Separating fact from fiction when it comes to dynamometer testing is an endeavor fraught with peril, particularly for the end user. Outside variables from the testing environment, standards of calculation, and other less-than-predictable factors can alter test data substantially from one test to the next, and that often makes using the efficacy of the data highly suspect.

In fact, even using horsepower as a unit of measurement has some caveats.

“Horsepower cannot be measured – it must be calculated,” explains Bret Williamson of SuperFlow. “You can measure torque, revolutions per minute, and acceleration, but you cannot measure horsepower. It is always derived by measuring something else.”

In a nutshell, horsepower is the ability to accomplish a specific amount of work in a given amount of time, like sending a car down the drag strip at a specific elapsed time. It’s a means of measurement that is at the core of dyno design.

Dyno testing can be a helpful tool, particularly when tuning. But when it comes to output, the only way to get a truly accurate measurement is on an engine dynamometer.

Here we’re looking at three different types of dynamometers with some insight from the experts at SuperFlow, Dynojet, and EFI University to get a better sense of how each measures power, how those designs affect the resulting data, and which scenarios best suit each dyno design to provide data that is as useful and accurate as possible.

Engine Dyno

While an engine dyno is the most effective way to get an accurate impression of an engine’s output because the measurements are coming directly from the engine itself, there are still a few factors worth considering.

“An engine dyno is not free from any of the same issues you’d have with any other type of design,” says Ben Strader of EFI University. “It still has mass and inertias that have to be dealt with. But the big deal with the engine dyno is that you eliminate some of the variables.”

Those variables largely relate to the ones introduced by the vehicle itself on chassis dynos—drivetrain losses through the transmission and differential, as well as losses through the tire and roller interface. Engine dynos have a few variables of their own though, like whether a particular engine dyno uses a water brake, an eddy current brake, an AC motor, and so on.

“All of those still have to deal with the inertia that’s developed when you accelerate or decelerate,” Strader explains.

“The inertial part of the equation only really happens during the acceleration phase. If your speed is constant, there is no inertia. So dyno manufacturers still have to understand what the amount of inertia in their system is and calculate for that. But the engine itself also has inertia, and most people tend to forget about that. So the dyno can have the most accurate calculation in the world for controlling its own inertia, but if you haven’t also calculated for the inertia of the engine itself, you still run into some of those issues.”

Fortunately the margin of error inherent to engine dynos, and its relation to engine inertia compensation, is still significantly smaller than those introduced in roller dyno designs.

“It’s definitely not something that most guys need to get carried away about,” Strader adds. “You’re stepping over dollars to pick up dimes in terms of the procedure complexity required to measure that, and the reality is that the industry standard is pretty well set in terms of how to do a typical engine dyno sweep.”

Chassis Dyno (Inertia)

While chassis dynos are convenient means of measuring output due to the fact that they’re designed to be used with the engine installed and fully functional in the vehicle, the fact that the data comes from the work exerted on the roller(s) from the wheels rather than from the engine itself introduces a number of issues in terms of measurement accuracy.

An inertia chassis dyno’s greatest virtue is its repeatability, which makes it very effective for comparing component combinations.

“Chassis dynos don’t measure engine output,” notes Mike Giles of SuperFlow. “They measure output at the contact patch between the dyno and the dyno roller. Those are significantly different things, and the latter is influenced by a great deal of factors not found in engine dyno testing.”

Many of those factors are focused not only on calculating drivetrain loss, but in the lack of procedural standards, which can have a significant effect on the results.

“I would argue that there are some manufacturers who make very accurate chassis dynos,” says Giles. “Unfortunately a significant portion of the operator base—and worse, their customers—simply don’t understand how they work or the data they are providing.”

With a roller dyno, the system creates its output data by using the equation that force is equivalent to mass multiplied by acceleration (Newton’s Law). The mass in this equation is the amount of inertia in the roller, so the data is essentially generated by the amount of time it takes a vehicle to accelerate that roller.

“The advantage is that there’s very little to go wrong with the dyno in terms of change over time and calibration,” says Strader. “That makes it very repeatable for back-to-back testing, so for shops that do work on specific models and do comparisons between parts, the before and after results can be very reliable.”

That repeatability can save time for specialized tuning shops.

“The key benefit of this design is that it’s easier and faster to operate with minimal initial set up and no calibration necessary prior to making a dyno run,” says Dynojet’s Will Fong. “It also provides more accurate and consistent data in the dyno graph, due to direct sampling from drum revolutions versus using a strain gauge such as on eddy current or another load system.”

The downside here is that while the data may be accurate for comparison purposes, that’s not necessarily the case in other circumstances.

“If it’s an inertia-only model (no eddy-current added) it can’t perform load tests such as steady state, step, or wind drag simulation,” Fong points out. “These are typically better for shops who simply want to validate power readings with a chassis dyno and perform minimal tuning. We see this with shops who use proven tunes and parts combos that provide consistent results across similar vehicles. They normally have to change very little in the calibration.”

The design’s limitation is inherently tied to how it gathers its data. “We’re not measuring the torque and horsepower at the engine here, we’re measuring it at the tire,” says Strader. “Then we’re making some assumptions and backwards calculating to what it would have been based on the engine RPM, not the tire RPM.”

That means the dyno has to look at the amount of torque applied to the roller and compare engine RPM versus tire RPM, consider factors like gear ratios, and make an assumption on available engine power based on those factors.

“The problem is, if I do a test on the dyno, and then I drain the 90w gear oil out of the differential and put in, let’s say 30w gear oil, it may not be good for the life span of the gear but it will likely create less drag, and that will make more torque and power available at the tire even though I didn’t change anything at the engine,” says Strader. “This is an issue related to any chassis dyno. The reality is that we are ultimately measuring left over engine power, after all of the losses.”

While that higher reading is accurate in that there is in fact more power available at the tire, since nothing was changed about the engine configuration, it shows the inherent accuracy drawbacks of a chassis dyno when tasked with measuring engine output.

Strader also points out that while many companies tout the accuracy of their chassis dynos based on their ability to predict drivetrain losses, outside variables can easily affect the data from one pull to another, regardless.

“The problem is that if I change tire pressure, or I change strap tension, or the temperature of the transmission fluid changes—any of those things are variables that are affecting the amount of total thrust available at the tire to roller interface that have nothing to do with engine performance.”

This in turn creates an issue in tuning, where the tuning interface is connected to the engine directly. “When I make a change on the laptop, I’m never really sure if the add or subtract value that I ended up with at the roller was because of what I did to the engine, or because of some other variable in the drivetrain that I didn’t account for—like doing three pulls back to back—which means the transmission and differential temperatures are higher.”

Chassis Dyno (Steady-State)

While a steady-state chassis dyno might look similar to its inertia counterpart, there are fundamental differences in how these dynos measure power output.

“If you want true horsepower readings, you must use a steady-state test mode,” says Superflow’s Williamson. “Here the inertial mass makes no difference. Given a proper load cell calibration on the dyno, you will get an accurate power reading.”

These chassis dynos employ the use of a retarding force, the most common being an eddy current brake. It’s essentially an electrical brake on the side of the roller that controls the maximum allowed speed of the vehicle being tested.

“What will happen is that when the roller gets up to that speed, it will begin to apply electrical current to make the magnetic brake slow down the roller,” Strader explains. “The more you press on the throttle, more load is created as the vehicle tries harder to spin the roller, and the brake increasingly tries to oppose that force in turn.”

The steady-state dyno is searching for a state of equilibrium where the vehicle is not accelerating or decelerating but holding steady at the predetermined level. That brake is attached to a load cell which allows a real-time physical measurement of the amount of torque the vehicle is applying to the roller. That prevents the need for the backward calculations inherent to inertia-style chassis dynos.

“Here, not only do I have the ability to hold the engine at any RPM or load, I can also program in a sweep to start and end at specific RPM and have the sweep rate be perfectly linear,” Strader says. “The testing then becomes different because, by doing a linearized test, I’m making that inertia variable something I can control.”

That control offers a wider array of tuning options.

“An inertia-eddy or pure-eddy dyno allows the tuner to drive the vehicle on the dyno in more ranges than available with an inertia-only dyno,” Fong explains. “Many dyno sessions are dictated by the customers needs and requirements. An example would be an airstrip attack type of car where the car will be at sustained high rpm’s for a longer period of time than a street/strip car. A load control dyno will allow the dyno operator run the car in similar conditions and datalog and/or see real time if the fuel demands are being met by the fuel system and/or tune.”

Because of that attribute of controlled versus variable inertia, steady-state dynos will usually read slightly lower than inertia dynos.

“You’ll always hear guys saying, ‘Oh, that dyno is the heart-breaker, or it’s the lie detector,’” Strader adds.

“Without a doubt, the steady-state test mode is the most consistently superior method of tuning—anybody who has the capability to do it will echo that sentiment,” Williamson says. “It’s only an arguable point with those who can’t do it properly.”

Dyno Day Preparation

So it’s clear that enthusiasts who are looking to dyno test their vehicles have a few considerations to make before diving in. Strader says there’s also a couple of things that you can do ahead of a dyno session to ensure that the data you’re getting is applicable to your needs.

“If you’re looking to just find out how much power your car makes, then you can probably just go to any dyno you want,” he explains. “In fact, I might recommend an inertia dyno just so you get the higher number so you feel a bit better, because we’ve already established that there is no such thing as an ‘accurate’ chassis dyno that will give you a number that represents what the engine really makes.”

While some shops have a regimented list of preparation procedures they perform before dyno sessions, it’s in your best interest to discuss with the shop how best to prep the car before coming in.

Strader suggests focusing on their testing methods.

“Ask the shop whether or not they have a procedure,” he says. “Do they have a checklist that they use every time they put a car on that specifies the way the vehicle is strapped down, the way they check tire pressures? Generally what you find is that the shops that are more prepared in that way tend to have a better chance at giving you a smooth dyno day that you can be happy with.”

But if you’re looking to do some tuning, Strader says you’d be doing yourself a disservice to use a shop that only has inertia-type dynos. “The problem is that all of the actual performance gains and mapping on a car that gets driven on the street are in the areas that are less than full throttle, and you’re incapable of tuning for that on an inertia dyno. It’s not that those are bad dyno designs, it’s just the wrong tool for that particular job.”

Want to know more about the nitty gritty of dyno tuning? EFI University has a facility set up to allow people to learn just that. After all, there aren’t many shops around that are going to let you loose on their dynamometers to figure it out. Then once you’ve honed in on your dyno of choice, SuperFlow or Dynojet can set you up with one of your own. In the meantime, we hope this comparison has helped shed some light on merits and drawbacks of different dyno designs and testing methods involved.