problem properly ... after I get my GPS data extraction routine working.

Bob Wilson

 

the accelerator pedal has two sensors, actually- one to send the input, the other to verify it is real. if there is a discrepancy, it sets a code.

if there is a discrepancy between the accel pedal and the throttle body, it sets a code and drops into fail-safe mode.

effectively, both accel pedal sensors would have to fail in a manner in which the output remains within the bounds and they both agree. the likelihood of this happening is ~0.

now. if this is a software issue, it's either in the ECM or the HV ECU. if the car is still under warranty, the owner needs to go to the dealership and tell them if they cannot fix the car, because this is a SAFETY ISSUE (key words), he wants an even trade for a new car. both computers should be replaced. even if they only do one, if it happens again after the warranty expires and there is record of a complaint prior to warranty expiration, the other should be covered because the issue happened prior to the warranty expiration.

if they open a repair order, and tell you it cannot be fixed, tell them you will not accept the car back and call the toyota customer experience center.

you want to focus on the word "safety" and how you do not feel safe in the car.

 

Wouldn't something mechanically snagging the pedal cause this?

 

yes, but that is different from the sensors failing. that is something pulling the pedal down outside of the driver's control. which is interpreted as a valid signal, because... technically it is. something is physically changing the position of the pedal.

 

Ok, this thing about the accelerator pedal has been bugging me. I am also thinking about the intermittant problems with the Airbus fly-by-wire sidestick, as in neither case was a code stored

The Prius accelerator pedal sensor schematic



Some of the possible DTC associated with the accelerator pedal



However, thinking of that pedal assembly Bob took apart, a *common* or single mode failure, very similar to what happened to the Northwest Airlines Airbus sidestick controller, may very well impact the signal output uniformly and *not* meet the requirements to set a DTC

The car actually would think the pedal was physically moved. Like in the case of that Airbus having the uncommanded roll, the pilots probably both yelled "Wth!" but the aircraft behaved as if it was a valid control input

This would be a very unusual and rare failure, as it was for the Airbus and as it probably is for the Prius - or any vehicle with throttle-by-wire. However, it's not statistically impossible

In the case of Airbus, since safety was an overriding priority, all sidestick controllers had to be inspected for wear and foreign material, eg spilled coffee. Changes were made to the wiring (Resister added to dampen voltage spikes) and to the software (Aircraft "detuned" to be much less sensitive in Flaps 2, Flaps 3, and Flaps Full)

Similar issues are present in the fly-by-wire Boeing 777. The control column sensors call for routine inspection and servicing

 

As far as I know, NHTSA might order an investigation if enough of the same/similar issues are reported by drivers. The auto manufacturer might initiate their own w/o NHTSA's order. This is how many recalls come about.

I'm unclear if there is an absolute threshold in terms of absolute numbers or a percentage that might be the trigger.

Unfortunately, some of the "unintended" acceleration issues are user error (driving stepping on the accelerator instead of the brake) or floor mat issues.

 

I recall in the case of the Ford Explorer / Firestone tire issue, the trigger was that the Firestone tires fitted on the Explorer had a much higher trend separation failure rate compared to other tires of the same type. When they investigated further, the "bad" tires were mostly a single size that came from a single Firestone factory, so they initially recalled just that size produced at that factory. Later, when it was determined to be a design flaw, Ford recalled addtional sizes of the Firestone tires.

So, in triggering an investigation, they are looking for a higher incidence of the problem compared to other similar cars.

 

On the first point, I suspect the seriousness of the issue also heavily weights the trigger

On the second point, yes there is enough "noise" that even NHTSA may conclude driver error

To bring up the example - again - of the Airbus with the sidestick problem, this was an *extremely* rare problem for the fleet. However, an uncommanded roll, especially in a takeoff or landing phase, could have fatal consequences

Even though the Airbus is designed with an "override" button on the sidestick, so the other pilot can take control from an errant sidestick, the momentary delay in doing so could be pretty serious

The NTSB database entries for the same tail number of the Airbus illustrate how quickly they escalated the seriousness of the incident. It went from being a pilot issue to being a major hardware/software inspection and repair issue fleet-wide

 

WANTED: NHW20 accelerator assembly.

Suggestions?

I think it is time to get one and take it apart to see how it goes together. Alternatively, ask the original poster if they would be willing to swap a salvage with their accelerator assembly for detailed analysis?

Bob Wilson

 

Bob

Now you're talking. As far as pinouts, the Gen I accel pedal sensor you took apart hints it has a lot in common with the Gen II. Although two hall effect sensors are involved, a failure of the common shaft could cause problems

Apparently, according to the shop manual, later models (+2005??) will monitor for "noise" like the kind you recorded with the full range travel. There is a DTC for that, but like all things, there is always a chance of a failure mode not anticipated and/or monitored for

 

um, ok, if you insist. good luck.

 

You aren't going to get the dealership to fix this problem because they probably think you are a crackpot. You have a couple of options:

1. Continue as normal, maybe it will go away, maybe it won't

2. Try to reproduce the same conditions when the problem occurred - forcing the failure and hopefully taking measurements to determine Wth is going on

3. Shotgun approach to solve problem - clean (or replace) TB, examine (or replace) accel pedal, inspect all connectors to ECU, etc.

4. Trade it in on another car.

 

This comment is silly. Galaxee has more real world experience than you can imagine.

Tom