This is a live parameter adjustment record of Betaflight 4.3 Nightly. The target is an 8S, 10-inch camera with about 30V. The author started from the default PID and existing filter baseline, and gradually improved the locking feeling through test flight, Blackbox spectrum, Step Response, D-first slider and P/FeedForward adjustment. The values in the video only apply to this large-size machine and can never be directly copied to a 5-inch machine or other configurations.
Establish a safe and comparable baseline first
The author flashed Betaflight 4.3 Nightly near November 4, 2021, and retained common configurations such as personal Rates and OSD; the PID basically started from the default, and the filtering was preprocessed due to the low-frequency noise of this large machine. Nightly firmware may contain unstable changes. Before flashing, you must save a complete diff/all backup, confirm the flight control target, and prepare a recovery path.
The purpose of the first test flight is not to pursue the limit, but to record the "meat", shaking and abnormal performance when pushing the throttle. The mass and moment of inertia of a 10-inch machine are much larger than those of a 5-inch machine, so the response will not be the same. All ground connection and configuration operations should be done with the propellers removed; keep a distance during outdoor test flights, check the motor temperature, and refuse to hand-connect large aircraft like in the video.
Distinguishing control response from mechanical noise with Blackbox
The author uses the gyro_scaled Blackbox log to view the gyroscope spectrum. The absolute height of the spectrum will be affected by the log length and display scaling. What is more worthy of comparison is the relative relationship between the peaks of each frequency band in the same log, and whether the noise still enters the D-term after filtering. A peak around 70 Hz near the yaw axis in the recording continued to be of interest and attempts were made to suppress it with static filtering.
More filtering is not better: it can reduce noise, but it also introduces delay. The correct cycle is to first confirm the stable existence of the peak, find the source of the rack or power system, then make targeted adjustments, and then re-test and record. If the noise still penetrates into the D-term, you should not just continue to increase D to mask the problem.
Test flight loop for D-first slider
Simple Tuning in Betaflight 4.3 simplifies PID scaling management. The video adopts a D-first approach: slightly increase D, conduct a short test flight, observe the locking feeling, oscillation, abnormal noise and motor temperature after stopping the action, and then check the Blackbox. D provides damping. If the amount is appropriate, it can reduce the sluggishness after the action is completed; if it is too high, it may amplify high-frequency noise and cause the motor to heat up.
The author gradually tried upwards from a more conservative magnification, and finally continued to verify within the allowable range of the interface, but this was only an on-site process for the 10-inch 8S machine. After each change, a landing inspection must be carried out. If there is continuous oscillation, whistling, abnormal temperature rise or a tendency to take off and rise, it should be stopped immediately. Do not copy the 1.2, 1.4, 1.8, 2.0, etc. magnifications in the recording.
Rebalance P with FeedForward
To observe the Step Response, the author temporarily lowers the FeedForward to make the response more easily reflect the relationship between P and D. After D has established damping, increase P to enhance the tracking of the target angular velocity; if the curve is too wide, it indicates insufficient following, and if it overshoots or oscillates, it indicates that the proportion needs to be rolled back. Then restore the settings related to the joystick response, so that the aircraft can follow the hand more easily while locking and stabilizing.
The final parameter adjustment is still a compromise, rather than pushing the slider to the limit. At the end of the video, the author believes that the hand feel has been significantly improved, but it also shows that time is limited and further research on filtering may be possible if fine-tuning is done in the future. Before saving the final configuration, you should complete flights with various throttles and maneuvers, check motor temperatures, compare multiple log segments, and retain a rollback baseline.
Operation steps
- Record Nightly firmware and initial configuration
Save the target, firmware date, full configuration backup, and filtering changes before performing the first baseline test flight.
- Read gyro_scaled Blackbox spectrum
Compare the noise peak value of each axis and the relationship before and after filtering, without drawing conclusions based on the absolute height of logs of different lengths.
- Check filtering for stable peaks
During the recording, focus on observing the peak value of about 70 Hz on the yaw axis, and verify whether static filtering can reduce it from entering the D-term.
- Slightly increase D and short test flight
Change only one gear at a time, check the locking feeling, oscillation, abnormal noise and motor temperature, and then check the log again.
- Temporarily drop FeedForward to view Step Response
Reduce the impact of direct joystick feedforward and observe the contribution of P/D to target following and overshoot.
- Raise P and restore stick response
Enhance target following after the damping is controllable, and then use FeedForward to adjust the degree of follow-up without pursuing the extreme value.
- Do final log and temperature verification
Review the Step Response, noise, motor temperature and flight feel under multiple throttles and movements before saving.
FAQ
Can I copy the PID slider value in the video to the 5-inch machine?
Absolutely not. The author clearly reminds that the parameter range of the 10-inch aircraft is very different from that of the 5-inch aircraft, and the values only represent the test flight process of this 8S aircraft.
Why do you need to check the motor temperature repeatedly when adjusting D?
D amplifies high frequency noise. If it is too high, even if the hand feels tight for a short time, it may cause oscillation and the motor to heat up.
Are the tallest bars in the spectrogram the worst noise?
Don’t just look at the absolute height. Log length and display scaling will affect height, and peaks under the same conditions should be compared relative to each other, frequency location, and whether they penetrate into the D-term.
Why temporarily lower FeedForward?
This makes it easier for Step Response to display the basic relationship between P and D; after completing the balance, resume feedforward to adjust the joystick and feel.
Is the Nightly firmware suitable for direct use in daily flying?
Nightly has additional risks. You must check the target, save backups, understand the recovery method, and verify it in a controlled environment. It cannot be used unconditionally as a stable version.
Full timeline transcript
Transcripts are arranged according to video time, making it easy to quickly locate the explanation content. Transcript language: Simplified Chinese.
Hello, hello, everyone. I’m Quarank. It’s such a good time to see these big eyes. Look at this.
Yinfeng Zhenren, do you feel like you are rushing to catch up with the sea and sky group? I am not flying today, so today I am going to adjust an airplane. I received and took this 10-inch airplane. I have probably finished this airplane. It has been almost a year, but I haven’t flown it much, I haven’t flown it much, and I haven’t adjusted it much. And now, doesn’t it feel like Betaflight 4.3 is about to be released? This is a question mark.
We will take advantage of the upcoming release of BLHeli 4.3 to try out this BLHeli 4.3 to see if we can change the firepower of this aircraft. We can also win more opportunities to be flown in the future. Among the many aircraft, don’t just sit on the bench all the time. I seem to hear him saying, hurry up, I want to fly, okay, then let’s adjust it.
To briefly describe the current situation, I just flashed the Nightly firmware of Betaflight on November 4th, and nothing else has changed. I just flashed my personal Rates, OSD, and other commonly used configurations, and then put them between PID and this Filter. The only thing that touched me was that I added some more Filters and lowered the lowest box of this Filter. Because the default box, why can’t my players fly? The lower one is too noisy.
Let’s fly it now to see what the default situation is like. This aircraft is an aircraft with this BIOS. I use the iFlight’s X80 ESC, so it only goes to the BIOS OK. Let’s still fly OK with this priest. The wave motion feels very fleshy for such a big machine. In fact, it is also fleshy, which is understandable because
The plane is bigger, what is its PID called? PID Authority, it is not that high. Think about you, those small Hummingbirds, that wind bird, it can spin to a stop, but because you have seen an eagle, it spins to a stop, it must not be as flexible, right, or as exciting as that small bird, right? OK, when we push the accelerator, we feel that this is a little bit wow, quite a lot, quite a lot, quite a lot of wow.
OK, that's the current situation in China. Let's see if we can adjust it better. Let's connect it to the computer. Today is also the first time I use this 10.3. Otherwise, today is also the first time I use this 10.8 configurator. Before, I usually used this CLI. So how do we adjust it today? You can first take a look at my green wave camera. My green wave is the default, and then adjust the q value of the line rate of the dynamic green wave to 300. The default is 500, which cannot fly.
D-term is adjusted to 75 which is the lowest and then here I lower it to 100 and now everyone just saw this and it feels like it can fly. We can take a look at the noise in a moment. What is the black box now? I am now at gyro scale. We can take a look at the noise and then all the settings in this PID are default. We can try this first. Let's turn it up a little for you and take a look at this black box mode. OK. Let's connect to the aircraft and open this black box and take a look.
Here is this Betaflight, this black core day, which should be what we smell flying. Take two pictures of this and go directly to see the noise. You can turn this up a little, turn it up a little, because the height of this Spectrum Diagram is actually ah, it doesn’t make much sense to say the height because the longer your day is, the higher its height is. What is meaningful is the relationship between each peak and their relative height.
So what do you think you can drag out to see if it is clear? We can take a look at this Gerro. There is a lot of noise between 110Hz and 50GHz. It should be noise. It can be about the same. We can actually eliminate it if it is higher. But generally speaking, it is okay because most of the signals are within this 10Hz.
The right axis of the Gyro does not have much movement, so the right axis signal is not particularly large. The only thing is that there is some noise in this 71Hz area. We will see if we can disable it through the settings later. Then after the Fuelter is completed, you can see that the noise in the higher places is all ground out. This is usually the same as the Gyro, and the flesh position is similar.
Try the simple tuning in 4.3. To put it simply, first adjust the P value to the highest position and then adjust it to the plane.
Start this crazy position and then lower it a little and then adjust the p value. Adjust the p value to a high enough position. Let's see if this method works. I'm too conservative and doubtful. OK. As we just said, add a Static Filter to this 70Hz to eliminate the noise of 70GHz. Let's lower it here. Let's add a 60 and let's test fly. It's so cold. Oh my God, look at me recording a video for everyone in the cold wind. Should I give you a like?
It's still a little wobbly, but it feels better than before. I just made the ground pool a little larger because it's a braking ground that blocks all movement, so the shaking just now will have a little effect. I just adjusted the ground to 1.2 or something, and then added this 60Hz filter barrel filter on the right axis. It feels like it's much more sensitive than just flying. This brake is much more sensitive.
It basically means that after you move it, it will settle there and feel that way instead of being there after you move it. OK, let’s come down. Those friends who want me to hand pick it up can go and take a look at this safety specification. This is such a big issue. I won’t hand pick it up. We connect to the computer. This time we directly enter the black box mode. Let’s see the data first. How about this date file? There is still some noise at 70Hz in this place. There is still some 70Hz noise on the right hand.
There is no sales fishing, and in the end it is worn on D-Term. It seems that we need to find a way to see if we can sell the 70Hz noise. I can take a look at this step response. There are not too many signals, but overall there is some overshoot. We have not reached the highest value because of the D value, right? This simple telemeter is, we use this D-first telemeter,
Increase D until something bad happens. OK Then let's continue to Increase D and then let's think about how to control this 70Hz noise. Because this will directly affect how high our D value can rise later, I prefer to lower it. If this place allows us to choose different future types, we can add a pt2 or pt3 or bycord to it. The effect may be better. No one has passed 50.
Because we can see that this is caused by the right axis of the right axis engine. Then we continue to use its future to adjust the D-Gain a little higher. We raise it from 1.25 to 1.4 and give it a try. Yes, we need to see this step response and disable Feedforward completely. It does not allow me to drag it. Maybe it is because of the problem with the experts mode. We can activate the experts mode. Now put this
Adjust the FeedForward to the lowest. Let's try again to make me warmer. I'm used to asking him to take my high dog, but now I bring some of these cameras. Hey, it feels like this, and it can go up. There is no chirping. The response of the joystick and the aircraft are obviously not so responsive. Because we just lowered the FeedForward to 0, but it doesn't matter. We can adjust it back. We can see that it is a little bit smooth and the flight is just right.
The temperature of the motor is quite good, almost lukewarm. Look at this combination. This 008 came out today. I feel that the noise of this 70-something box is still there, but it is a little lower than just now. It is a little lower than just now, but I don’t know if it is worth it. But I don’t know if it is worth it. Our PD ratio still feels a bit overshoot. Why? If we adjust the D value to such a high level, it is still a bit overshoot.
It may be because our D_MIN is not turned off. Let's take a look at D_MIN first. D_MIN is still turning off this D_MIN. It is not allowed to turn off D_MIN yet. It has not turned off D_MIN. It has not turned off D_MIN. But let's try to use DMAX to continue to increase it to see how much DMAX can be raised. I was just a little bit. Hey, I was at 1.4, right? Let's raise it to 1.5. At this time, we need to pay attention to see if the aircraft will be OK.
It may shoot directly into the sky because it is too big. It feels okay. It feels like it needs to be locked. It is very stable, but it feels like it doesn't want to move. My joystick frame doesn't move very crisply when I hit it. We can do it later by adding FeedForward and raising the fart. I personally don't particularly like this D_MIN. Let's take a look. We are using this Spider of Betaflight 4.3 today. There is no way. There is no way.
When we were young, we used to adjust the ground situation to see how we can adjust it. Alright, let's see if I can increase and increase the low multiplier. Just use this slider to lower the low. Let's see how much it was just now. It was 1.6. Let's adjust it to 1.8. Because this kind of large aircraft is slightly different from the small 5-inch aircraft. The range of these values may be similar to the 5-inch opportunity.
The gun cedar will be very big because our paper is relatively large now, so we must pay attention to whether it will rush up suddenly when taking off. No, it didn't rush up, and I didn't hear any special noise. Now we are ascending online. Hahahaha. I turned off the Exprim mode to prevent it from turning off. Turning it off will not allow us to fight online. Now we will turn it on 1.8
Let’s adjust it to 2.0. OK, it’s reached the top. I’m no longer allowed to adjust it online, even in expert mode. Okay, let’s take a look. If it doesn’t have any of the bad things mentioned in this concept, then let’s let it go to 2.0. Let’s try it. It feels like it can be adjusted higher. Oh, it seems to be a little bit louder, but it’s still within the controllable range. I don’t feel it’s reached the limit yet.
That's it for now, let's make it 2.0. Let's adjust the step response. The D value of the handle bar ensures that it is effective against noise... No, no, the D value ensures that its ability to brake between various interferences is better than its anti-interference ability. Let's take a look at adjusting the butt. We can see the height of the butt. What the Tune interference color says is
Once we know the height of the DJI, we have already adjusted the DJI to the highest level, right? Let’s adjust the PID in BrainP up. If the PID is adjusted once, then I will turn it off. That's it. Generally speaking, the PID ratios will not differ too much, so let's adjust this to the higher the better. Let's take a look and adjust it to close to two. Should we just go to two?
We still want to try it out first. I think the knife is fine. OK, let’s break out of this dark day. Okay, then let's try this P. The ratio between the roots and the ground is one to one. Let's try it and it can fly. I think it's almost enough to be able to fly. There is no need to adjust it too much.
It feels like you have to follow the hand obviously, because the P value will help the aircraft and help sell this error. When you move the stick, the P value will help push the plane until you want it to go this way and you want it to reach this angular speed. Oops, it feels very good. I haven’t even added FeedForward. The feel is almost good. There is a little bit of this, just like blowing this wind.
The Turping sound is a little bit loud, but overall I feel that it is flying well. We can take a look at this StepResponse and let’s do some weak operations. Take it back and take a look at this StepResponse. You can see that the D-60, DMAX810, and PID are 90. Please be careful. Do not build these on a 5-inch machine because the parameter difference between a 10-inch machine and a 5-inch machine is quite large.
We don't have FeedForward here, so we don't have absolute control. We can directly look at this step response. The new Chinese 001, alright, this time we directly look at this step response, because the noise feels quite controllable. This one is a little wide, and it feels like the P value is not high enough. Then there is a slight process, but not too much. It feels like the P value is a little high, so we can lower the P value.
Lower it a little bit because there is not too much overshoot. Then we increase the joystick gain response to 120, which is about 130. About 130, which is more in line with the hand. Not too much. A little more in line with the hand. Overall, I am quite satisfied with this Tune. The 4.3 Slider has really simplified a lot of things. You don’t have to calculate the ratio between PIDPIDs, etc. Apple Flip is also a bit afraid to Flip this 10-inch computer.
It will be very uncomfortable if it explodes. Oh no problem. The sensitivity of the 10-inch and 5-inch cameras is indeed not as good as that of the 5-inch camera. As I just said, the difference between an eagle and a sealing bird is that you can't expect an eagle to spin around in a bushy bucket, right? OK, let's adjust the field here today. Look at my mobile phone. See you in the next issue today.
Timers and timers for simulators.
