The basic idea is good, but not novel. You will not get anywhere near 20-bit performance this way. Measuring any physical quantity (except time) to within one part in a million hardly ever happens in practice. 20 bits is the equivalent of a breath of air compared to the weight of a car. Consider that the placement of the slots will need to be accurate to a millionth of a circle, (about 1 second of arc) or you will get missing codes. You will not get a sine wave so distortion-free that you can interpolate it to thousands of different levels. You will also need to deal with the effects of physical optics. As an optical aperture closes, the light transmitted in a given geometric path becomes non-monotonic due to optical diffraction. Your light detector will have shot noise that will become significant as the rotation speed increases. Stray light needs to be eliminated. I suggest that you try to build a 20-bit rotation encoder, put it on the same shaft as a commercial 20-bit encoder (like one from Renishaw) and present the results of your comparison in a follow-up video. If you can get the first 14 or 15 bits to agree, you will have done admirably.
Thank you for watching and commenting. This video basically explains the principle of making a high-resolution encoder as you mentioned. The encoder shown now is an encoder with 24-bit resolution by interpolating 2048ppr + 13bit. In order to compensate for the optical distortion you mentioned, an interpolator with a built-in real-time correction function was used. (The interpolator uses the function shown in the video.) Next, I will explain by making a video demonstrating an encoder with 20bit or more using a photo IC and a high resolution interpolator. I will demonstrate and explain the part where the optical signal is distorted and how to compensate for it. Your comments were a great help in making the video.
Good information. Thanks.
Thanks for the feedback!
The basic idea is good, but not novel. You will not get anywhere near 20-bit performance this way. Measuring any physical quantity (except time) to within one part in a million hardly ever happens in practice. 20 bits is the equivalent of a breath of air compared to the weight of a car. Consider that the placement of the slots will need to be accurate to a millionth of a circle, (about 1 second of arc) or you will get missing codes. You will not get a sine wave so distortion-free that you can interpolate it to thousands of different levels. You will also need to deal with the effects of physical optics. As an optical aperture closes, the light transmitted in a given geometric path becomes non-monotonic due to optical diffraction. Your light detector will have shot noise that will become significant as the rotation speed increases. Stray light needs to be eliminated. I suggest that you try to build a 20-bit rotation encoder, put it on the same shaft as a commercial 20-bit encoder (like one from Renishaw) and present the results of your comparison in a follow-up video. If you can get the first 14 or 15 bits to agree, you will have done admirably.
Thank you for watching and commenting.
This video basically explains the principle of making a high-resolution encoder as you mentioned.
The encoder shown now is an encoder with 24-bit resolution by interpolating 2048ppr + 13bit.
In order to compensate for the optical distortion you mentioned, an interpolator with a built-in real-time correction function was used. (The interpolator uses the function shown in the video.)
Next, I will explain by making a video demonstrating an encoder with 20bit or more using a photo IC and a high resolution interpolator.
I will demonstrate and explain the part where the optical signal is distorted and how to compensate for it.
Your comments were a great help in making the video.