5iMX宗旨:分享遥控模型兴趣爱好

5iMX.com 我爱模型 玩家论坛 ——专业遥控模型和无人机玩家论坛(玩模型就上我爱模型,创始于2003年)
12
返回列表 发新帖
楼主: roland-c
打印 上一主题 下一主题

统计一下自稳会歪的飞控

[复制链接]
21
 楼主| 发表于 2018-1-31 14:15 | 只看该作者
felix727144 发表于 2018-1-31 14:00
没有用过naze32, 和 f3是用的同一个固件吗,硬件有什么区别?
看了下,naze32也是cf固件,硬件上

固件是同一个(不同的编译目标),imu的采样速度不一样(可调)

欢迎继续阅读楼主其他信息

22
 楼主| 发表于 2018-1-31 14:22 | 只看该作者
felix727144 发表于 2018-1-31 14:00
没有用过naze32, 和 f3是用的同一个固件吗,硬件有什么区别?
看了下,naze32也是cf固件,硬件上

问题果然以传感器为主...NAZE MPU6500
F3 MPU6050
F3EVO MPU9250


23
发表于 2018-1-31 14:25 | 只看该作者
本帖最后由 felix727144 于 2018-1-31 14:30 编辑
roland-c 发表于 2018-1-31 14:22
问题果然以传感器为主...NAZE MPU6500
F3 MPU6050
F3EVO MPU9250

但是这几个传感器参数貌似差不多,所以还得需要多多测试,这个活不轻松。。6050和6000性能基本一致,6050多了个i2c但是采样频率不能达到8k,9250多了地磁,陀螺仪性能似乎强大了些。
https://www.invensense.com/produ ... ng/9-axis/mpu-9250/
https://blog.dronetrest.com/iner ... pu6500-vs-icm20602/

MPU6000
Right now the MPU6000 is the most popular IMU sensor used on just about all the best flight controllers. Reason being is that it it can sample the gyro at 8kHz via an SPI bus, and also has great vibration tolerance. One thing to watch out for is to make sure this sensor is connected via SPI to the processor to get full use of the high speed updates.
MPU6050
The MPU6050 is exactly the same as the MPU6000 above, however it talks on I2C protocol, it is not often used on flight controllers because I2C is just too slow for this sort of thing and cant read the gyro at 8Khz. I would stay away from any flight controllers using this sensor.

MPU-9250 Nine-Axis (Gyro + Accelerometer + Compass) MEMS MotionTracking™ Device

The MPU-9250 is the company’s second generation 9-axis Motion Processing Unit™ for smartphones, tablets, wearable sensors, and other consumer markets. The MPU-9250, delivered in a 3x3x1mm QFN package, is the world’s smallest 9-axis MotionTracking device and incorporates the latest InvenSense design innovations, enabling dramatically reduced chip size and power consumption, while at the same time improving performance and cost.

The MPU-9250 MotionTracking device sets a new benchmark for 9-axis performance with power consumption only 9.3μA and a size that is 44% smaller than the company’s first-generation device. Gyro noise performance is 3x better, and compass full scale range is over 4x better than competitive offerings.
The MPU-9250 is a System in Package (SiP) that combines two chips: the MPU-6500, which contains a 3-axis gyroscope, a 3-axis accelerometer, and an onboard Digital Motion Processor™ (DMP™) capable of processing complex MotionFusion algorithms; and the AK8963, the market leading 3-axis digital compass. The MPU-9250 supports InvenSense’s market proven MotionFusion. A single design can support the MPU-9250 or MPU-6500, providing customers the flexibility to support either device in different product SKUs.

Improvements include supporting the accelerometer low power mode with as little as 6.4μA of and it provides improved compass data resolution of 16-bits (0.15 μT per LSB). The full scale measurement range of ±4800μT helps alleviate compass placement challenges on complex pcb’s

The MPU-9250 software drivers are fully compliant with Google’s Android 4.1 Jelly Bean release, and support new low-power DMP capabilities that offload the host processor to reduce power consumption and simplify application development. The MPU-9250 includes MotionFusion and run-time calibration firmware that enables consumer electronics manufacturers to commercialize cost effective motion-based functionality.




24
 楼主| 发表于 2018-1-31 17:09 | 只看该作者
felix727144 发表于 2018-1-31 14:25
但是这几个传感器参数貌似差不多,所以还得需要多多测试,这个活不轻松。。6050和6000性能基本一致,6050 ...

渐渐觉得这个问题变成玄学了。。。
25
发表于 2018-1-31 17:23 | 只看该作者
roland-c 发表于 2018-1-31 17:09
渐渐觉得这个问题变成玄学了。。。

而且只有f4或者f7的飞控才支持32k采样,一般可能就是8k采样或者2k。
所以我觉得不歪可能是凑巧。这个需要大量测试。或者彻底搞懂姿态算法。

26
 楼主| 发表于 2018-1-31 18:15 | 只看该作者
felix727144 发表于 2018-1-31 14:00
没有用过naze32, 和 f3是用的同一个固件吗,硬件有什么区别?
看了下,naze32也是cf固件,硬件上

基本上确定是三轴传感器的问题。震动下,六轴传感器不能准确读数,依赖于三轴积分。如果积分误差太大,飞机就会歪

27
发表于 2018-1-31 21:39 | 只看该作者
关注,曾经也遇到过同样的问题。

怀疑与程序有关。
APM 3.2.1  貌似就很稳定,飞多暴力都不歪。

28
发表于 2018-1-31 22:46 | 只看该作者
roland-c 发表于 2018-1-31 18:15
基本上确定是三轴传感器的问题。震动下,六轴传感器不能准确读数,依赖于三轴积分。如果积分误差太大,飞 ...

其实我想表达的是三轴陀螺是有局限的,原理上就有缺陷,所以只能通过软件修正

29
发表于 2018-1-31 23:04 | 只看该作者
同一台機拆過有可能就會歪
估計某頻率的震動會耦合放大
導致加速度計讀取不能
機體有任何變動都有可能改變了耦合諧震的頻段
我們這種土法跟本抓不到的
要討論不同設備的搭配就更難了

塑膠機軟軟的會吸震
我常玩的華牌150飛六七分鐘後開自穩都不會飄
其他機架越硬的只要摔了就會越不穩
這些機子的自穩都沒在用的
慣性控制基本上就是會累積誤差
要能可靠的就地面座標系統控制還是得靠另外的參考點
30
发表于 2018-2-1 01:09 | 只看该作者
F3永远不歪真的很神奇。。。CC3D还真从来没歪过
您需要登录后才可以回帖 登录 | 我要加入

本版积分规则

关闭

【站内推荐】上一条 /2 下一条

快速回复 返回顶部 返回列表