Philips Semiconductors ation USB interface device with parallel bus PDIUSBD12 MODE O(NON-ISO MODE ENDPOINT ENDPOINT DIRECTION MAX PACKET SIZE NUMBER TRANSFER TYPE ENDPOINT TYPE Default Generic Out Generic 1 Generic Out OUT MODE 1(ISO-OUT MODE) ENDPOINT ENDPOINT TRANSFER TYPE ENDPOINT TYPE DIRECTION MAX PACKET SIZE NUMBER INDEX (BYTES) Control out OUT 0 Control Generic Out Generic OUT Generic In Generic Isochronous OUT MODE 2(ISO-IN MODE) ENDPOINT ENDPOINT TRANSFER TYPE ENDPOINT TYPE DIRECTION MAX PACKET SIZE (BYTES) 0 Control out OUT Control in Default Generic Out Gene OUT 16 1 Generic Isochronous 1284 MODE 3(ISo-IO MODE ENDPOINT ENDPOINT MAX PACKET SIZE NUMBER INDEX TRANSFER TYPE ENDPOINT TYPE DIRECTION YYTES Control Out OUT ontrol in Default 012345 Generic Out OUT 1 Generic Generic In sychronous out OUT 6 Isochronous In 1. Generic endpoint can be used either as Bulk or Interrupt endpoint 2. The main endpoint (endpoint number 2) is double-buffered to ease synchronization with the real time applications and to increase 3. DMA access is for the main endpoint (endpoint number 2)only. 4. Denotes double buffering The size shown is for a single buffer. MAIN ENDPOINT The main endpoint (endpoint number 2) is special in a few ways. It is the primary endpoint for sinking or sourcing relatively large data. As such, it implements a host of features to ease the task of transferring large data: 1. Double buffering. This allows parallel operation between USB access and local CPU access thus increasing throughput Buffer switching is handled au cally. This results in transparent buffer operatio ports for DMA (Direct Memory Access)operation. This can be interleaved with normal l/O operation to other endpoints 3. Automatic pointer handling during DMA operation. No local CPU intervention is necessary when 'crossing the buffer boundary. 4. Configurable for either isochronous transfer or non-isochronous(bulk and interrupt) transferPhilips Semiconductors Product specification USB interface device with parallel bus PDIUSBD12 1999 Jan 08 6 MODE 0 (NON-ISO MODE): ENDPOINT NUMBER ENDPOINT INDEX TRANSFER TYPE ENDPOINT TYPE DIRECTION MAX. PACKET SIZE (BYTES) 0 0 Control Out Default OUT 16 0 1 Control In Default IN 16 1 2 Generic Out Generic OUT 16 1 3 Generic In Generic IN 16 2 4 Generic Out Generic OUT 644 2 5 Generic In Generic IN 644 MODE 1 (ISO-OUT MODE): ENDPOINT NUMBER ENDPOINT INDEX TRANSFER TYPE ENDPOINT TYPE DIRECTION MAX. PACKET SIZE (BYTES) 0 0 Control Out Default OUT 16 0 1 Control In Default IN 16 1 2 Generic Out Generic OUT 16 1 3 Generic In Generic IN 16 2 4 Isochronous Out Isochronous OUT 1284 MODE 2 (ISO-IN MODE): ENDPOINT NUMBER ENDPOINT INDEX TRANSFER TYPE ENDPOINT TYPE DIRECTION MAX. PACKET SIZE (BYTES) 0 0 Control Out Default OUT 16 0 1 Control In Default IN 16 1 2 Generic Out Generic OUT 16 1 3 Generic In Generic IN 16 2 5 Isochronous In Isochronous IN 1284 MODE 3 (ISO-IO MODE): ENDPOINT NUMBER ENDPOINT INDEX TRANSFER TYPE ENDPOINT TYPE DIRECTION MAX. PACKET SIZE (BYTES) 0 0 Control Out Default OUT 16 0 1 Control In Default IN 16 1 2 Generic Out Generic OUT 16 1 3 Generic In Generic IN 16 2 4 Isochronous Out Isochronous OUT 644 2 5 Isochronous In Isochronous IN 644 NOTES: 1. Generic endpoint can be used either as Bulk or Interrupt endpoint 2. The main endpoint (endpoint number 2) is double-buffered to ease synchronization with the real time applications and to increase throughput. 3. DMA access is for the main endpoint (endpoint number 2) only. 4. Denotes double buffering. The size shown is for a single buffer. MAIN ENDPOINT The main endpoint (endpoint number 2) is special in a few ways. It is the primary endpoint for sinking or sourcing relatively large data. As such, it implements a host of features to ease the task of transferring large data: 1. Double buffering. This allows parallel operation between USB access and local CPU access thus increasing throughput. Buffer switching is handled automatically. This results in transparent buffer operation. 2. Supports for DMA (Direct Memory Access) operation. This can be interleaved with normal I/O operation to other endpoints. 3. Automatic pointer handling during DMA operation. No local CPU intervention is necessary when ‘crossing’ the buffer boundary. 4. Configurable for either isochronous transfer or non-isochronous (bulk and interrupt) transfer