In our "Dynos Lie, Dyno Dash" article in the June 2003 issue, we proved that the readouts of the same 350Z on different brands of dynos can vary greatly. Now we'll uncover tricks that can be used to generate misleading dyno results. We'll also provide tips on how to better optimize the consistency of readouts while revealing some of the compromises of dyno testing. The idea is not merely to catch dyno cheaters, but to also better understand how certain actions/factors can alter dyno test results.
Hot Run vs. Cool Run
The first dyno run when the engine is cool usually produces the biggest power. As heat builds in the engine, the peak power number usually settles at a lower level. This is especially true in turbo applications where heat rises. Intercooler efficiency and the placement of cooling fans can play a role in the heating trend.
The scheme here is to take a hot baseline and compare it to a cool product-test run. A lower-whp baseline coupled with a higher-whp test makes the product appear to make more power than it actually does.
Daisy Chain/Rapid-Fire Runs
Performing back-to-back-to-back pulls without ample cooling time between runs can cause an unnatural buildup of heat. This will adversely affect the before-and-after testing of components. The goal is to conduct testing at consistent, normal engine operating temperatures.
The Density Intensity Game
Another way of cheating is to play the density intensity game - conducting baseline runs in the heat of the afternoon and run the after-tests in the cool evening air. While some dynos can adjust for atmospheric conditions, the bottom line is denser air makes more power.
Beat the Drum
The crooked dyno operator can reprogram the weight of the drum, which will allow the dyno to create anomalous power numbers.
The High Ground
The same crooked dyno operator could elect to change the elevation value in the computer, which would have a critical impact in power generation. Naturally aspirated engines are especially susceptible to this ploy.
Changing the load by tightening or loosening the tie-down between runs can compromise any comparisons made with the dyno testing.
Turning on the air conditioning during baseline testing will cost a few horses and make the after-test all the more impressive.
The Old Switch-a-roo
Swapping in a Civic DX graph in place of your Integra GS-R baseline will make for some big power gains later.
ReStrapping the Car When possible, have work conducted while the vehicle is still strapped to the dyno after baseline testing. On some dynos the positioning of the tires on the rollers and the tautness of the straps can change the load characteristics seen during testing. Items like intakes, exhaust systems and most headers are the types of products that can be installed without unstrapping the car.
Different Day Dynoing
This one can be hard to avoid if you are having extensive work done. Check to see if the dyno in question has a weather station that can correct for any big differences. Beyond that try to arrange testing for about the same time of day as baseline runs were performed. This will lessen the effect.
Things to Know
A key element to producing comparable dyno charts on a global basis is the Correction Factor, SAE Standard J1349, which applies the following weather station data--atmospheric pressure 29.23, air temperature 77 degrees Fahrenheit and humidity 0 percent--to all tests. These exact figures allow apples-to-apples comparison of runs from different cars, different facilities, etc. So graphs with SAE-corrected power were made to this standard. The dyno in question must have a weather station and the proper software in order to generate SAE-corrected data.
An option on Dynojets, smoothing takes a jagged graph and delivers a more readable curve. Smoothing of the graph can be done by entering a value between one and four at the appropriate place in the computer.
The alignment of the vehicle being tested does affect the outcome of a dyno test. So if a car is dyno'd, then run over a curb, thus altering the toe or camber, the car may not have repeatable runs. This is especially true on Dynojets.