Known issues in PBP 2.60A

• 64-bit system requirements

Devices affected: All Enhanced 14-Bit parts (16F1937, 1827, etc.) when using MPASM 5.36
Commands affected: All commands
Date encountered: 07/2010
Error class: Critical (compile errors in most cases, but some silent bugs)

Microchip's release of MPASM 5.36 changes functionality of the assembly operator "high" for the enhanced 14-bit MCUs. This thouroughly wrecked our command and macro libraries for all families of parts that share the enhanced 14-bit instruction set.

Fix:

We released a patch within a few days of being notified of this issue. All users should use the patch to update from 2.60 to 2.60A.

Devices affected: 18F2420/2520/4420/4520 Family
Commands affected: WRITECODE
Date encountered: 03/2010
Error class: Apparent (no error, but consistently fails initial testing)

The WRITECODE command will not write to all locations of code space because the block size is declared incorrectly for this family. Write operations will seem to ignore portions of code space, while writing successfully to other locations.

Workaround:

In the PBP install folder, locate the .INC file for the device that you are compiling for (18F2420.INC, 18F2520.INC, 18F4420.INC, or 18F4520.INC). Open the file in a text editor and find the line "BLOCK_SIZE EQU 64". Change the setting to 32: "BLOCK_SIZE EQU 32".

Devices affected: PIC18F46J11 Family
Commands affected:  ADCIN
Date encountered: 12/2009
Error class: Apparent (no error, but consistently fails initial testing)

The ADCIN command always doubles the channel parameter.  AN0 reads correctly, channel-1 reads AN2, channel-2 reads AN4, etc.

DEFINE ADC_CLOCK writes to the wrong register, breaking all ADCIN commands.  Even if you omit the define, it still writes a default value to the wrong register.

Workaround:

For the DEFINE ADC_CLOCK issue, always write "DEFINE ADC_CLOCK  0" and then set the actual clock selection by writing to the ADCON1 register using direct access:

DEFINE ADC_CLOCK  0    ' safe setting
ADCON1 = %10000011 ' set ADC clock 3, right justify

For the channel select, you could read even-number channels with a half-value setting.  The better solution is to use a manual method by manipulating the registers directly to perform the conversion:

ADCON0 = %00000101      ' set to channel-1
PAUSEUS 50                   ' pause to settle

ADCON0.1 = 1                 ' start conversion
DO WHILE ADCON0.1 = 1   ' wait for conversion
LOOP

adval.highbyte = ADRESH   ' store result
adval.lowbyte = ADRESL

Devices affected: standard 14-bit instruction set (PIC16, PIC12)
Commands affected:  SERIN/SEROUT
Date encountered: 10/2009
Error class: Apparent (no error, but consistently fails initial testing)

Under some circumstances, SERIN and SEROUT commands are subject to baud rate errors that are beyond those allowed in the RS-232 specification.  This will mostly affect applications where the baud rate is 9600 and the system clock is 4MHz or less.  Symptoms mostly occur on receive (SERIN), and manifest as garbage or incorrect data being returned.

Workaround:

The easiest, safest fix is to switch to an alternative serial command (SERIN2/SEROUT2, DEBUGIN/DEBUG, or HSERIN/HSEROUT).

If this isn't possible due to the constraints of your design, contact support.  There is a library fix that can be implemented immediately, but it will interfere with the official patch when released.  A patch is in the works, but we are waiting on a few other improvements that we want to include.

Devices affected: 16F1826, 16F1827
Date encountered: 08/2009
Error class: safe (always reported)

There are invalid port aliases in the PBP header file for these parts.  The following assembly errors will be generated:

	Symbol not previously defined (PORTC)
	Symbol not previously defined (TRISC)

Workaround:

Edit the files 16F1826.BAS and 16F1827.BAS (found in the PBP install folder) as follows:

	PORTL   VAR     PORTB
	PORTH   VAR     PORTA  ' PORTC invalid, change to PORTA
	TRISL   VAR     TRISB
	TRISH   VAR     TRISA  ' TRISC invalid, change to TRISA

Command affected: WRITE
Date encountered: 07/2009
Error class:  safe (always reported)

When a WRITE command is attempted with a word variable in the value position, but the WORD modifier is omitted.  Also occurs when a calculation is used in the value position, as this may force a word temp variable to be used.  (Correct operation is for the command to write only byte-0 of the variable when the WORD modifier is omitted.)

	wordvar    VAR    WORD

	WRITE 0, wordvar

One of the following assembly errors will sometimes be generated:

	... symbol not previously defined (WRITE)

	... undefined symbol 'write'

Workaround:

Add the following define at the top of the code.  This define will have no effect on unaffected code, nor will it increase code space usage.

	DEFINE  WRITE_USED  1

Previous version 2.50C has no bug reports, but is incompatible with MPLAB versions 8.20 and later.  Versions 2.50, 2.50A, and 2.50B each have minor issues that I won't list here as they have been fixed in the downloadable patch.  Use the patch!

Windows 64-bit compatibility requirements:

Users of 64-bit systems must instruct PBP to use the Microchip Assembler, MPASMWIN.  In MicroCode Studio, this is a simple check box setting in the Compile and Program Options dialog.  On the Compiler tab, check the box labeled "Use MPASM".  MPLAB must be installed.  It can be downloaded from the Microchip site or installed from your PBP install CD.