Brian Thomas Boulter
Applied Industrial Control Solutions LLC
4597 E Sprague Rd.
Independence, OH, USA 44131

© ApICS ® LLC 2000

Automotive industry Dynamometer applications that are designed to test drive-shaft components such as transmissions and/or differentials may use an electric motor in place of a gasoline/diesel engine to simulate the shaft torque dynamics of an engine based test stand. These test stands can also be designed to include pulses that emulate the drive-shaft stresses that result from individual piston firing torques. This paper describes the hardware and software that is required to perform this task.

The Engine-less Dynamometer system architecture shown in Figure 1 can be augmented to include the firing pulse torque generation algorithm and hardware. A complete description of the Engine-less Dynamometer can be found in "Designing a Dynamometer Test Stand with an Emulated Gasoline/Diesel Engine Power Source: A White Paper." The augmented figure showing the INPUT stand Drive and Controllers is shown in Figure 2.

Figure 1. Engine-less Dynamometer System Architecture

The Pulse Generator must inject the pulses at an analog input to the drive. The drive must also have a bandwidth that is higher than 4p times the maximum desired pulse frequency. It must also have the option of injecting the pulse torque reference directly into the analog side of the vector algorithm, NOT through an A/D converter and then summed with the sampled side of the vector algorithm.

The addition of the pulse generation algorithm, and the computer that is used to compute the pulse shape shown in Figure 2 enables the Test-Stand to adequately simulate the torque pulses associated with the firing of cylinders in gasoline /diesel engines.

Figure 2. Firing Pulse Generation

The efficacy of the pulse generation algorithm is dependent on the ratio of the computation performance of the computer producing the pulses and the amount of pulses per second. For example, an output sample rate that produces 2 computed torque pulse amplitudes per pulse will produce a completely inadequate representation of a torque pulse. On the other hand, an output sample rate that produces 100 computed torque pulse amplitudes per pulse will produce a very accurate representation. This relationship can be seen in Figure 3. As a rule of thumb, the minimum number of amplitude samples per pulse is 10.

Figure 3. Output Sample Rate

Not shown in Figure 3. are additional problems associated with aliased signals. An output pulse sample rate that is 10 times higher than the maximum output pulse rate will significantly reduce these effects.

The firing sequence generator in Figure 2. Computes the frequency of the pulse train, based on the speed feedback of the INPUT stand shaft, and a set of engine set-up parameters.

where: = Feedback [rpm] of the INPUT stand motor shaft.

n = number of cylinders to simulate

X = 1,2,3 or 4, depending on the engine type (i.e. V-8/6, Straight 4/6, 4/2 stroke)

The pulse shape generator calculates the amplitude of each sampled pulse output based on a predefined torque pulse shape, the fixed D/A sample rate, and the calculated pulse frequency.

The shape of the torque pulse that is used as the template in the pulse shape generator algorithm, will vary depending on the engine cylinder size, the engine type, the instantaneous fuel rate, the firing frequency, and the output of the TORQUE MAJOR torque regulator.

Briefly, the pulse shape is constrained by the following :

• The normalized time average of the area under the computed pulse curve should be equal to the normalized time average of the output of the TORQUE MAJOR regulator.
• The height of the torque pulse is a non-linear function of the fuel-rate and the engine speed.
• The two above statements imply that the width of the pulse shape will vary depending on the output of the torque regulator, the instantaneous fuel rate, and the engine speed.

The first bullet indicates that the calculation of the pulses should be done at a considerably faster rate than the scan time of the INPUT MAJOR torque regulator. This is to ensure that the time average of the pulse generator is at least 10 times faster than the effective corner frequency of the INPUT MAJOR torque regulator (i.e. f_{SAMPLE RATE} > 10/(p x T_{SCAN TIME}))

The exact non-linear relationship between the height of the pulses, the instantaneous fuel rate, and the engine speed can either be determined heuristically (with measurements for a given engine) or analytically. The analytical solution may prove to be very involved.

A load sharing scheme may be used to split the torque reference from the INTPUT MAJOR torque regulator directly, and the output of the pulse generator algorithm. For example a 50/50 load sharing scheme will result in directly feeding 50% of the output of the INPUT MAJOR torque regulator to the INPUT MINOR torque loop, and dividing the output of the pulse generator by 50%. The ideal set-up would be realized when the direct feed from the INPUT MAJOR torque regulator is set to zero, and the entire torque reference is derived from the pulse generation algorithm.

Given that the minimum number of computed pulse amplitude samples per firing pulse is 10. The sample rate at which the output pulse computer must calculate the pulse amplitude must be:

Similarly, the corner frequency of the drive [rad/sec] must be greater than:

The software must be capable of computing the amplitude of the output pulse at a rate greater than f_{SAMPLE RATE[Hz]}.

Supervisory referencing and sequencing is handled in the Level 2 computer, which communicates with the Automax regulators, the drives, and the pulse generation computer using standard communication protocols. Each application may require a custom communication package, and needs to be thoroughly investigated before proceeding with the design. Industry standard protocols are evolving at a brisk pace. Depending on the needs of the test set-up, the best communication package for each application must be identified prior to finalizing the design.

Considerable care must be given to the selection of the D/A converter used in the pulse output generator. It should have a linearity specification better than +/- 1%, and an s.n.r. better than 65 [db]. In addition, the D/A should be located as physically close as possible to the Analog input of the drive. This is to minimize noise corruption of the pulse signal. Care must also be taken to ensure that there are no ground loops in the system, this is because of the use of an analog signal as a critical reference signal in the control loop.

An engine-less test stand may be configured to generate torque pulses that simulate the torque pulses that result from cylinders firing in a gasoline or diesel engine. The effectiveness of this approach is dependent on the use of a fast computational engine for the purpose of calculating the pulse shape.