General
A dynamometer, or "dyno" for short, is a device used to measure power and torque produced by an engine. There are two types of dynos; One that gets bolted directly to an engine, known as an engine dyno, or a dyno that can measure power and torque without removing the engine from the frame of the vehicle, this is known as a chassis dyno.
Chassis dynamometers are capable of very accurately measuring the speed, torque and power that is delivered to them. With the appropriate hardware and software, they are also capable of applying a well-controlled loading to the vehicle under test. The typical loading modes used with a chassis dynamometer are constant force, constant speed, or a vehicle-simulation value.
What are the Differences Between Chassis and Test-Stand Dynamometers
It is important to remember that a chassis dynamometer reports the power, force, and speed experienced by the dynamometer’s roll shaft(s). A power figure obtained for an engine using a test stand dynamometer will (and should) inevitably be higher than the power figure obtained using a chassis dynamometer, for (among others) the following reasons:
- On a test stand, there are no torque-converter/clutch, transmission, driveshaft, differential or axle bearing losses.
- On a test stand, there are no losses between the tires of the vehicle and the rolls of the dynamometer.
- On a test stand, some or all of the engine accessories may be disconnected.
- On a test stand, the engine intake air, water and oil supplies may be externally controlled.
- On a test stand, the exhaust may be different than the exhaust system used on the vehicle.
Why Are There Differences in Reported Power Between Dynamometers?
All of the following factors can influence the power measured by a chassis dynamometer:
- Tire compound, pressure and temperature.
- Engine, transmission and differential temperatures.
- Lubricant types.
- Method of vehicle restraint (downward pressure will waste power).
- Type of testing performed: fast decelerating sweeps will generate the highest values, steady state tests will generate intermediate values, and fast accelerating sweeps will generate the lowest values, due to the internal power requirements of the engine and drive-train in the vehicle under test.
>
- Atmospheric condition corrections to different “standard” conditions.
- Operator driving differences (can be very significant!).
- Data acquisition options, particularly smoothing/averaging and clipping functions.
While all chassis dynamometers should report the same power output for the same vehicle, this is seldom the case. Most commonly, the difference in reported power values for the same vehicle on different dynamometers can be traced to one or more of the factors listed above.
What is "Virtual Inertia" and why is it important?
Mustang’s Chassis Dynamometers utilize Eddy Current Loading Technology in conjunction with Mustang’s patented Virtual Inertia™ Controller Technology. Virtual Inertia is a patented method of simulating driving conditions on a chassis dynamometer. This technology makes true Road Testing possible without leaving the safe confines of the shop – a benefit that is not available with competitive dynamometer products.
During automotive tuning, a chassis dynamometer is an important tool to determine whether or not the changes you have made in the ECU are beneficial. There are two types of dynamometers on the market; an inertia style and a loading style. A chassis dynamometer that can load the vehicle will allow for a more accurate tune on the vehicle than one that is an inertia style. When you are tuning a vehicle on the dynamometer, you want the amount of load being placed on the vehicle to be as close to the amount of load the vehicle sees when it is on the street. This includes weight of the vehicle, wind friction and road friction. The only way you can do this is with a loading dynamometer.
When using an inertia style dynamometer, the only load the vehicle sees is the weight of the drums of the dynamometer. This type of dynamometer is adequate if you are only doing horsepower pulls, but it will not allow you to do accurate tuning. When tuning a vehicle, you want to be able to tune every load site in the ECU’s fuel and timing map. This is done by adjusting how much load is on the vehicle. With an inertia style dynamometer, you will only be able to do testing in one load column because you cannot adjust the load on the vehicle.
For example, let’s tune a late model Chevrolet Camaro. This car weighs about 3500 pounds. When running this vehicle on an Inertia Dynamometer, the only load the engine sees is the load created by the rollers. This load is about 2600 pounds. Now let’s tune this vehicle. You start by leaning out the air-fuel ratio and advancing timing. The car runs great on the dynamometer. Now let’s pull the car off the dynamometer and run it on the street. You will notice that the tune you did on the dynamometer is going to be to lean and the timing is going to be advanced too much. The reason for this is because you tuned the motor in a 3500 pound vehicle to push around 2600 pounds. So when you put the car on the street, the motor is now pushing around about 1000 pounds more and it has to deal with wind and road friction. At 100MPH, wind friction can absorb almost 60 horsepower depending on how aerodynamic the vehicle is. This will easily cause the motor to experience detonation which will eventually damage the motor. Another issue with the amount of load being to light compared to what it usually sees is the ability to accurately data-log information. As a result of the vehicle accelerating so quickly, the computer may not have enough time to accurately data-log all of the important information (i.e. temperatures, flows and pressures). This will hinder the tuner’s ability to analyze all of the vital information needed to accurately tune the vehicle. Moreover, turbocharged vehicles have a difficult time building full boost on inertia style dynamometers. Turbochargers run on exhaust gases. When there isn’t enough load on the motor to produce the proper amount of exhaust gases, the turbocharger won’t spool up to full boost. For this reason, you will need to adjust the boost controller to make full boost on the Inertia Dynamometer. After making adjustments on the Inertia Dynamometer to build Boost, a tuner will need to adjust the boost controller again before going on the street - else you will make to much boost due to the extra load.
Now let’s tune this same car on a Mustang Dynamometer. Before testing, you input the vehicles weight and HP@50MPH. This allows you to use the vehicle simulation and accurately load the vehicle. When the car is on the Mustang, it sees the same amount of load it sees when it is on the street – weight, wind friction and road friction. It is the closest thing you can get to performing a road tune on the vehicle. Therefore, when you adjust the fuel and timing values in the ECU they will be accurate when you take the vehicle on the road. Having the same amount of load on the dynamometer as on the street will also allow the vehicle to accelerate at the same rate it does when on the street. This will allow the computer to accurately datalog all of the information you will need to properly adjust air-fuel ratios and timing. Also, because you are applying the proper amount of load, turbocharged vehicles will spool up at the appropriate time and produce the desired boost levels.
If you are looking to perform just power pulls, an inertia style dynamometer will be acceptable. But if you are looking to accurately and SAFELY tune vehicles to run on the street, a loading dynamometer is the only way to go.
Please contact a Sales Engineer at Mustang Dynamometer if you have any questions about this Tech Tip.
Basics of PID Closed-Loop Control
Dynamometer’s that incorporate a loading device, such as a PAU, motor or water brake, generally employ some type of closed-loop control to achieve and maintain the desired loading and/or speed values. Inertia-only dynamometers obviously have no need for any closed-loop control, since no loading device is present. All Mustang Dynamometer control systems employee the PID form of closed-loop control.
PID closed-loop control constantly compares the desired loading or speed value to the actual measured loading or speed value, and makes fine adjustments to the PAU control signal to make the measured loading or speed match the desired loading or speed. The letters “PID” stand for “Proportional”, “Integral” and “Derivative”, which represent the different ways by which the output signal is modified based on the desired values and measured values. For each element in the PID control loop, a constant is used to adjust the magnitude of adjustment the PAU control signal will experience due to various types of errors between the measured and desired values.
The “Proportional” term constant is used to scale the system’s response to a simple error between the desired and measured loading values. Thus, the “Proportional” term is never 0 unless the measured value exactly matches the desired value. Additionally, The “Proportional” term is used to provide over-all scaling of all three terms of the PID control loop. The “Proportional” term handles low to medium frequency response requirements.
The “Integral” term constant is used to scale the system’s response to a following error between the desired and measured loading values. The “Integral” term is used to handle situations here the measured values continuously lag behind the desired values, in the same direction of error. In almost all dynamometer-loading schemes, the desired value changes rapidly, and the measured value chases the desired value both upwards and downwards. Thus, the “Integral” term is very seldom required, and its use is not recommended in our control systems. The “Integral” term handles relatively low frequency response requirements.
The “Derivative” term constant is used to scale the system’s response to the rate of change of error between the desired and measured loading values. The “Derivative” term responds to sudden changes in both or either of the desired loading or measured loading values, and to the rate of change of the error between them. The “Derivative” term is never 0 unless the rate of change of error is 0. The “Derivative” term handles medium to high frequency response requirements.
The basic logic for our implementation of the PID closed-loop control logic is given below. The “Desired” and “Measured” values may be torques, forces, speeds, etc. The logic given below happens 100 times per second (by default).
Error = (Desired – Measured)
ProportionalTerm = (PConstant * Error)
IntegralTerm = IntegralTerm + (PConstant * ICconstant * Error)
DerivativeTerm = (PConstant * DConstant * ((Error – LastError) / UpdateTime))
PAUVoltage = ProportionalTerm + IntegralTerm + DerivativeTerm
OldError = Error
Since the mathematics above remain constant regardless of the value being controlled (e.g. torque or speed), the sign of the “P” constant is critical. This can be observed by considering the “ProportionalTerm” above; if the measured torque is less than the desired torque, the PAU signal must be increased, whereas if the measures speed is less than the desired speed, the PAU signal must be decreased. A sign-error on the “P” constant will result in a loop that works backwards, in effect saying to itself, “The actual load is too low, I must decrease it!”
“Tuning” a PID-type control loop refers to the process of determining the P, I, and D constants required to achieve the desired control characteristics. Generally, the time required to achieve a 90% response to a step change in the desired loading value and the long-term stability of the control loop are the primary considerations in tuning a PID loop.
People new to the concept of closed-loop control frequently ask if there is a method of directly calculating the optimal P, I, and D constant values. Control theory tells us that, in theory, this is possible. In reality, tuning a PID loop requires that some initial PID constant values be selected, and iterative tuning be performed. The initial PID constants selected must be of the correct sign, and must be relatively small. By appropriately selecting the initial PID constants, a very slow, but probably stable, control loop may be achieved. Iterative tuning, wherein the PID constants are gradually varied and the control loop’s response evaluated, is then used to increase the PID constants until the desired control loop response characteristics are achieved. For the truly motivated reader, the PID control loop math outlined above can be used to select initial PID constants, knowing that the PID loop’s output value is fed to the PAU controller, and has a range of 0.0 volts to 5.0 volts, and that the control loop is updated (by default) at 100 times per second.
Mustang Dynamometer strongly recommends that the default PID
values for your dynamometer type be used.
Operators familiar with PID closed-loop control may wonder why there are individual PID constant sets for each test routine, since a dynamometer can generally only control the torque on or the speed of its rolls.
The reason that individual PID constant sets are supported for each test routine is to allow the operator to configure tests that use the same type of control (torque or speed) to have different response characteristics.
Why is Mustang Dynamometer a superior dynamometer?
Mustang Dynamometer systems incorporate the best elements available, from mechanical systems through control software – here are some of the details:
Strain gauge load sensing
< 0.1% non linearity
- High frequency response
Low noise via shielded cabling and filtered differential input channels
- 12- or 16-bit analog inputs, for 0.025% to 0.0015% of range resolution
- 5th order, maximally flat analog filter, 14 – 21 Hz, set just below large magnitude mechanical vibration speeds
Extremely precise speed and acceleration measurement
- Tough and durable geared tooth encoders
- Custom, high repeatability Hall effect sensors
- Synchronized anti-noise maps, map out noise due toothed gear machining imprecisions and mounting imprecisions
- 100 mSec time base for speed and acceleration, nearly perfectly in phase with torque measurments
- Dedicated hardware for edge to edge timing, rather than simple teeth-over-time measurments
- Custom toothed gears with single pulse/revolution channels for map synchronization
- Speed measurement is precise to 1 part in 400,000, and (electrically) accurate to 50 parts/million (based on crystal oscillator’s 50 ppm tolerance)
- Acceleration is precise to 1 part in 20,000, and (electrically) accurate to 50 parts/million
- For 4WD/AWD dynamometers, front and rear speeds and accelerations are measured independently, and used to obtain a weighted average based on dynamometer inertia on the front and rear rollers – this removes large reported inertial loads due to oscillations through the front-rear driveshaft or belt (this is always present, and must be handled in software, infinitely stiff connections are not available in real world systems).
Eddy Current Power Absorbers (PAUs)
- Provide the smoothest available load, no periodic (rotational) load variations at all.
- Provide fast (<50 mSec) response to load changes
- No large electrical power requirements
- No water cooling requirements
- Tough and durable
- Relatively Low cost
- Current controlling eddy current brake drive electronics
- Provide fast eddy current brake control, using current feedback
Road load simulation with inertia simulation
- Applies the same load to the vehicle on the dynamometer as on the road
- Very precise, high dynamic response control
- Control logic and components have been verified (when identical parts used in emissions testing systems) by the California BAR’s dynamometer certification laboratory, using the lab’s “dyno-dyno”.
- Control logic and components have been verified (when identical parts used in emissions testing systems) by Sierra Research using their dynamometer instrumentation system.
- Power curves are done at the same acceleration rate that will be experienced on the road, resulting in measured values that can actually be experienced.
- Closed loop controller updates all inputs and outputs and performs control logic at 100 – 300 Hz (10 mSec to 3 mSec) update rate.
- Parasitic losses of vehicle on dynamometer (as opposed to dynamometer only) can be used, similar to US EPA certification dynamometers, to ensure the load applied accurately simulates the on road loading.
Built in coastdown verification routine
- Verifies correct configuration, calibration and functioning of the dynamometer inertia value, roll diameter value, speed encoder pulses/revolution value, parasitic losses data, loading device control, and system clock.
Do you have an adaptor to test boat engines, both outboard and inboard?
Mustang Dynamometer primarily specializes in testing equipment for inboard boat engines. Contact us for more details, and to discuss specific project requirements.
