Oznog
Active Member
That does beat out the HolySpokes/SpokePOV bike wheel LED displays.
I can see how difficult this was to implement in full color, and do realtime updates into moving wheels.
Still... $12,500-$19,500?? Most expensive microcontroller project not in space or blowing up a third world country!
https://www.leftlanenews.com/2006/03/25/forget-the-spinners-pimp-rims-go-high-tech/
I'm just trying to analyze the task. I calculate the wheel turns around 10Hz at 50mph. But each LED and each color in it would need a PWM channel.
The PWM has to run pretty fast too, and the idea of using fewer drivers by selecting one LED, driving it hard, then selecting another LED and driving it (like a row/column select used in most any LED display, including multicolor ones) wouldn't work because the LEDs move so fast that it would make them appear in different places. The whole row must be driven at all times and with a PWM freq>>wheel rotation rate.
At first I thought maybe they went a Single Board Computer route, but the task actually doesn't require that much bandwidth or memory. It's all about a complex hardware interface which an SBC probably won't help with. There's a chance they had to have an ASIC made to handle all the PWM jobs but they also may have just put in a crapload of I2C/SPI PWM chips in parallel. The most PWM channels on a chip I saw on Digikey were 6-channel chips. Looks like maybe 24 tricolor LEDs, 3 per wheel, man, would take 216 channels, with output transistors and resistors!
Or maybe I'm discounting the potential to do scanning too soon. Lemme see- if we want to avoid distortion maybe you want to scan the whole row in under 1/1000th a rotation, so you'd have to drive the whole row at 10kHz. With one PWM it'd need to drive 216 channels, and if there were 8 clocks per channel (9 bit color) it would still need to run at over 17MHZ which is pretty unrealistic for a PWM or the selector switch. OK, multiple PWMs reduce the requirement, and I can see ways to do it, but I'm still in awe at the complexity of what initially seemed like a simple task.
The bummer here is they won't do anything spectacular at a standstill or low speeds, and yet they're going to be illegal to use on the road. So I'm not sure how you're going to show them off much legally.
Still another question is how they transmit the advertised real-time updates to the wheels. You'd either have to put a ring of IR receivers on the wheel or a ring of transmitters around the axle. Or go with a radio link.
I can see how difficult this was to implement in full color, and do realtime updates into moving wheels.
Still... $12,500-$19,500?? Most expensive microcontroller project not in space or blowing up a third world country!
https://www.leftlanenews.com/2006/03/25/forget-the-spinners-pimp-rims-go-high-tech/
I'm just trying to analyze the task. I calculate the wheel turns around 10Hz at 50mph. But each LED and each color in it would need a PWM channel.
The PWM has to run pretty fast too, and the idea of using fewer drivers by selecting one LED, driving it hard, then selecting another LED and driving it (like a row/column select used in most any LED display, including multicolor ones) wouldn't work because the LEDs move so fast that it would make them appear in different places. The whole row must be driven at all times and with a PWM freq>>wheel rotation rate.
At first I thought maybe they went a Single Board Computer route, but the task actually doesn't require that much bandwidth or memory. It's all about a complex hardware interface which an SBC probably won't help with. There's a chance they had to have an ASIC made to handle all the PWM jobs but they also may have just put in a crapload of I2C/SPI PWM chips in parallel. The most PWM channels on a chip I saw on Digikey were 6-channel chips. Looks like maybe 24 tricolor LEDs, 3 per wheel, man, would take 216 channels, with output transistors and resistors!
Or maybe I'm discounting the potential to do scanning too soon. Lemme see- if we want to avoid distortion maybe you want to scan the whole row in under 1/1000th a rotation, so you'd have to drive the whole row at 10kHz. With one PWM it'd need to drive 216 channels, and if there were 8 clocks per channel (9 bit color) it would still need to run at over 17MHZ which is pretty unrealistic for a PWM or the selector switch. OK, multiple PWMs reduce the requirement, and I can see ways to do it, but I'm still in awe at the complexity of what initially seemed like a simple task.
The bummer here is they won't do anything spectacular at a standstill or low speeds, and yet they're going to be illegal to use on the road. So I'm not sure how you're going to show them off much legally.
Still another question is how they transmit the advertised real-time updates to the wheels. You'd either have to put a ring of IR receivers on the wheel or a ring of transmitters around the axle. Or go with a radio link.
