Tang Nano 9K

The Tang Nano 9K can support various video resolutions depending on the design complexity and frame rate: 1) Standard resolutions - 640x480 (VGA), 800x600 (SVGA), and 1024x768 (XGA) are easily achievable. 2) HD resolutions - 1280x720 (720p) at 60fps is well-supported for many applications. 3) Full HD - 1920x1080 (1080p) is possible but may be limited to lower frame rates or simpler processing depending on design complexity. 4) Camera input - depends on the camera module used, with OV5640 supporting up to 5MP. The limiting factors are FPGA resources (8,640 LUTs), SDRAM bandwidth (64Mbits), and HDMI interface capabilities. For video processing applications, 720p60 is a good target that leaves resources for processing algorithms.

Target 720p for most applications. 1080p is possible for simple passthrough or limited processing. Contact us for video processing reference designs.

Yes, the Tang Nano 9K can implement machine learning inference for small to medium neural networks: 1) Network size - can implement networks like simple CNNs for image classification on small input sizes (e.g., 32x32 or 64x64). 2) Quantization - INT8 or lower precision quantization is recommended to fit within FPGA resources. 3) Performance - expect inference times of milliseconds to tens of milliseconds depending on network complexity. 4) Applications - suitable for simple object detection, gesture recognition, or classification tasks. 5) Tools - use Gowin's AI IP cores or implement custom accelerators in the FPGA fabric. For more demanding ML applications, consider moving to larger Arora FPGAs like the GW2A series. The Tang Nano 9K is excellent for learning and prototyping ML on FPGAs before scaling up.

Start with small networks and quantization. Contact us for AI implementation guidance and reference designs for the Tang Nano 9K.

Using the camera interface on Tang Nano 9K involves several steps: 1) Camera module - connect a compatible camera like OV2640 or OV5640 to the camera connector. 2) Interface protocol - implement the camera interface protocol (typically DVP parallel or MIPI CSI-2) in the FPGA. 3) Image capture - design logic to capture pixel data from the camera. 4) Frame buffering - store captured frames in the onboard SDRAM. 5) Processing - implement your image processing algorithm. 6) Output - display processed video via HDMI or store to SD card. Gowin provides camera interface IP cores that simplify this process. The OV5640 is recommended for higher resolution (5MP), while OV2640 is sufficient for lower resolution applications. Check the pinout and voltage levels when connecting the camera module.

Use Gowin's camera interface IP to simplify development. Start with lower resolutions and simple capture before adding processing.

The Micro SD card slot on Tang Nano 9K can be used for various purposes: 1) Configuration files - store multiple FPGA bitstreams and load them on boot. 2) Data logging - record sensor data or processing results. 3) Image storage - save captured images from the camera. 4) Video recording - store compressed or uncompressed video streams. 5) Firmware updates - field upgradable designs by loading new bitstreams from SD card. 6) Lookup tables - store large data tables for processing algorithms. The FPGA design needs to implement an SD card controller (SPI or SD mode) to access the card. Gowin provides SD card interface IP cores. Standard FAT32 formatting is typically used for compatibility. Card capacity support depends on the controller implementation - most designs support up to 32GB cards.

Implement SD card controller in your design. Use for configuration, data logging, or storage as needed for your application.

The Tang Nano 9K can be used for professional development, with some considerations: 1) Prototyping - excellent for proof-of-concept and algorithm development before custom PCB design. 2) Small production runs - suitable for low-volume products where custom PCB cost isn't justified. 3) Evaluation - great for evaluating Gowin FPGAs for potential use in products. 4) Limitations - onboard peripherals are fixed, power supply is integrated, and form factor may not suit all applications. For professional development, the board provides a cost-effective platform for: algorithm validation, IP core development, performance benchmarking, and customer demonstrations. Many professional developers use Tang Nano 9K for initial development, then design custom PCBs for production. The skills and IP developed on Tang Nano transfer directly to custom designs.

Use Tang Nano 9K for prototyping and evaluation. Design custom PCBs for production volumes. Contact us for custom design support.