Engineers: How did VW actually do it?

[aap1966]

Seriously.
We all know (?) it was a "software" mediated get around, but exactly how was it achieved? How did the car "know" it was being tested, and what did it do differently to return the rigged results.
(Assume ignorance but intelligence in any answers.)

Thanks.

[Charles Freeborn]

As I understand the car's brain can detect when the analyzer is plugged in and uses a different set of parameters than when it's being driven normally. I wouldn't imagine it's all that hard for the software to tell when it's being mounted, so to speak...

[aap1966]

Yeah, but is it that the car is stationary but with wheels turning, or the car following an arbitrary pattern of engine loads that the computer recognises as a test pattern?
Did the computer then alter the air:fuel ratios? Or run the car hotter (or cooler)?

Morality aside, the engineering here strikes me as borderline brilliant. So I'm curious to know how they pulled it off so well, I mean, they've been getting away with it for years.

Also, how many in the company knew? The knowledge must have been pretty effectively quarantined for not one "disgruntled employee" to have made a quite phone call to EPA / Greenpeace / NYT, who-ever?

[pavulon]

Front wheels rolling, back wheels stationary = emissions test?

[pavulon]

If the "fix" is a software flash, how long until tuners are selling a stage zero re-flash returning it to baseline?

[enzo1]

[asphaltgambler]

The 'firmware' is coded to communicate at specific baud rate by each manufacturer. OBDII is set to a much slower ( 10 Mbps - I believe). When any OBDII based tool is connected and starts the 'handshake' @10mbps - the firmware recognizes it and performs what ever task is coded and manipulates what it has control of

[Tervuren]

There are lots of ways.

If tested on a dyno, only the driven wheels are moving. Detect a difference with the ABS/Traction control of one set stopped, other moving, and its easy.

Detect a control being plugged in, easy.

Parking sensors detect the car is in a room/confined space, easy.

[rick-l]

There are three ways stated so far

1. Diagnostic tool plugged in
2. drive cycle pattern
3. wheel speed difference between front and back

Which one is it? All three have drawbacks.

From the quoted article... Is the EPA test chamber barometrically controlled? I'd believe the article if that were true

[robertocravallo]

Ask the guys at CATERPILLAR. They got caught doing the exact same thing a while back. So it's not german engenering after all...

[HardDrive]

What is astonishing about this whole thing is how many people had to know about it. There is no way some rogue engineer did this. A lot of people were involved, and someone told them to do it. One has to wonder why the CEO stepped down so quickly. It very well go right to the top. And then one must ask....did the German government know about it?

[1990C4S]

Originally Posted by rick-l

There are three ways stated so far

1. Diagnostic tool plugged in
2. drive cycle pattern
3. wheel speed difference between front and back

Which one is it? All three have drawbacks.

From the quoted article... Is the EPA test chamber barometrically controlled? I'd believe the article if that were true

1. is not the answer, otherwise a Scangauge or OBDII reader would put the car into test mode, which apparently does not happen.

Wheel speed on a dyno is reportedly one component (front wheels turning, rear wheels stationary). But if they used this method Europe it would fail, as they sell AWD diesels there.

There is a steering angle sensor built into the ZF electric power steering box, so the car knows exactly what the steering angle is, and I would guess the car is not occupied for an EPA test.

The speeds use by the EPA are very closely controlled, with 0.1 mph I believe, so a few seconds at either the low or the high speed would be another factor. http://www3.epa.gov/nvfel/methods/la92dds.txt



And yes, I believe the atmospheric pressure is controlled.

[Tervuren]

Originally Posted by 1990C4S

1. is not the answer, otherwise a Scangauge or OBDII reader would put the car into test mode, which apparently does not happen.

Wheel speed on a dyno is reportedly one component (front wheels turning, rear wheels stationary). But if they used this method Europe it would fail, as they sell AWD diesels there.

There is a steering angle sensor built into the ZF electric power steering box, so the car knows exactly what the steering angle is, and I would guess the car is not occupied for an EPA test.

The speeds use by the EPA are very closely controlled, with 0.1 mph I believe, so a few seconds at either the low or the high speed would be another factor. http://www3.epa.gov/nvfel/methods/la92dds.txt



And yes, I believe the atmospheric pressure is controlled.

They didn't cheat European emissions regulations, the USA has its set of "extras".

[cockerpunk]

they used the wheel position sensor.

wheels turned: driving
wheels straight on + other things: testing

fairly easy to detect that the test was taking place. they punish the method, not to tough to find that one combination of throttle inputs, and other control inputs to detect the test method.

[red-beard]

How did the cheat work?

Just speculation, but a richer mixture and more EGR would reduce NOx

[PetrolBlueSC]

IF () THEN () statement. If hooked up to equipment, then operate in test mode.

[GH85Carrera]

The fallout from the Volkswagen diesel scandal is spreading fast to the company's other famous brands, including Porsche and Audi, and across the Atlantic to the U.S. The scandal reached down into the company's engineering corps as the CEO of Volkswagen's US business, the research and development chief from Audi and the engine chief from Porsche, which are part of the Volkswagen Group, are said to be following Volkswagen's CEO out the door of the company, according to multiple reports Thursday. The impending departures are a sign that the Volkswagen scandal is ready to grow to much larger proportions.

and how it can be done

Normally it is a case of pointing the finger down the food chain until someone cannot point any further.
Engineers are normally the Middle Class people, They don't have the money for a prolonged out of work. So there is only so much risk they would take at their job. It would be nice that everyone would be brave enough to stand up and say "This is wrong, I won't do it!" however with the risk of getting fired, plenty of the power is still in the higher ups.
Also it is quite possible that the engineers could design something without full realization on what they are doing.
I need you to code a function when variables are between this range, that returns a value in this range.
The function is created.
Then it will go to an other engineer, We have the function to optimize fuel efficiency in place, could you add this function to your code.
The function is added.
The higher ups can organize their orders so the engineers doesn't have the full picture of the scope. However when things goes down they will see in the comment that engineer who had made the function and the other who added it. They get canned, for doing their work, and never had an inkling on what they were doing on the grand scheme of things.

and how it was found
Sales of Volkswagen diesels have been on the rise in the US in the last few years, despite this more stringent regulation. ‘Clean diesel’ is a goal for the industry, and a European non-profit, the International Council for Clean Transportation, got in touch with researchers at the Center for Alternative Fuels and Engine Emissions (CAFEE) at West Virginia University. The question of how Volkswagen was able to produce a diesel engine in compliance with US regulations was high on the list of priorities, and the team was well-equipped to test Passats, Jettas, and Golfs in real-world conditions.
Despite passing emissions tests required by the US government, CAFEE found these engines were out of compliance. This apparent discontinuity can be brought to light by examining how diesel emissions are measured. An EPA notice of violation explains this was done by a bit of code functioning as a ‘defeat device’ that would sense when the vehicle was under test. Software installed in the electronic control module (ECM) would detect when the car was undergoing emissions testing by reading, “various inputs including the position of the steering wheel, vehicle speed, the duration of the engine’s operation, and barometric pressure.” Under these conditions, the ECM would use a different ‘map’ that would reduce torque and NOx emissions. Under normal conditions – when the vehicle was not being tested for emissions – a separate ‘map’ would be used that would increase acceleration, torque, or fuel economy. CAFEE was able to determine this because of a portable testing rig; instead of testing emissions in a garage on a dynamometer, the researchers performed their tests in real-world conditions, driving around Los Angeles, from LA to Mount Baldy, and from LA to San Diego.

[John Rogers]

When I had my 2000 Ford F350 7.3 diesel tow vehicle I added a different intake and 5 inch exhaust both by AFE and an aftermarket programmable Edge with Attitude computer that had "stock", "economy" "tow" and "race" preset modes. The truck was ODBII compliant and when CA started checking diesels for smog they plugged in their reader and looked at the computer settings, blipped the throttle to check for excess smoke and that was all. I had the Edge unit in "stock" mode and the smog fellow never saw any thing. He did keep looking for a catalytic converter as one of his manuals said 2000 year Fords were supposed to have one but the others said no. The sticker on the right side valve cover did not list one so I passed.

I noted the readings which were the time the injectors fired, when they fired as to engine RPM and fuel pressure. I plugged my AutoEnginuity software in after I got home and pulled up the data windows for these readings and then changed the computer settings to see what happened and the biggest shock was in "economy" mode where injector fire time was about 65% of stock which is why the millage went up to 20 MPG vice 14 MPG for stock! The software would tell the brain in the truck, hey here I am, I want to get some readings so when ever I needed to read values they were handy. I would guess that since the function of the pinouts are readily known, what they do for various model car/engine combos is also known that it is easy to have the software tweak the settings.

I did note that the last smog test I had run on my Chevy HHR 2.4L, the tester plugged in his wireless reader, read things for about 2 minutes and I was done. No rolling road, no inspection or anything?!

[1990C4S]

Originally Posted by Tervuren

They didn't cheat European emissions regulations, the USA has its set of "extras".

Then why are 11,000,000 vehicles affected?

[wildthing]

This would've all been fine if they just put three manual settings: ECO, Sport, Sport Plus, or whatever they want to call it.