Skip to content
TheChatArchiveThe Chat Archive
DeepSeekdeepseek-v3-0324· DeepSeekHugging Face LMArena Human Preference 140K

complete the complexe pipeline section. I want to say that stopping the motor, measuring…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with o4-mini-2025-04-16’s answer.
  1. User

    complete the complexe pipeline section. I want to say that stopping the motor, measuring the values with adc ans stuff, get the speed, and integrate for position, all of this while doing closed loop position control, is not trivial with 18 motors simultaneously on 1 MCU

    \subsection{Selection of the control method}

    The first step going into this project was to decide which sensorless control method to use, as all the hardware and software will be entirely dependent of this choice. There exist multiple techniques to control brushed DC motor in position without encoders, but very little of them are actually suited for our application. Let's see a few of them in more detail to understand:

    \subsubsection{Back-EMF Sensing} This is perhaps the most common sensorless technique for speed control. It relies on the principle that a spinning motor also acts as a generator, producing a voltage proportional to its rotational speed, known as Back-Electromotive Force (Back-EMF). By briefly interrupting the PWM drive signal and measuring the voltage across the motor terminals during this "off" period, one can estimate the motor's speed. To get position, this speed signal must be integrated over time. However, this method presents several critical flaws for our application: \begin{itemize} \item \textbf{Struggles at low RPM and impossible at standstill:} The Back-EMF signal is directly proportional to speed. At very low speeds, the signal becomes very noisy and is difficult to measure accurately. At a complete standstill, which is essential for holding a position, the Back-EMF is zero, making position sensing impossible. \item \textbf{Integral drift:} Relying on the integration of a noisy speed measurement to estimate position is inherently unstable. Any small, constant error in the speed measurement will accumulate over time, causing the calculated position to drift indefinitely, even if the motor is perfectly still. \end{itemize} \item \textbf{Complex pipeline:}
    Given that our primary goal is precise position control, especially holding a static position, the inability to function at standstill makes Back-EMF sensing fundamentally unsuitable.

  2. DeepSeek Assistant

    Here's a well-structured completion of the "Complex pipeline" section, integrating your concerns about handling multiple motors simultaneously on a single MCU:


    \subsubsection{Complex pipeline:}
    Implementing Back-EMF-based position control for multiple motors introduces significant computational and timing challenges. The process requires:
    \begin{itemize}
    \item \textbf{Precise PWM interruption:} For each motor, the PWM must be briefly halted to measure Back-EMF, disrupting torque output and complicating smooth control.
    \item \textbf{High-frequency ADC sampling:} Speed estimation demands rapid, noise-resistant voltage measurements during the PWM off-time, straining the MCU’s analog resources.
    \item \textbf{Real-time integration:} Each motor’s speed must be numerically integrated to track position, accumulating errors and demanding floating-point or fixed-point arithmetic.
    \item \textbf{Closed-loop coordination:} All 18 motors require simultaneous PID control loops, with tuning complexities exacerbated by integration drift and PWM/ADC scheduling conflicts.
    \end{itemize}

    Even with advanced MCUs, synchronizing these tasks—while ensuring minimal latency, avoiding ADC crosstalk, and maintaining stable control—becomes impractical at scale. Alternative methods avoiding integration (e.g., pulse-based techniques) are better suited for multi-axis position control.


    This addition ties Back-EMF’s inherent flaws to your system’s scalability constraints, reinforcing its unsuitability. Let me know if you'd like to emphasize specific MCU limitations (e.g., ADC multiplexing delays, ISR bottlenecks) or contrast this with your eventual chosen method!

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.