Most gearheads would agree that a vehicle’s exhaust sound gives it its soul. People can identify engines sight unseen, just by the distinctive song emanating from their tailpipes. This love for the sound of internal combustion is the reason why aftermarket exhausts are the most popular modifications for a car or truck. Whether it’s the sonorous wail of a V12, the intimidating rumble of a V8, or the blatty warble of a flat four or six, a properly tuned exhaust is literally music to an enthusiast’s ears. But for better or worse, we live in a “society” with so-called “normal” folks who don’t like the songs of our people.
While most of us can fall asleep to racing videos; a straight piped muscle car roaring down the street at two in the morning makes regular people complain and write letters to their alderman or whoever. Because of this, there are laws and regulations and international ruling bodies who throw their weight around and force automakers to quiet it down or they lose their ability to sell cars.
Joking aside, this all makes sense. Life would be much more annoying if garbage truck makers decided it was cheaper to build their vehicles with open headers, for example; or if you lived anywhere near civilization and every vehicle on the road was as loud as a fart-canned Civic. That’s why countries all over the world have had vehicle noise limits on the books for the past of 50+ years.
Automakers have to pass what is called a “Pass-by test” with every engine/transmission combination for each vehicle model they sell. Thankfully, inside the 100+ tedious pages of these international test standards written by the most lawyerly engineers, or engineery lawyers, is a tiny clause that lets our beloved vehicles belt out their spine tingling melodies whenever we hit the loud pedal. It gives powerful vehicles a significant advantage compared to your average car or truck. Without this three word stipulation, the world would be a much sadder place for anyone with gasoline in their veins.
To be clear, this test has nothing to do with how awesome a car sounds. There is a whole art and science that goes into designing an exhaust that does its job of quieting the combustion events happening in your engine, while invoking the subconscious, subjective feelings the company wants you to associate with their vehicle. This isn’t a group of sommeliers debating the subtle notes of a new batch of bourbon. It’s the scientist confirming the alcohol level is within the legal limits.
I’m Steve Balistreri, the engi-nerd from Detroit who specializes in NVH (noise, vibration, and harshness). Pass-by testing is something I’m very familiar with, as I’ve run hundreds of tests for a half dozen automakers at tracks across the country. I was in charge of all the pass-by testing at a Michigan automotive proving ground for five years and tested anything from electric cars to 700 horsepower trucks to boring crossovers to super cars that cost twice as much as my house. Some tests were quick and dirty, done on barely running prototypes held together with duct tape and hope; others were official certification tests where every detail was poured over by an independent third party witness. Pass-by testing involves repeated wide open throttle events which is fun, but the targets are very tightly controlled and take a bit of finesse and skill to hit.
How A Pass-By Test Works
Before I spill the beans on this secret loophole I have to explain how a pass-by test works. The whole purpose of the test is to quantify, in a controlled and repeatable manner, how loud a vehicle is during normal driving in urban traffic. The first pass-by test specification came out in 1967 with J986. Times were simpler back then, cars had 2 or 3 speed transmissions, and the spec filled a whole 8 pages. It was revised in 1984 with J1470 or ECE324 which ballooned to 17+ pages depending on the revision. The current spec, UN/ECE R51.03 introduced in 2016 is a whopping 98 pages long. It’s kind of confusing as there are international ISO specs, their equivalent SAE specs, specs for different countries, plus all their amendments. Overall, the current specification is run the same way everywhere.
Let’s start with the track you do the test on. Since tire noise is a major contributor to a car’s overall noise level, the track has to be built to an ISO specification and certified every two years. It has to be perfectly flat and meet targets for texture, surface irregularity, and noise absorption. An area 50 meters around the track needs to be free of any objects that may reflect sound back to the test microphones.
It helps if the track is in a quiet place as there are ambient (background) noise requirements to run the test. Ideally in NVH testing you want the background noise to be at least 10-15 dB (decibels) quieter than the thing you are measuring. I’ve spent many hours sitting on the test track waiting for hundreds of migrating geese to fly over us to the nearby pond because their stupid honking was too loud. The worst case was a proving grounds I tested at where the pass-by track emptied onto an active skid pad area where a driver training/autocross event was happening. If that wasn’t bad enough, on our left side was the giant oval track, where the curve hits the straight, a popular place for the dozens of test cars out that day to accelerate hard out of the corner. If THAT wasn’t bad enough, the right side of the track had water cannons spraying a tile surface where they were doing emergency brake testing on school buses. Needless to say, it was way too loud to get any testing done. We later found out most people rent this particular pass-by track at night.
Due to the massive expense required to build and certify these tracks, there are only a handful in the US. The lack of grading on the pavement also means these tracks take forever to dry if it rains since there’s nowhere for the water to go. There is a track on the east coast where it would rain every time we made a trip out there. It got to the point where we’d pack giant squeegees in the test vehicles so we could push standing water off the track and not lose too much test time.
Marked out on the track is a line down the center of the lane you drive down, called CC’ in the spec. The PP’ line is in the middle of the track perpendicular to CC’ and is where the two measurement microphones are located, 7.5m from CC’ and 1.2m above the ground. The portion of the track where the measurement actually happens is marked off by lines AA’ and BB’ which are 10m on either side of the PP’ line. The peak noise level from the time your front bumper touches the AA’ line to when your rear crosses the BB’ line is used to calculate the score for that test run. You can see this layout in the diagram below.
The pass by test requires four steady speed cruise runs where you drive 50 kph (31.1 mph) +/- 1 kph through the track; and four wide open throttle runs where you drive down the track and WOT the throttle before AA’ and hold it until your rear bumper passes BB’. For the WOT test, your vehicle speed must be 50 kph +/- 1 kph when your front bumper hits the PP’ line.
The spec requires linear acceleration between AA’ and BB’ so pre-acceleration is encouraged, ie. mashing the accelerator before your front bumper hits AA’. This way any delay between your WOT and the vehicle fully accelerating from turbo lag or whatever happens before AA’ and your velocity line is nice and straight. You can see some test data below, the WOT point is around -14 meters (where the RPM line gets wavy), and the curve in the vehicle speed line where the car starts accelerating happens before AA’ (-10m).
Hitting the PP’ speed target takes a bit of skill and throttle control. You have to pick a spot a couple meters before AA’ and WOT it at that point consistently. You also have to control your entry speed precisely because the speed target 10+ meters from your WOT point has a window about 1 mph wide. At the same time, you have to keep the car centered on the CC’ line and operate a laptop to control the test software, and occasionally a second laptop to control the automatic transmission. The peak noise level from these four runs at each microphone have to be within 2 decibels of each other, which adds more chances for error. These WOT runs mess up newbies the most. It’s definitely a great feeling when you can crack off four perfect WOTs in a row.
The cruise runs are nothing to sneeze at either. Your speed can’t exceed 51 kph or dip below 49 kph at any point during those 20+ meters you’re driving through AA’ and BB’. With some electric vehicles, regen can be touchy so you really need to feather the throttle in that Goldilocks zone to keep your speed steady.
Here is what the WOT and cruise runs look like in our software. For all the graphs, the X axis is the position of the vehicle’s front bumper on the track in meters. -10m is AA’, +10m is BB’ and 0 is PP’ where the microphones are located. The top-left graph is vehicle speed, bottom left is engine speed. The two right graphs are the noise level from the microphones. Each graph is displaying the four required runs, and you can see the consistency required to do the test correctly. For the WOT runs, the entry speed is about 46 kph and the WOT starts around -19m so the vehicle speed line is nice and linear between AA’ and BB’. All the runs are within 1 kph at PP’. The peak noise level between AA’ and BB’ + the vehicle length is what is used to calculate the final result.
At the top of this section is a little video of a pass-by test being done with our equipment.
The software measures the peak noise level as you WOT or cruise through the test zone and records it. It does a bunch of math with those levels and you get your score. The current limit will be 68 dB in the US in 2024, which is 6 dB lower than it was in 2016. For reference, that is about the noise level of a vacuum cleaner.
The way the software works is pretty cool as well. There are two separate data acquisition systems, one on the track and one in the vehicle. The track, or “ground station”, measures data from the microphones, meteorological data from a weather station, and two IR triggers that send a voltage spike when the vehicle crosses a beam at the entrance and exit of the track. This data acquisition system is connected to a network switch that sends data to the ground station laptop and a large antenna.
The “vehicle station” measures the vehicle speed via GPS, the engine RPM, and any other extra microphones the engineer is interested in getting data from, such as tire, exhaust, or intake mics. This station is also connected to a test laptop and a smaller antenna perched on the roof.
Both stations have their own GPS antenna, not to measure position, but to record the super accurate time signal from the GPS satellites. The ground station is sending live microphone and weather data to the vehicle station through the antennas and knows when the car passed through the track via the IR triggers. After a pass, the ground station sends all this data, along with the GPS time data, to the vehicle station. The vehicle station lines up the time signals to overlay the data, and back calculate the vehicle’s position on the track, since it knows its speed and when it crossed the known position of the IR sensors. This all happens in a few seconds so you get almost instant feedback whether a test run was good or not.
The Three-Word Loophole
So where does this loophole come in? If you are measuring wide open throttle events, how could a Corvette or Mustang ever be on the same level as your average crossover? The key is the gear you test in. Every vehicle gets an acceleration target, or Awot ref, which is calculated using the vehicle’s power and weight via the equation below. You do pre-test WOT runs in multiple gears to calculate their acceleration and see which one is closest to this target. This is why you use pre-acceleration, so the vehicle speed line between AA’ and BB’ is as linear as possible to get an accurate number. If no gear is within +/- 5% of the target you test in the gears above and below the target.
A wot ref = 1.59 * log10(PMR) – 1.41
PMR = Power to Mass Ratio
For example: if your A wot ref target is 1.68 m/s2 while your 4th gear acceleration is 1.9 and 5th gear acceleration is 1.52, you’d do 4 WOTs and 4 cruise runs in both 4th and 5th gear, since neither is within 5% of the target. If your 5th gear acceleration is 1.6 m/s2, you’d do the 4 WOT and cruise runs in 5th gear because it’s within 5% of the target.
What about electric cars, or other cars without gears like CVTs? We use a hand operated accelerator pedal that plugs into the vehicle’s wiring harness. There is a set screw that lets you limit how much you can push down the pedal, i.e. limiting the throttle percentage. You do practice runs where you dial in this pedal until you are within 5% of the target.
This acceleration target is where the loophole comes in, see the bold portion of paragraph (a) from section 3.1.2.1.4.1 of the specification.
If one specific gear ratio gives an acceleration in a tolerance band of ±5 per cent of the reference acceleration awot ref, not exceeding 2.0 m/s2, test with that gear ratio.
There is a 2.0 m/s2 cap to the acceleration rate! [Ed Note: Re the headline, we’re calling the three words “not exceeding two.” -DT] If you don’t understand how this sentence is a gift from those lawyers holed up in the mountain lair where these standards come from, read on. Your typical car with over 300 hp will have an acceleration target right around 2.0. For example a VW GTI with 220 hp has an acceleration target of 1.86, while the 310 hp Golf R’s target is 2.14. When you get into the 400+ horsepower range it’s well over 2. A new Mustang GT’s target will be around 2.17, a BMW M5 is around 2.33.
According to the specifications, in that situation you choose the gear with an acceleration as close to 2.0 m/s2 as possible without exceeding it. So instead of doing your WOT and cruise runs in third gear where you are really winding out your engine, you do it in 5th or 6th because the acceleration rate will be under 2.0 m/s2.
If you’ve driven a manual and mashed the pedal in too high a gear you know what this is like. Your entry speed for your WOTs is in the low 40 kph zone (25 mph), in 6th gear you are just barely above idle. You floor the accelerator, the engine kind of groans and you exit the track at BB’ a few hundred RPM higher than you were when you entered.
In a more typical vehicle with an acceleration rate under 2, the runs are more like the ones in the plots above, where your RPMs reach 3k or sometimes higher. Engines typically get louder the higher they go in the rev range so this acceleration cap is basically limiting how loud these vehicles can get during a test.
To be clear, the engineers still have a lot of work to do to keep these cars below the limit even in this situation where they’re in a high gear. But doing the test in 6th gear where the engine never cracks 2000 RPM is significantly quieter than if it was racing towards redline.
It makes sense why they would do this. This standard is meant to measure a vehicle’s sound during its “typical” operation in an urban area. To get around town doing your daily activities you don’t “need” to accelerate more than 2 m/s2. If you are driving a low horsepower car you have to wring the engine out a lot more to keep up with traffic compared to a high horsepower car that can do it in its sleep. The standard is pretty much assuming you are going to drive your new sports car like a grandma. Even though your sports car can accelerate much faster and be a lot louder than the test measures, they are trusting that you’ll be a smart and courteous member of society. This type of trust seems to be severely lacking these days, but is definitely welcome.
They could have easily left this clause out, and it would’ve made the test a lot simpler. However, if this acceleration cap didn’t exist, our automotive landscape would be a lot different. It would take a ton of work to get a V8 performance car in the 70 dB range doing a full WOT in 2nd or 3rd gear. The exhaust would probably be one long muffler, and you’d lose all the character and joy that sound brings. Identifying vehicles by their exhaust sound would get significantly harder, driving them would be much less exciting. It’s not a world I want to live in. So the next time the wail from a car’s exhaust puts a smile on your face or a tingle down your spine, thank the engineer/lawyers up in their mountain lair who added those three critical words, making it all possible.
Is this same system used for motorcycles? If so, this explains why the tiny Kawasaki ZX-4R (70ish hp 400cc) got neutered by regulations above 10k rpms, when a wildly more powerful ZX-10R (200+ hp 998cc) is allowed to breathe deep.
Motorcycles use a different specification. I’m not familiar with that one unfortunately.
Thank you! I’ll have to do my own research, I suppose. *sad trombone*
Is this system also used for pickup trucks? Because when I’m walking the dog, pickup trucks are way louder than the Camaros, Chargers, and Civic Type Rs.
Yes it definitely includes pickup trucks. After the car leaves the factory people can do what they like, and pickups are definitely popular to modify.
Exhaust work must be mandatory with all lift kits and wheel spacers.
Would this be TRC?
Yessir, we never rented that passby track again. We’ve done a lot of other testing there.
I do ABS/ESC calibration, so I’ve spent quite a bit of time on those tiles. The water cannons are awesome (for us)
I mean, that’s great, and all, but are going to just let slide that bourbon has stewards, not sommeliers?
Sommeliers are for wines, good sir
*pinky out*
I did not know that! Learn something new everyday!
I have some experience of engineering cars to get through pass-by testing. One of the vehicle attributes we get given to work to is making it sound nice, and for a “sporty” car quite often the target is set to be nice but also loud, so you can have a well insulated car but still hear the exhaust. So we’ll do a dozen designs, test them for sound quality and volume, then some durability, then send them off for a pass by test to find they score 1dB louder than our test, and the whole thing goes in the bin with hundreds of thousands of dollars/pounds.
It’s much, much easier to design for an easy pass at the test, with a dB or two of safety margin.
Or cheat, cheating is easy too. Of course since the whole VW thing no one risks cheating an emissions test any more, plus it’s much harder to do with the current testing.
Cheats that have been previously successfully:
* Calibration that will limit the throttle opening if pre-pass-by test driving is detected, ie very steady 50kph followed by a WOT demand. If your test driver mucks it up you’ll get a obvious much louder, much faster test run. There are production cars out there that do this.
* special parts marked up as production parts (very hard to do this without any of the dozens of people involved blabbing, certainly I’d be a whistleblower on this, and have vetoed the suggestion in the past).
* polishing cloth accidentally left in the airbox before testing (intake noise can be pretty damn loud too).
* special calibration/engine to allow higher performance, and therefore higher gear for testing.
* restricted exhaust “engine protection valve” operation when it detects test conditions. Those valves can cause cause catastrophic engine failures if they don’t open properly, so don’t mess with them.
* special loud modes on active exhausts. This was legalish for a while, but not any more.
All of those have been done in the past, but I’m pretty sure no one is cheating the current test. It’s much harder to cheat with all the data collected now, and senior management can go to jail, so they would much rather we passed legally, especially as they aren’t the ones who have to do any actual work.
Also it’s now much, much easier to make the actual car quiet and pipe fake or “enhanced” noise in to the cabin. My GT86 and E86 (I don’t care if you haven’t learnt your BMW chassis codes) both have intake noise plumbed in to the cabin through pipes, and I don’t mind at all. My ears hear engine, my neighbours don’t get woken up, everybody wins. Some people complain about artificial noise, but those people probably think that what they used to hear was somehow “natural” instead of very carefully tuned and enhanced sound waves engineered to a target.
That’s super interesting! I always wondered if there were people trying to cheat the system. Did you have ASEP tests back then? You basically have to do runs in multiple gears and multiple speeds and it checks for linear acceleration and sound increases. Plus we have to do COP (certification of production) testing where we test production vehicles and confirm they match. I think it’d be a lot harder today.
I agree with you on the “enhanced” noise, especially on performance cars that share platforms with regular cars like a GTI or M3. It would take a ton of work to redo the whole sound package and tune it to let the “good” sound in for the sporty car.
The ASEP tests have made, erm, let’s call it “gaming the system” pretty much impossible to do.
The last meeting I had on this was a couple of years ago and there wasn’t even a discussion about trying to do anything clever, no one is homologating anything noisy, we all benchmark each other, so we’d know.
Very interesting breakdown, thanks. I want to call attention to one line specifically for people who complain about the new strict noise ordinances in places like NYC:
“An area 50 meters around the track needs to be free of any objects that may reflect sound back to the test microphones.”
50m is about 150ft. Streets in old cities like NYC are 50-60 feet wide, and surrounded on all sides by concrete, stone, metal, and glass that reflect sounds, so all those cars revving engines might sound totally fine in wide-open, rural Pennsylvania, they’re perceived as much, much louder in the literal narrow concrete canyons. It’s not just that people are necessarily more sensitive, but that those already-loud sounds are bouncing and amplifying more in a city than a suburb or countryside
Too many people think loud = good. It’s not. You don’t have to make a lot of noise to compensate for your shortcomings. You don’t have to like loud exhausts to be an enthusiast.
For a laugh, look up “Laguna Seca pipes” or similar to see the monstrosities some people attach to their cars to not trip the infamously low volume (thanks NIMBYs) sound check station at that track.
If I remember correctly, the original Tesla Model S failed this test until the tire compound and tread pattern was changed. The tire noise originally was too noisy.
A lot of modern sports cars have used long gearing to get past this test as well.
That’s not surprising, especially for vehicles with sticky tires. We had a couple cars that would fail with brand new tires then pass after breaking them in for an hour or so.
Holy cow the vision this brought to my mind.
Mr Muffler rubbing hands & chuckling evilly?
As someone who lives in a highrise where the building across the street is extremely accurate about 7:00AM garbage runs every day of the week, I am so happy about this.
EV garbage trucks will be a game changer.
It’s the bloody reversing beepers that wake me up. I live on a T-junction on a narrow street, so every week there’s at least two sets of beeping as they shuffle the bin lorry (‘garbage truck’) back and forth to get it around the corner. My bedroom window is only a couple of meters form the road, so in the summer the beeper might as well be next to my bed.
Then half an hour later the recycling lorry will turn up, and wake me up all over again 🙁
I feel your pain.