How Much Power Does the Volkswagen TDI Lose in “Cheater” Mode? | Hacker News Reader

How Much Power Does the Volkswagen TDI Lose in “Cheater” Mode?

tflcar.com

67 pointsdanso46 comments

This test seems very poor. They didn't measure the NOx emissions so we don't know if it really entered "cheater mode". According to a comment there, it was probably trying to compensate for what it detected as the rear wheels spinning and they need to turn off DSP. It doesn't sound like they really knew what they were doing so I'm not sure I trust it.

Yeah, also a power run on a dyno is nothing like the sustained cruising-speed tests carried out for emissions tests. I'm not sure that during a power run the car would realize it is being tested. At least, it wouldn't if I was the one implementing the cheat-mode based on my knowledge of emissions testing procedures.

All of this testing can be seen as very poor, however, I would argue that we are running the wrong tests. Let me explain...

Imagine you were developing a car. As part of the project you would need to test, test, test and test the drivetrain and the engine. For your test data to be relevant you would want a controlled environment, e.g. car on perfectly flat road with no pot holes, turns or aerodynamic consideration. You could then adjust the variables and check the results - power, torque, emissions, noise level, oil use, fuel use, temperatures and so on.

In this development mode the test rig works pretty good. Figures can be produced that are important for the engine/drivetrain development. Variables can be changed and results measured. The figures - defeat device aside - are actually true with the normal margins for statistical measurement (not every engine will be 100% exactly the same even if it came off the same production batch).

The problem - defeat device aside - is when regulatory bodies use this data for what they need to know, i.e. real world performance. They accept data that they know is not real world and accept it as 'fact'.

If we look at our own testing for building apps, websites and such like we take a modular approach, testing our dev boxes with some type of 'seige' that does not factor in real world internet connectivity and bandwidth, not to mention how customers might use our products in the wild. We can even prove to our clients that their site isn't slow, sharing our metrics with them. Yet, in the real world things don't quite attain those metrics. Luckily nobody gets harmed if our 'squirrel picture app for cats' app falls short.

So, it is the procedures that are wrong and our standards bodies that are doing it wrong (by blindly accepting manufacturers' test data). We are lucky that VW have been cheating as we now are having the conversation about the testing methodology.

For your test data to be relevant you would want a controlled environment, e.g. car on perfectly flat road with no pot holes, turns or aerodynamic consideration. You could then adjust the variables and check the results - power, torque, emissions, noise level, oil use, fuel use, temperatures and so on.

That's not how cars are developed. There is some of this for sure, but even "static" dyno tests involve varied workloads to simulate an urban environment. The European Driving Cycle[1] is a fairly basic example of this.

[1] https://en.wikipedia.org/wiki/New_European_Driving_Cycle

I have an idea for a different test method.

You take your cheating Volkswagen TDI without software fix to the local dragstrip on a day with little wind and do a series of test runs, you document your 1/4 mile time.

Then you send in your car for a fix to Volkswagen most likely a ECU software update as that is cheapest fix.

You retest your car at the drag racing strip, record the same number of runs.

If the numbers between the test run differ much you can sue Volkswagen as the car may have worse performance with the legal Nox emission software.

Here is a device that will let you test your car perfomance on normal roads. http://www.vboxmotorsport.co.uk/index.php/en/products/perfor...

Disclaimer I own a car with the affected engine.

That's a better test method, but since the VW fix is not available, it's worth trying something else. Although in this case, it seems the methodology is suspect: traction control applied brakes may be the source of the lost horsepower, and emissions weren't measured so there's no confirmation that the enhanced emission control mode was enabled either.

The methodology here is highly suspect. First of all they are using an AWD dyno which can be unlinked to test a FWD or RWD car. It is expected that the car will dyno lower when driving the rear wheels as well as the front. Not only does it have to turn the drums of the dyno but also the rear wheels, tires, and brakes of the vehicle. Second, a key component of the cheat as reported by the media is keying off of the presence of the OBDII instrumentation. As far as I can tell they did not have any way of simulating this condition.

Not all emissions tests are conducted with OBD2 link. In some relaxed regimes, OBD2 interrogation is the only component of the emissions test, and the car passes if the car determined there is no faults. I lived in a county with emissions testing and the OBD2 link was used as a pre-test before the dyno run, and to feed the VIN number into the state's computer to provide some assurance about the identity of the car being tested.

Some vehicles, like my Toyota, indeed behave slightly differently if there is a diagnosis computer connected. I don't know if this applies to the OBD2 link as well as the toyota-specific connector under the hood, but it's pointed out in the factory shop manual. At any rate this knowledge has prevented me from trying the "scangauge" type of data collection device.

Not really. The car is a FWD car. So when the rear wheels aren't being spun up, the car should produce ever so slightly more measurable power instead of less. Because, as you stated, it didn't have to spin up the rear wheels, etc.

The dyno can tell if both wheels are being driven or just the front ones. If it couldn't, then the results wouldn't be off by 10% but they'd be off by 50% as the dyno wouldn't understand that by having the two drums linked they got twice as heavy and it'd do the calculations based off of the inertia of a single drum instead of both drums.

>will dyno lower when driving the rear wheels as well as the front

well, it did BETTER on four wheels spinning, so how is that a bad test?

> The acceptable particulate level in Colorado emissions testing is 35% opacity. That means the smoke and particulates that are coming out of the exhaust pipe can block up to 35% of light and pass the test. This is a lot of smoke.

This is the most shocking part of this... can you really pass with that much smoke coming out the tailpipe? That's absurd and I find it hard to believe. What's even the point of testing based on those regulations?

The 35% is a meaningless number without more details. It's not shocking at all if a 1km column of smoke blocks more than 35% of the light passing along it. So we need to know more about the test to decide if it's shocking or not - specifically the distance the light travels through the sample.

Integration time is also important. Short bursts of dense smoke can be equally polluting as a continuous stream of light smoke.

While not conclusive, I give these guys credit for replicating a behavior fitting the "cheater" profile. But I am surprised that not spinning up the rear tires triggered that behavior.

The loss in power measured here will be very noticeable to diesel owners. Diesels have narrow power bands. You don't spend much driving time anywhere near the top hp zone. The torque curves are what matters. A 10% loss means you will probably have to rev up another 10%, say 3300 rather than 3000, to get at the same power. While a professional reviewer stepping into a car for the first time might not notice, the guy taking his personal vehicle in sure will. VW owners will not be happy if the coming fix reduces torque by this much.

I suspect this "cheat" is done by enriching the mixture a little to lower the temp and reduce NOx. Conversely, VW might have been running things too lean under normal operation. A lean (oxygen-rich) mixture burns hotter, resulting in more power and efficiency but much of that unburnt O2 ends up combining with N2 to form the nasty N0x junk. So if the fix means effectively pinning the cheat on all the time, fixed cars will probably end up burning more fuel with less top-end power.

I know that a lean burning petrol engine will emit more NOx, but as an American who is not a truck mechanic I don't really know much about diesel. Does it work the same way in the diesel cycle? On the other hand, I am reasonably sure that more fuel = more CO2 emissions (or whatever emissions are catalyzed into CO2.)

It's the same basic chemistry. You've got O2+N2 (air) plus hydrocarbon (H-C=C-H sort of things) all mixed together and heated. Ideally all the 02 bonds with the H and C to form C02 and H20. If the mixture is lean and hot enough then there isn't enough H and C for all the O2 and it instead bonds with the N2 to form NOx.

The only real difference with the diesel is that the hydrocarbons are longer, which is why diesel is thicker and less volatile. Diesel needs to be hotter to break down these longer hydrocarbons and this is achieved via a higher compression ratio (14:1 instead of 10:1). That extra heat is an issue for N0x production if the mixture gets too lean. But if the mix is too rich, with not enough 02 for all the H and C, then you get incomplete combustion. That black soot is essentially just carbon that couldn't find any oxygen to bond with.

To avoid soot, so-called clean diesels run slightly lean and then rely on a catalytic converter to turn any N0x back into N2 and O2, but a dry cat can only go so far. BMW and others inject urea into the cat where it become ammonia (NH3). The O in the NOx grabs the H from the NH3 and the two Ns form N2. NOx + NH3 = N2 + H20.

Not very convincing.

On most cars, you can't completely turn off traction control. The "ESP off" button is there driving with tire chains, it just increases the threshold of the safety systems but doesn't disable them. That could explain the warnings in the dashboard, and also the reduced torque. High torque on low-traction surfaces leads to wheel spin.

Off topic but I hate cars that don't allow traction control (TC) to be turned off it can be dangerous.

We had Ford Transit vehicles at work and in the winter if there is a lot of snow the TC will kill 90% of engine power trapping you in the snow.

I've almost been killed by TC on a Ford Transit activating at the worst moment. When I first drove the vehicle I had the vehicle stop abruptly when the TC kicked in and a car almost ran into the back of me.

You can disable it by taking out a fuse but it also disables the ABS and airbag. Plus it was a leased company vehicle so I couldn't do that anyway.

It's like a practical joke here kill 90% of the engine power at the worst moment.

For you warm climate people drifts of snow at the end of driveways or on the road are common in winter regions. Usually you power through them if possible, wheels straight, foot down on the pedal all the way.

In the end I found out I could go fast and then put the vehicle in neutral and coast through as best I could.

I should say for ice TC is fantastic but it's not common for pure ice to be on a road, some parts but most times it would be melted by applied road salt.

ISTM the Transit's TC might just not be tuned very well? That might be related to the model year or might be something a Ford dealer could fix with a firmware update. Friends I've had in snowy places have sworn by their AWD vehicles. Personally I've scared myself half to death driving a big RWD truck in the snow.

Living in a snowy climate yes AWD and 4X4 are great yes (even FWD) it's the traction control that's the problem.

You can tell it wasn't designed by people who have ever seen snow. The Ford Transit is manufactured in Turkey but I don't know if that's a reason for the terrible traction control (and poor heat, and failing wheel bearings, and always cracking windshields).

The most telling aspect of this is not the 10% power drop but the massive torque drop at lower RPMs. That indicates that VW is limiting turbo boost at low RPMs while in this mode. That makes sense, as increasing turbo boost increases pressure, which increases combustion temps, which increases NOx production.

VW owners may be looking at the kind of retrofits California requires for older heavy trucks.[1] The muffler is replaced with a filter which has to be cleaned periodically. This seems to be working; it's rare to see a truck belching smoke in California. So there's a path to a fix available.

[1] http://www.arb.ca.gov/msprog/truckstop/truckstop.htm

I always see large trucks on I-5 regularly spewing huge amounts of smoke when they hit the gas. It gets so bad that people speed around them to avoid the smog getting into their cars. This is mostly large commercial semis and dump trucks. Are you talking about passenger cars?

The black stuff from diesels is soot, not smog. Soot is just unburnt carbon. The term 'smog' covers a whole manner of nasty chemicals, mostly oxides of nitrogen and sulfur.

Soot is relatively harmless compared to smog. A diesel producing soot is running rich and therefore is arguably cleaner than a "clean-diesel" running lean and pumping out oxides. Those carbon-belching trucks are far better for your lungs than the smoke-free VWs pumping out N0x. Beware the pollution you cannot see.

Good post! Years ago I had a slight head gasket leak, and needed to drive the vechicle just one more day before I took the head off and did the nasty job.

I had a little bit of extra smoke comming out of the tail pipe, and I got pulled over by a Police Officer. I told the officer, the extra smoke was mainly water vapor, but was still expecting a ticket.

Well, the officer said he used to be a mechanic, and knew I was likely telling the truth. He let me off with a warning.

I think those days are over though. I have seen people freak out over the wiff of cigarette smoke. Yesterday, I heard this lady yelling at a janitor over bleach smell. She squelched , "You're giving me cancer!". I don't think he understood English, and was probally better off?

I'm all for cleaning up the air, but people need to educate themselves on what exactly is going to give them, or the environment problems.

A leaky head gasket will usually produce a white smoke that smells a bit like pot smoke. That's the ethylene glycol in the antifreez/coolant being burnt.

The new anti-bleach craze is laughable. I had one woman at the local dog beach tell me "You know bleach is the same chlorine they use in swimming pools and we know that stuff causes cancer." I spend at least an hour a day swimming laps. Have done so since I was eight. And I at one point lived in a country where we regularly added small amounts of bleach to drinking water to disinfect it. If chlorine causes cancer I am doomed.

I doubt that NOx can be captured by such a filter. And according to the article particles level is already well below the limit.

Wow, only a 10% loss? That hardly seems worth cheating... then again, I don't run a car company, so I guess it's the kind of thing they think they need to do. bummer.

The real question is how it converts to mpg. A whole bunch of manufacturers have gotten in trouble for exaggerating fuel economy, and I for one certainly consider it when comparing cars (to an irrational degree, in fact). If it's the difference between 38mpg and 34mpg then that might be worth lying about.

Especially if you're trying to meet US CAFE standards. Curious if VW is going to need to buy a bunch of ZEV credits now in CA (with Tesla being the primary beneficiary).

Both CAFE and Smog Check-style programs are gamed terribly, across multiple manufacturers... but VW gets the blame, which it rightly deserves.

Testing of emissions needs to happen at the tailpipe while driving with mobile equipment in the real-world with certain types of traffic / road conditions.

Mileage needs to be calculated by metered fuel measurement on a test track with both highway- and (stop-and-go traffic) city-driving.

In addition to either two, Tesla vehicles and other full electric vehicles cannot come quickly enough to supplant gasoline/diesel/CNG/LNG/flex fuel/etc.

PS: I just California Smog Check'ed my '85 Vanagon, which has direct-port fuel injection, mass-airflow sensor and rudimentary "computer" (Digijet) after replacing the catalytic converter, O2 sensor and having a VW specialist shop tune it up. NOx (nitrous oxides) went from barely failing 15 and 25 mph limits to near zero. HC (hydrocarbons) and CO were also barely detectible. Emissions on it (a 90 hp motor) are currently at the levels of a PZEV or a new vehicle, and it gets 24 (real-world) MPG (9.8 L/100km).

Already, CA Smog Check station owners are foreseeing additional mandated equipment investments to perform tests on modern vehicles not just via the OBDII port under the steering wheel. This is a good thing but it's hard to sell small-business owners on capital and labor expenses up against the status quo/sunk costs fallacy.

So are they claiming that the software drops the engine into "cheat" mode by detecting when the front wheels were not spinning (which implies dyno-testing)?

Is it confirmed that is how cheat mode worked?

If I were writing this I'd be more inclined to look for known test cycle behaviour (ie, accelerate for a known time, then constant speed then stop, start etc etc) and drop into cheat mode when it is detected. I guess this could be a problem for unknown cycles though?

Other groups have stated the same thing [1]. The car already has to detect dyno testing to turn off traction control and other things that are dangerous under these conditions, adding a flag to tweak the mixture the car works with under speed isn't so difficult after that. Trying to detect a pattern would lead to much more complicated logic, especially if you get a tester who decides to do things in a different order.

[1] http://www.consumerreports.org/cro/cars/volkswagen-emissions...

>if you look lower on the RPM curve the difference is significant. The car lost 15 horsepower around 2,800 RPM, and 32 lb-ft of torque near 2,700 rpm

so, looking at the picture, the power/torque at 2800rpm is about 120/260 while in "cheater" mode it is 105/228. And that ratio is preserved through whole 2000-3000rpm range, the range that really matters as most of the driving happens here.

So now, given that "fixing" the problem will cause power loss, and consequently adversely affect fuel economy, wouldn't a number of VW owners opt not to bring in their vehicles for the recall? What's the incentive for VW owners to voluntarily bring in their cars for a regressive (in terms of power and mpg) fix?

Also given that this effects the quality of the product they bought significantly, wouldnt they be entitled to a refund?

The incentive is that, in states that have mandatory smog checking, you will not be able to renew the registration on your car until you get it fixed, rendering it illegal to drive

Unless you find one of the many, many shonky smog-check places, which surely can't be that hard.

Except that it might be hard to get shonky with a VW and keep it under the radar.

I believe VW have promised to replace everyones entire engines. I'm stupefied over how much this will cost, i've seen numbers around to 100B mark. Their market cap is somewhere near half of that. Presumably they think it's better than the brand damage otherwise. I think it's fair to assume they'll go bankrupt over this.

I seriously doubt Germany would let VW go bankrupt, the government owns twenty percent of the company and they are too big too fail.

Do you have a link for this? Replacing even 500k engines in the US is indeed stupefying.

The BBC did a real-world driving test and directly examined the exhaust for NOx emmisions and they were exponentially worse than the lab results.

If the exhaust was going into your car, you'd be fatally poisoined in short order.

So thanks for the cancer you've given to all the pedestrians and cyclists VW (and all the other lying corporate a-holes).

Not all toxins are carcinogenic. In particular, VW seems to have been leaning out the mixture, resulting in lower levels of soot and carcinogenic polyaromatics. I'm not aware of any studies showing oxides of nitrogen to be carcinogenic.

Why would a company take such a huge risk for a paltry 10% gain in engine power, and a probably even smaller gain in mpg? This seems like a very dumb thing to do.

In order to sell cars. The competition is/was doing it also and it was widely known practice. [1] For some reason VW is getting all the blame.

[1] https://www.youtube.com/watch?v=pjUr3RQRERM

Offtopic but I couldn't help it... It looks like they're still running win xp! I suppose the pc is air gapped yeah..

Reply on news.ycombinator.com