This time the comparison between ICM42688 and MPU6000 focuses on Blackbox data, rather than drawing conclusions based solely on feel: after adjusting the internal filtering of the ICM42688 under test, the gyroscope noise before entering Betaflight is significantly reduced, and the D-term residual noise is also lower; however, the results only represent the body used, firmware draft and flight conditions, and cannot be directly extended to all flight controls.
Why re-evaluate ICM42688
MPU6000 has long been regarded as a low-noise gyroscope by many FPV players, but it has been discontinued and new flight controllers need reliable alternatives. The ICM42688 is a candidate due to supply and performance potential, but controversy often arises from the comparison of different firmware defaults, flight control layouts and filtering configurations. The value of the video is not to declare that a certain chip is "kingdom" under any conditions, but to show how to relate the internal filtering configuration to the Blackbox results.
Tested using Betaflight 4.6 PR #13351, which was in draft status at the time, and a set of ICM42688 settings provided by Charles Wong. This PR exposes controls related to the chip's internal registers and filtering. Automatic transcription has poor recognition of these precise names and values, so this article will not guess at CLI commands; when you need to reproduce the experiment, you should directly check the PR in the link, the current Betaflight documentation, and the parameters supported by your own firmware version.
Compare before and after configuration on the same machine body
Video establishing pre- and post-modification test conditions before applying ICM42688 internal filtering, anti-aliasing filtering, and a static Gyro low-pass setting on the Betaflight side. The film mentions a static Gyro lowpass around 250 Hz, but since the draft parameters and names may have changed, one should not copy this single number into the current version and expect the same results. A truly comparable experiment should keep the airframe, blades, motors, sampling rate, flight path and throttle range as consistent as possible.
The subjective experience of the flight from about 04:15 onwards was described as only a small difference. The tester has not flown for two weeks, and the venue is a small backyard, so he cannot maintain full throttle for a long time, so there are not enough conditions to prove the improvement in body feeling under racing load. It is important to include this limitation in the conclusion: if the two flights have different intensities, the effects of chips, filtering, status and routes cannot be distinguished simply by "smoother" or "smoother".
Why Blackbox is more convincing than its feel
The modified Blackbox shows that the original gyro noise has dropped significantly before entering Betaflight software filtering; the filtered D-term noise also remains low. For PID control, this is more specific than the waveform "looking smoother", as D-term amplifies high frequency changes, and too much residual noise can increase the risk of motor heating, harsh sound, or control output jitter.
The log before modification contains more original noise, some of which remains after Betaflight filtering and is further amplified in the D-term. This before-and-after comparison supports the conclusion that "the tested configuration is valid", but it cannot alone prove that the ICM42688 chip is better than the MPU6000 in all flight controls. Power supply noise, flight control soft installation, frame resonance, motor and propeller status will all change the spectrum; the baseline must be re-recorded after changing to another aircraft body.
Risks of Draft Firmware and Current Reproduction Methods
At the time of recording, the relevant PR has not yet been merged into the Betaflight mainline. The video clearly reminds you at about 07:30: Flashing the draft firmware is at your own risk. When reading this test today, the first step is not to find the old firmware to download, but to check whether the PR was finally merged, whether the parameters were renamed, whether the target flight control supports it, and whether the current stable version has adopted different default values.
CLI diff/dump and recoverable firmware version should be saved before modification, flashing, sensor identification and receiver/failure protection check should be completed in the propeller disassembly state, and then low-risk test flight in a safe open area. Only if the baseline and modified logs repeatedly show the same trend under similar throttle and route, it is worthy of further testing; if abnormal vibration, motor overheating or sensor error occurs, you should roll back immediately instead of continuing to fly in pursuit of a certain spectrum screenshot.
Operation steps
- Apply only settings confirmed by document
Configure the ICM42688 internal filtering and Betaflight filtering according to the current documentation for the firmware you are using; do not guess CLI commands and precise values based on the ASR or this article. Enter the setup instructions at about 02:41 in the video.
- Record unmodified baseline Blackbox
Maintain the current status of the airframe and filter, record logs using repeatable routes and throttle ranges, and record the installation status of the blades, motors and flight control; the video begins to explain the pre- and post-test arrangements at approximately 03:44.
- Repeat flight and compare original Gyro to D-term
Reproduce similar routes and throttles in a safe area, and then compare the gyroscope spectrum before filtering and the D-term noise after filtering. The video shows the modified Blackbox results at about 05:16.
- Finally check the current PR and firmware documents
About 07:30 in the video, it is emphasized that the draft used at that time had not yet been merged into the main line. Before jumping to conclusions, confirm the final status of PR #13351, current parameter names and target board support, and ensure you have restoreable CLI and firmware backups.
FAQ
Does this test prove that ICM42688 is definitely better than MPU6000?
No. It proves that the noise of the tested aircraft is significantly improved under specific draft firmware and settings; the log needs to be re-recorded after changes in the chip, flight control layout, rack and firmware.
Can I directly copy the CLI commands from the video?
Not recommended. A draft PR was used during recording, and transcribes cannot reliably restore parameter names. The final status of PR #13351 and the current Betaflight documentation should be reviewed and only commands explicitly supported by the native firmware should be used.
Why watch both original Gyro and D-term?
The original Gyro can display the sensor and body noise before entering Betaflight filtering; D-term will amplify high-frequency changes, and whether the residual noise after filtering enters D-term can better reflect the control and motor burden.
Does the small difference in subjective feel mean that the settings are invalid?
Not necessarily. The field and flight status of the video limited somatosensory comparisons, while the Blackbox still showed noise reduction. The test should be repeated multiple times under matching conditions, and the motor temperature and flight performance should be checked at the same time.
Full timeline transcript
Transcripts are arranged according to video time, making it easy to quickly locate the explanation content. Transcript language: Simplified Chinese.
ok First of all, let me tell you why I want to try this. The reason for this is that I saw Charles Charles Wong, who seems to be a pilot in Texas and a very fast pilot. Then he posted a post and said about this ICM42688. What did he say? New Champion over MPU6000 means that it will be a better gyroscope than MPU6000.
Friends who fly at short speeds believe that the cost is the same. Everyone knows that the NPU6000 is the gyroscope with the smallest noise in this flight control in the past ten years. However, this thing should have been discontinued, and it is unlikely to buy too many new NPU6000s in the future. In the future, everyone will definitely use other gyroscopes. And I have used the SEM4288.
More than a year, more than a year, because after I switched to iFlight's flying holes, iFlight's flying holes all used ICM-4268, so I was interested in this, this, this iron increase, and then he said what he said, first of all, you have to brush this PR. This PR is a PR in Betaflight 4.6. We know that Betaflight is still 4.5 4.6. This is a new feature and will be included in 4.6.
Then we can click on this PR to see what it is. This is a PR from a very senior Betaflight Dev called Karate Broad. Let’s see what it does. Configure AA and it should be added to this UI Filter, right? Expose SCM-388 register in the CLI means some internal registers of SCM-388. It exposes it and you can configure it directly. SCM-388
Then several new parameters were added here, GeraldDoubt, GeraldDoubtSquare, and GeraldBiftShift. These things together seem to be several configurations of the frequency of the green wave. I will give you a closer look. Then there are several parameters of the UI green wave. In that post, Charles said that he wanted to change these parameters. I haven’t looked at this carefully, but if you are interested, you can check it out for yourself.
Back in Pose, first adjust the AFF filter to 2029Hz, shift it, and then increase the UI filter a little higher, Ui15, OR-VanRing, and then configure a static gyroscope low-pass green wave of 250Hz, which is almost the same as MPU6000.
In Fueler, pay attention to designing a low force wave of 250. Of course, those commands must be run. OK, next, let’s fly it first without swiping it, and then compare it. After swiping it, from the aspect of feel, data, and black box data, let’s see if there is any progress, whether it has really gotten better, and see if it is correct to swipe it out. OK, I just lost a few batteries before and after brushing the shoes.
Before we look at this data, let’s talk about whether there is any difference in the feel. Personally, I don’t feel that there is a particularly big improvement, but is it there? There must be some, but I don’t feel it is particularly obvious. There may be two reasons. One is because I haven’t reached speed for more than two weeks, so this feeling may not be fully back yet, and I may not be so sensitive to subtle differences.
There is another possible reason because my backyard is useless. This backyard is relatively small and there is no way to deal with the noise at full throttle. In fact, in many cases, this is more obvious at full throttle. So for the above two reasons, I may not feel the big difference, but the data can often give us more accurate answers. So I will practice my aircraft here and let's take a look at the black box data. Now that I have refreshed it, let's take a look at the refresh directly.
Because this PR mainly improves the noise, so we directly look at its noise feeling. You can see that this is the raw data that comes in from Betaflight before filtering. Before filtering, it is the raw data that comes in from Gyro Silver. Because this PR can directly change the internal calculator of Gyro Silver, which is the Gyro Silver’s own.
Tutorial is set with filtering, so when Betaflight sees this signal, it has actually been filtered internally. As you can see here, even before Betaflight filtering, the noise is already very small, as you can see. Then after filtering below Betaflight, there is almost no noise. Then to the low-quality noise, you see that the low-quality noise is almost very small. You can take a look first. This is the noise after the improvement after flashing.
Let’s take another look at what the cause looked like without this improvement. It’s better to find a larger section. In fact, it feels more obvious, right? Look at this, the original toro signal, the cause is much higher than just now. Then Betaflight waved it, eliminated a lot of the cause, but there are still some left. These are amplified on the low side. You can see that there are obvious two ancient packets on the low side. It is better than what we just did.
Will the noise on the D value be much higher after this PR? And I don’t have much full pylorus action here. There is almost no full pylorus action. I tested it on the 6S and there was almost no way to fill the pylorus. So I believe the difference will be more obvious when there are more full pylorus on this real track. So here I plan to flash all my aircraft to this new version.
But everyone must pay attention when brushing up, because this PR has not been submitted to the main trunk and is still a draft, so be sure to pay attention to safety. You are responsible for your own safety. OK, that’s all I have to say today. If you think this video is helpful to you, please don’t forget to click three times. I haven’t talked about this PID and this black core analysis and black core data analysis for a long time. If the reverse of this video is okay,
I would like to share with you some of my experience in the past two or three years. So if you want to know more, please like and forward it. Thanks. This is QDRK, see you next time.
