There is no more effective way to deceive than by using the truth. Simply delivering the truth in a carefully twisted manner is a long-honored tradition of politicians and advertising executives alike. The performance aftermarket is seldom considered a bastion of honesty and decency, but the proliferation of cheap-to-operate Dynojets has done a lot to clean up what could mildly be termed "optimistic" horsepower claims: Claims that once were the rule rather than the exception.
They have also made higher honest power gains more common, as developing effective parts is cheaper and easier when quick, repeatable dyno runs are possible. As usual, though, the news is not all rosy. Though I often refer to the Dynojet as our standard unbiased provider of truth, it is still quite possible to make them lie. You don't necessarily have to be devious to achieve this deception either, (though it does help) simply being sloppy can be enough.
Since most people don't have dynos in their driveways, we have to rely on others, usually on those who are selling us something, to test our cars. Sometimes these dyno operators are sloppy, sometimes wishfully optimistic, and occasionally downright dirty. To give you a chance of spotting the unrealistic dyno results, here are the easiest ways to lie with a Dynojet:
Other modifications (Oh, that car was turbocharged?)
This is perhaps the oldest trick in the book, and a favorite for the liar with a conscience. Say you are quoting power output on an exhaust system. It is natural to expect that someone buying an exhaust would also have some other performance parts, so it is not too far fetched to test a slightly modified car. So you stick a stock exhaust on that modified car, test it, and then put your exhaust on and test it again. This is still an honest assessment of what the exhaust did, but if that car happens to be a Civic DX, and the other modifications happen to be a 15-psi turbo system, the 30-hp gain now claimed for the exhaust will be more than a little unrealistic.
Hot vs. Cold
Most cars make more power on their first Dyno pull while everything is still nice and cool. If you allow a reasonable cooldown period between runs (only a minute or two is needed on most cars) all the subsequent runs should be consistent with each other. The natural inclination of anybody trying to show a power gain, though, is to use a lower, hot run as the baseline, and a nice, strong cold run to showcase their part. I have seen people do this with no malicious intent, just elation that their part made so much power. "As soon as I put the power pulse super-mega-flagenator on the car, it made 17 hp on the first pull!" Never mind that it made only 13 hp on every subsequent pull.
Because there are so many different levels of cold runs, the only realistic data, unfortunately, is to compare consistent hot runs. Whenever possible, we actually monitor coolant temperature and make each run within a five- to ten-degree window. Our Nissan Sentra SE-R, for example, is always tested between 185 and 195 degrees.
Different Gears
Back in the May, 1998 Technobabble, I dyno tested a Sentra SE-R in every different gear. Second, third and fourth were very similar, but first and fifth were quite a bit lower. Testing a car in either first or fifth is a little odd, and would probably be noticed by the car owner if they were observing the test. The differences are much more significant on a turbocharged car, though, as the higher gears slow the engine's acceleration and allow the turbo to spool up earlier and stronger. If you do a second gear pull with a turbocharged car, the engine will often outrun the turbo--by the time the turbo is spinning fast enough to make boost at 3000 rpm, for example, the engine will already be going 4000 rpm; by the time the turbo is fast enough for 4000, the engine is at 5000.
This is why turbocharged cars feel so strong in higher gears when pulling up a hill, or whenever they are heavily loaded. If you compare a turbocharged pull made in second gear to a pull made in fourth, the fourth gear run will almost always be stronger. We typically dyno naturally aspirated cars in third gear, and turbocharged cars in fourth.
Another difference that can occur with gearing comes when there is a 1:1 gear ratio available. Most rear-wheel drive gearboxes will have a gear that is 1:1--typically this is fourth gear, but in six-speed gearboxes it is often fifth. In the 1:1 gear in many gearboxes, power is not actually transferred through a gearset; the input and output shafts are just locked together. This eliminates the losses through the gear teeth, and results in a higher power output in that gear.
Random Variation
There is always a small variation from run to run. Some of these variations, like the difference between a hot engine and a cold engine, can be predicted and controlled. Others, like the temperature of the transmission, the alignment of the wheels on the roller, and the slight difference between the actual intake temperature, and the dyno's correction temperature as measured elsewhere in the dyno room, are harder to control. Cars that tune themselves while you are testing can make this problem even worse. Run-to-run variations of less than 1 percent are normal, but if you take the lowest baseline run and compare it to the highest test run, you can easily make something look better than it is. The best way to avoid being snookered by this one is not to split hairs. A 1 or 2 hp difference is not really significant enough to feel, even if it is real and repeatable, so why worry about it?
Corrected or Uncorrected
Dynojets can read corrected or uncorrected horsepower. Because different weather conditions can result in different air densities and different oxygen concentrations, the weather can have a significant effect on power output. The SAE has a standard set of correction factors that can be used to normalize all power outputs to what they would be at sea level, on a 60 degree day, with 0 percent humidity. Every Dynojet has a small weather station built in to feed the appropriate temperature and barometric pressure readings to the computer so it can calculate this factor. The difference between 0 percent and 100 percent humidity is about a seven percent correction. A temperature change from 60 to 90 degrees, on the other hand, will have an effect of about a 2.8 percent. A difference in elevation from sea level to 5000 feet is worth a whopping 20 percent!
If you use uncorrected data, the changes in power output due to weather conditions could be misconstrued as being a result of something else. For example, if you have a dyno sheet showing that a header made a 15-hp gain, but the runs were done on different days and the plot was made with uncorrected data, you can't be sure if that is really a 15-hp header, or an 8-hp header and a 7-hp weather change.
Corrected numbers, on the other hand, can be suspect in certain cases as well. Turbocharged cars running at high altitude, for example, might be more accurately represented by uncorrected numbers. Say you are testing an FD3S RX-7 in Denver, where the elevation is approximately 5,000 feet. Shiv Pathak, master of our FD3S RX-7 project, reports that he always sees higher boost levels at high altitude. The reason is simple. The wastegate opens when boost is 12 psi higher than the normal sea-level reference air behind the wastegate actuator diaphragm (air that has been stuck in there ever since the diaphragm was sealed somewhere in Hiroshima). As the air density drops at high altitude, the actual pressure in the boosted intake manifold remains constant. The boost gauge, though, reads pounds of boost over ambient pressure. If the ambient air pressure in Hiroshima was 14.5 psi when that diaphragm was sealed, but it is only 13.5 psi when Shiv drives through the mountains, his boost gauge will read 1 psi higher than normal.
The SAE correction factor used by Dynojet assumes that lower air pressure at the sensor box means lower air pressure in the intake manifold, though, so at 5,000 feet the dyno is applying a 20-percent correction factor to compensate for a loss of air density that the engine never sees. This is fine if you are doing all your tests in Denver, but if you do one test in Denver and one test in New Orleans (the highest mountain in New Orleans is 12 feet above sea level) uncorrected numbers will be more accurate.
The SAE correction factors are only accurate over a relatively limited range, and the Dynojet software is smart enough to warn you when two runs with wildly different correction factors are being compared. The software in New Orleans can't check your glovebox for that last dyno printout from Denver, though, so you'll have to warn yourself.
Messing With the Weather
This is where the dyno operator has to be extremely devious. If you mess with the readings that the computer uses to calculate the correction factor, you can alter the corrected output significantly. The one reading that the built-in sensors do not take automatically is humidity. The dyno operator has to enter the humidity correction themselves. Since the humidity is manually entered into the computer it is the easiest to alter. While humidity numbers are obviously suspect, the temperature can be faked pretty easily as well. We actually decided to try this one, just to see how easily it could be done.
Jackson Racing's Dynojet is set up with the weather data box mounted on a perpetually-shady portion of the dyno room wall. The temperature probe hangs under the box in the open air about two feet off the ground--right where most air intakes pull their air supply. Oscar Jackson pointed out that he has seen these boxes mounted where they were more easily accessed, and has even seen the temperature sensor hanging on a divider wall next to the computer, or in a drawer on the dyno bench. In the drawer, an unscrupulous dyno operator could put his or her hand around the sensor before doing a run, bumping the ambient temperature reading up into the 90-degree range. With it hanging on a well-placed wall, the sensor could be flipped from the shade into direct sunlight, where it could slowly bake up to a nice, warm temperature.
The wire on Jackson's sensor was only about two feet long, so we couldn't get it into the sun, and a shield prevented holding the sensor in a warm hand from having much effect. Instead, I cupped my hands around the sensor and blew on it. Within 30 seconds the dyno was reading 95 degree ambient temperatures even though our baseline run made a few minutes earlier had been in 66-degree air. We made another run with the engine breathing 66-degree air, but the dyno correcting for 95-degree air. Our corrected power jumped from 136 hp to 143 hp.
Air Conditioning
This seems almost too easy to mention, but the fact is that it is one of the easiest and most effective ways to make a dyno lie. Just run the baseline pulls with the air conditioning on. The only ways you can tell if the air conditioning is on from the outside of the car is to look at the clutch on the air conditioning compressor, look at the fans on the radiator (most cars will kick on a cooling fan whenever the air conditioning is on) or look at the face of the dyno operator. No sweat? Be suspicious.
Since the Sentra SE-R we were testing automatically disengages the air conditioning compressor at full throttle, we actually had to adjust the throttle position sensor so that the ECU never noticed it was at full throttle. This alone has some effect on power, but based on subsequent runs without the A/C, the difference was only 1 or 2 hp. The difference caused by the compressor, on the other hand, was10 hp.
Since many cars do disengage the compressor at full throttle, this trick does not work on every car. Jackson pointed out, however, that although the Integra GS-R is supposed to be one of those cars, he has seen the compressor run at full throttle on multiple occasions.
So what can you do so you don't fall for fake dyno figures? There is no foolproof way, short of dynoing everything yourself (and not making any of these easy mistakes on your own), but a critical eye is always a good idea. Look at the printout from the Dyno. Dynojets always print the weather conditions and show the correction factor (labeled as CF). Correction factors between 0.97 and 1.03 are pretty normal. Outside that range, you should be on the lookout for large differences between the runs you are comparing. You should also have some sort of idea what to expect from each modification. If you tweak your adjustable cam sprockets, don't expect 15 hp across the board--4 or 5 hp over part of the powerband is more realistic. How are you supposed to know what to expect on every single part? Well, I can think of a magazine you might want to read...