Differences Between CF1 & CF2

The 68LK332-based PERCF2I5 will be a near drop-in alternate to the CF1, and will work with the same CF1 programming tools and offer additional features such as the ability to operate either Type I or Type II CF cards. The PERCF2I5 will have the 512KB of SRAM and will sell for $395: the same price as the PERCF1C.

A slight height increase to accommodate larger TypeII CompactFlash cards will be the only physical difference between the CF1 and CF2. The standard operating temperature is -40C to +85C. The pin-out remains the same, but the CF2 will bring out 16 TPU (Time Processor Unit) signals where the CF1 brought CTM (Counter Timer Module) signals. Some recoding will be required for functions more complex than simple digital I/O, but the TPU also brings some exciting new capabilities. Let us know if you want to be included in early CF2 information e-mails.

Key electrical differences:

1. Pin C5 (was a no connect of CF1) is now VBAK, which is the switched output of VREG and VBBK for use by external circuitry that needs power at shutdown or off.

2. Pin C36 (CTM31L on the CF1) is now a don't connect.

3. Pins C22 to C37 (CTDx and CTSx) are now TPU1 through TPU15 (in that order).

4. Pin B20 is still CLKIN, but requires a zero ohm resistor be moved from one set of pads to another.

5. The 68332 and the TI MSP430 communicate over SPI and so control lines SCK, MISO, and MOSI are no longer available as simple digital I/O lines - even for systems that do not have external SPI devices. QPB slot 15, which was never a good choice for activating an SPI device since it drives all PCS lines high and this is the idle state, is now reserved for onboard use between the 68332 and the TI MSP430.

6. Three of onboard LEDs (4 LEDs in 2 bi-color parts) connect to the MSP430 instead of the 68332 so timing will be a bit different. The LEDs however are placed on the outer edges of the top of the board for much better visibility.

7. The MSP430 takes over the key role of battery backed RTC from the built-in RTC module in the CF1's 68338. The BIOS delay, elapsed, and countdown timers all relied on very fast access to time information and the SPI link to the MSP430 can't hack it. Instead the CF2 takes over TPU channel 0 running a custom Pseudo-RTC function that synchronizes to the MSP430 at startup or after LPSTOP and SHUTDOWN and gives similar fast access.

8. The CF1's separate 32,768kHz RTC crystal and 40kHz CPU crystal have been replaced by a single 40kHz crystal that runs the MSP430 and gets fed back into the 68332. There will no longer be any synchronization drift issues between PIT inspired sampling intervals and time reported by the RTC.

9. Type II CF socket can support IBM Microdrive with external power.

Key software differences:

1. CTM specific functions don't exist, so any BIOS calls beginning with CTMxxxx() will need to be replaced with workalikes from the TPU. Any timing functions you've written using specific features of the CTM will need to be rewritten to use equivalent functions from the TPU feature set. All simple digital I/O that references pins by number (PinSet(25), PIORead(33), etc.) will work essentially the same except for pin-36.

2. Except for simple digital I/O settings, TPU functionality will not automatically be re-enabled after SUSPEND. This probably means you will have to call your TPU function initialization routines after suspending. I'm looking into an unobtrusive registry system to restore the automatic operation.

3. LPSTOP is now a PicoDOS call instead of an inline macro to support TPU resynchronization with the MSP430 RTC.

4. QPB slot 15, which was never a good choice for activating an SPI device since it drives all PCS lines high and this is the idle state, is now reserved for onboard use between the 68332 and the TI MSP430.