Blackbox curve is essentially flight control data sampled by time. Betaflight reads the gyro, receiver and other states in the continuously running PID loop and calculates the PID and motor output; Blackbox then records every few loops at the set frequency. When reading the graph, first look at the time on the horizontal axis, and then look at the real unit and scaling of each trace. You cannot directly compare the values just because the two normalized curves are at the same height on the screen.
From PID loop to Blackbox data point
The video compares the curve to a stock chart: the system reads the current state at regular intervals and plots it as data points. Betaflight's PID loop can be thousands of Hz. Each loop reads the gyroscope angular velocity and rocker target, and then mixes them into each motor output after PID calculation. Blackbox is not another controller, it just extracts and saves some variables from this real-time process.
Logging frequency is usually lower than PID loop. Betaflight records gyro, setpoint, PID, motor and other data every few cycles according to the Blackbox settings. Therefore, the log must not only retain the dynamics of interest quickly enough, but also consider the flash memory capacity and writing capabilities. The exact frequencies available depend on the flight controller, firmware, and analysis goals.
The sampling frequency determines the highest frequency that can be seen
The video takes a 5-inch camera and 1000 Hz log as an example: According to the Nyquist principle, ideally the highest resolution can be about 500 Hz. This range covers the noise that many common parameter adjustments focus on, but 1000 Hz is not a fixed optimal value for all models; higher speeds, special filtering, or other analysis tasks may require different settings.
Undersampling can cause aliasing, making high-frequency components appear to fall at the wrong frequencies. Before increasing the log rate, you should confirm that the device can record stably without a large number of frame drops. First make it clear whether you want to analyze control response, low-frequency frame motion or high-frequency motor noise, and then choose an appropriate sampling rate.
The horizontal axis is time, and the vertical axis is not a unified unit.
The horizontal axis represents the log time. The variables shown simultaneously on the vertical axis can vary greatly: a gyro can reach thousands of degrees per second, while battery voltage only varies in the tens of volts range. If placed directly on the same coordinate scale, smaller variables will be attached near the horizontal axis, so the analyzer will scale or normalize different traces to observe the shape.
The scaled screen height is not equal to the actual value. To read a point, look at the raw value and units in the sidebar, marker, or tooltip. Video example: The positive and negative values of roll gyro represent opposite rotation directions, but the positive direction of each axis should still be based on Betaflight coordinate convention and the current analyzer label.
Which curves can be compared to each other
Only curves with a clear and consistent relationship between units and scaling can be directly compared using height. setpoint and gyro are often used to observe the target angular velocity and actual following; PID items, motor output, voltage, etc. must first check their respective units and scales. Arbitrarily stacking two lines with similar appearance may lead to completely wrong conclusions.
When reading logs, you should first hide irrelevant traces in the legend, select an axis and a short time window, and then use actual values to answer specific questions. This article is a curve overview and does not contain recommendations for filtering or PID values for the next stage.
Operation steps
- Confirm PID loop and log frequency
Record the flight control loop and Blackbox sampling settings, and understand that the log only saves the state every few loops.
- Select sampling rate according to analysis target
Use the Nyquist principle to estimate the visible frequency range while confirming that flash memory and writing are stable.
- Read the horizontal axis time first
Locate the short window where the event occurred, and then expand the required gyro, setpoint, PID or motor trace.
- Read the real value and unit from the sidebar
Do not treat the normalized screen height as the original value, use a marker or tooltip to view the actual data.
- Only compare curves with the same unit and scaling
For example, check the following of setpoint and gyro; other variables must be confirmed first before explaining the scale.
FAQ
Does the Blackbox recording frequency have to be equal to the PID loop?
No need. Logs are typically set to sample every few PID cycles to balance frequency range, capacity, and write load.
Why do 1000 Hz logs often say that 500 Hz can be seen?
According to the Nyquist principle, the highest frequency that can be resolved by ideal sampling is about half the sampling rate; in practice, filtering and aliasing must also be considered.
If the two curves are the same height on the screen, are the values the same?
Not necessarily. Analyzers may scale or normalize different variables separately and must look at the true values, units and scale.
Which curves are best followed by direct observation?
setpoint and gyro are often used to compare target angular velocity and actual response, but you should still confirm that the analyzer uses consistent units and scaling for both.
Full timeline transcript
Transcripts are arranged according to video time, making it easy to quickly locate the explanation content. Transcript language: Simplified Chinese.
Hello everyone, I am Lao Li. Today we will talk about the three main bones. Oh, no, I'm quarter rank. Today we continue to play Betaflight’s black box. In the last issue we talked about how to find the black box file and open it. In this issue, we will take a look at how to interpret these black box curves. Why did I talk about stocks just now? It’s because these curves have many similarities with this stock trend chart.
Friends who have seen the stock trend chart should know how the stock trend chart comes from. In fact, it is to look at the current stock price every second or half a second and draw a data point. This is actually very similar to our black box. As you all know, the PID algorithm is somewhat different from our usual algorithms that run once and stop. It loops continuously. In Betaflight, we can also set the frequency of this loop, such as the common 8000Hz or 4000Hz.
This frequency determines how long the processor will run the PID loop. In each cycle, the processor will read Toloy's angular velocity, the receiver's wave sensing signal, etc. Then through the calculation of the PID controller, it is determined how much change each axis should make, and then mixed into the throttle of the four motors to try to achieve this change. So what does this PID loop have to do with black box data?
In fact, you can understand that the black box cognitive output is part of the PID loop, but it is not the main one. According to the output frequency of the black box we set, Betaflight will decide to record various current data every few PID cycles, such as the reading of gyroscope 1, PID value, motor output, etc. We can set the output frequency on the black box tab of Betaflight. The more professional data is called the sampling frequency. Generally speaking, for a 5-inch machine, 1000Hz is enough.
Because if you have studied signal sampling, you will know that 1000Hz can sample noise up to 500Hz. Generally, the noise in 500Hz is what we are more concerned about. For the 5-inch level, we will adjust the parameters later. We will explain in detail how to understand the generation of this black core data. We will understand the horizontal elbow of this graph. It is actually a time elbow. It is exactly the same as our stock movement. After understanding the vertical elbow, what is the vertical elbow? The middle school teacher told us to look at the coordinate elbow first.
Why does the vertical axis have no units? This is actually because we collect a lot of various data, and their value ranges are very different. For example, the value of the gyroscope may change between -3000 and -3000 degrees per second, but the battery voltage value may only change between 16V and 25V, for example. If you simply plot these values under a coordinate system, the voltage value may always appear to be lying on the horizontal axis.
Because 25 is too small compared to 2000, right? So in order to solve this problem, these curves have been normalized. The so-called normalization can be simply understood as reducing them to 0-1. The actual operation of this curve may be a little more complicated, but this understanding is enough. If you want to see the actual size of this value, you can also see it in the column of the figure next to it. For example, we take the reading of the gyroscope on the ROH axis as an example. The current point
The corresponding value is 65 degrees per second, which means the aircraft rotates to the right at an angle of 65 degrees per second. The corresponding Fuji 43 degrees per second means the aircraft rotates to the left at an angle of 43 degrees per second. This is what these governments mean. Of course, different curves may have slightly different government meanings.
[Speech cannot be reliably recognized here]
Okay, this video may take more than five minutes. In the next video, I will show you how to adjust your green wave settings through several commonly used curves. Thank you for listening to my BB. You are welcome to click three times. See you in the next issue.
