Can low latency and high precision be achieved for hardware-in-the-loop controllers?

Precision, hardware-in-the-loop (HIL) systems are often used to develop and test control algorithms for real world systems. The HIL is a virtual twin of a real-world system which can be used to develop the control algorithms for the system, saving time, cost and lowering risk. Faults and dangerous scenarios can be introduced into the virtual twin without risking the real system. Example use cases include anti-lock brake systems, grid protection, industrial process control, and industrial robotics. These use cases require high precision combined with low latency to achieve the necessary accuracy and responsiveness of the relevant control algorithms. This can be a challenge.

Hardware in the loop example

A single channel HIL system example is shown below. The HIL system consists of both measurement and drive signal chains. The HIL measurement signal chain is used to condition and sample the system controller’s stimulus (output), while the drive signal chain is used to represent the response to the system controller’s stimulus. The HIL host emulator runs a model of the real-world system’s physical plant or attached sensors such that its response to the stimulus represents the system’s dynamics. This allows developers of the control algorithm to validate and verify the algorithm.

Figure 1. Single-channel HIL example.

System Challenges

• Low-Latency, Real-Time determinism and Synchronization – The HIL system’s response cannot lag in response to the stimulus. It must respond as quickly as the real-world system. Hence, the measurement chain must have adequate bandwidth and low latency so that the emulator host has time to process the stimulus and generate the corresponding response on time.
• Model fidelity – While the model running on the host must accurately represent the physical plant being controlled, the signal paths that process the stimulus and the response must not degrade the accuracy of the system.
• Scalability – While a single channel example is shown here, most systems are multi-channel, as there are multiple sensors involved in the control of the system. These sensors will have varying signal characteristics, and the HIL system must be able to adapt to the different signal characteristics.

Example Solution for the Stimulus Signal Path

The diagram below illustrates a solution for the signal chain that processes the stimulus input. It includes:

Figure 2. HIL Stimulus input Signal Chain Solution

• Input protection using the ADG5421F, which tolerates over-voltage up to +60V, and can operate on a wide power supply range (up to +22V dual, +44V single).
• A fully differential programmable gain instrumentation amplifier (LTC6373) that enables adaptation to different sensor inputs.
• The ADA4896-2, an extremely low noise (1nV/√Hz typ.), low distortion (-115 dBc @ 100kHz), wide band amplifier with rail-to-rail output, as the ADC driver.
• The AD4630-24 Easy Drive SAR ADC, with dual channels, 24-bit resolution, 2MSPS each channel, +9 ppm INL (max.), and SNR = 105.7 dB (typ.).
• The LTC6655 low noise voltage reference with high initial accuracy (+025%) and low drift (2 ppm/⁰C)
• If needed, isolation using either the MAX22665 or the MAX22444, which offer different RX/TX channel combinations that can adapt to the optional digital output modes offered by the AD4630-24.

In addition to its excellent noise and linearity performance, the AD4630-24, when running at 2MSPS, has a worst case latency of 500 ns to the first bit of the data sample. It also offers multi-lane data transfer that is low-latency friendly. The LTC6373 PGIA offers selectable gain ranges from 0.25 V/V to 16 V/V for signal adaptability, and its high input impedance (5x10¹² ohms) prevents loading of the host stimulus signal chain, preserving fidelity. Its wide supply range (up to +18V) means that it often can directly interface to the host output.

Analysis and Simulation tools

The features and performance of this signal chain can be explored using ADI’s Signal Chain Designer (SCD) tool and its LTspice circuit simulator. Signal Chain Designer is a feature-rich tool that supports signal chain construction and analysis. The analysis capability includes noise/SNR, frequency response, DC errors, power consumption, and relative layout size. Analyze this signal chain solution linked here.

LTspice is a powerful circuit simulation tool which can be used to analyze time domain, frequency domain and noise characteristics of the signal chain. The LTspice simulation files for this signal chain can be accessed at this link.