NXP Semiconductors
DRAFT | D | D |
| AFT |
RAFT | RAFT AFT | |||
|
|
| DR | DR |
DLPC2917/19
ARM9 microcontrollerRAFT withDRAFTCANDRAFTand LINDRAFT
| T | DRAFT |
|
| T |
| |
| DRA |
| DRA |
| DR | ||
| F |
|
| F |
| ||
The flash memory has a |
|
| DRAF | ||||
|
| ||||||
initially via JTAG. |
| DRAFT | |||||
disabling JTAG access. Suspension of burning or erasing is not supported. |
| DRAFT | D | ||||
|
| ||||||
The key features are: |
|
| |||||
|
|
|
| DRA | |||
• | Programming by CPU via AHB |
|
|
|
| ||
|
|
|
|
|
| ||
• Programming by external programmer via JTAG |
|
|
|
|
|
| |
• | JTAG access protection |
|
|
|
|
|
|
• |
|
|
|
|
|
| |
8.1.2 Description |
|
|
|
|
|
| |
After reset flash initialization is started, which takes tinit time, see Section 12. During this |
|
|
|
| |||
initialization flash access is not possible and AHB transfers to flash are stalled, blocking |
|
|
|
| |||
the AHB bus. |
|
|
|
|
|
| |
During flash initialization the index sector is read to identify the status of the JTAG access |
|
| |||||
protection and sector security. If JTAG access protection is active the flash is not |
|
|
|
|
| ||
accessible via JTAG. ARM debug facilities are disabled to protect the |
|
|
|
|
| ||
contents against unwanted reading out externally. If sector security is active only the |
|
|
|
|
| ||
concerned sections are read. |
|
|
|
|
|
| |
Flash can be read synchronously or asynchronously to the system clock. In synchronous |
|
| |||||
operation the flash goes into standby after returning the read data. Started reads cannot |
|
|
|
| |||
be stopped, and speculative reading and dual buffering are therefore not supported. |
|
|
|
|
| ||
With asynchronous reading, transfer of the address to the flash and of read data from the |
|
| |||||
flash is done asynchronously, giving the fastest possible response time. Started reads can |
|
| |||||
be stopped, so speculative reading and dual buffering are supported. |
|
|
|
|
|
| |
Buffering is offered because the flash has a |
|
|
|
| |||
interface has only 32 bits. With buffering a buffer line holds the complete |
|
|
|
| |||
word, from which four words can be read. Without buffering every AHB data port read |
|
|
|
| |||
starts a flash read. A flash read is a slow process compared to the minimum AHB cycle |
|
|
|
| |||
time, so with buffering the average read time is reduced. This can improve system |
|
|
|
|
| ||
performance. |
|
|
|
|
|
| |
With single buffering the most recently read flash word remains available until the next |
|
|
|
| |||
flash read. When an AHB |
|
| |||||
as the previous read transfer, no new flash read is done and the read data is given without |
|
| |||||
wait cycles. |
|
|
|
|
|
| |
When an AHB |
|
|
|
|
| ||
involved in the previous read transfer, a new flash read is done and wait states are given |
|
|
|
| |||
until the new read data is available. |
|
|
|
|
|
| |
With dual buffering a secondary buffer line is used, the output of the flash being |
|
|
|
|
| ||
considered as the primary buffer. On a primary buffer hit data can be copied to the |
|
|
|
|
| ||
secondary buffer line, which allows the flash to start a speculative read of the next flash |
|
|
|
| |||
word. |
|
|
|
|
|
| |
LPC2917_19_1 | © NXP B.V. 2007. All rights reserved. |
|
|
|
Preliminary data sheet | Rev. 1.01 — 15 November 2007 | 15 of 68 |