RCM2110 issue

[spth]

I can't get my RCM2110 board to work (it has a Rabbit 2000C, just like my RCM2000, RCM2020, RCM2200, and RCM2210 boards that work fine using the same programming cable):

philipp@notebook6:~/OpenRabbit$ src/openrabbitfu --verbose --verbose --run --serialout examples/hello/hello-RCM2200.ihx /dev/ttyUSB3
Reset Rabbit.
Sending initial loader.
set baudrate to 2400
Sending 266 initial loader triplets.
set baudrate to 57600
Secondary loader format detected as Dynamic C 9
CPU: 0x0003 (Rabbit 2000C)
Sending secondary loader.
sending 5517 secondary loader bytes.
Negotiating baudrate.
set baudrate to 460800
set baudrate to 57600
warning: received NAK for subtype 0x46
set baudrate to 230400
reading data frame
parsed 140 bytes
CPU:  0x0003 (Rabbit 2000C)
Freq: 11059200
sectorSize: 0x1000
numSectors: 0x0040
flashSize:  0x0040
Flash speed: 0 ns
RAM speed: 0 ns
Due to its file extension, "examples/hello/hello-RCM2200.ihx" is considered to be in Intel hex format (of up to 256 KB size).
sending examples/hello/hello-RCM2200.ihx... done(bps: in=0, out=0)                   
Rebooting and running installed code.
set baudrate to 2400
Reset Rabbit.

Maybe there is an issue with the board? But maybe something OpenRabbit does is so close to the tolerances that it happens to fail with this one board, and this is still an OpenRabbit issue?

[tomlogic]

Definitely odd. Any chance you can test the hardware against a Dynamic C 9 installation to confirm that it's an issue with OpenRabbit and not the hardware?

Some differences between RCM2200 and RCM2110:

• Ethernet interface slightly different, but you probably aren't doing anything with those pins in your sample code.
• RCM2200 uses port D for serial port B, but you're probably using serial port A in your tests.

Do you have a simpler test program that just flashes an LED using the 32kHz clock as a timing source? That could eliminate all of your BIOS/startup code as being a possible cause.

[tomlogic]

Here are some macros defined in the compiler for the RCM2110 and RCM2200 -- they're identical, so they should both work with common startup code.

-----
Compiling RCM2110 "11.0592" "22MHz RCM2110, 128K SRAM, 256K Flash"
 CLOCK_DOUBLED=1
 ENABLE_SPREADER=1
 FLASH_WSTATES=0xc0
 MB0CR_SETTING=0xc0 | 0x00 | (0<<4) | (0<<5)
 MB1CR_SETTING=0xc0 | 0x02 | (0<<4) | (0<<5)
 MB2CR_SETTING=0xc0 | 0x05 | (0<<4) | (0<<5)
 MB3CR_SETTING=0xc0 | 0x05 | (0<<4) | (0<<5)
 NUM_FLASH_WAITST=0
 NUM_RAM2_WAITST=0
 NUM_RAM_WAITST=0
 RAM2_WSTATES=0xc0
 RAM_WSTATES=0xc0
 _DC_CLK_DBL_ undefined
 _FREQ_DIV_19200_=18
 _BOARD_TYPE_=RCM2110
-----
Compiling RCM2200 "11.0592" "22MHz RCM2200, 128K SRAM, 256K Flash"
 CLOCK_DOUBLED=1
 ENABLE_SPREADER=1
 FLASH_WSTATES=0xc0
 MB0CR_SETTING=0xc0 | 0x00 | (0<<4) | (0<<5)
 MB1CR_SETTING=0xc0 | 0x02 | (0<<4) | (0<<5)
 MB2CR_SETTING=0xc0 | 0x05 | (0<<4) | (0<<5)
 MB3CR_SETTING=0xc0 | 0x05 | (0<<4) | (0<<5)
 NUM_FLASH_WAITST=0
 NUM_RAM2_WAITST=0
 NUM_RAM_WAITST=0
 RAM2_WSTATES=0xc0
 RAM_WSTATES=0xc0
 _DC_CLK_DBL_ undefined
 _FREQ_DIV_19200_=18
 _BOARD_TYPE_=RCM2200
-----

[spth]

The OpenRabbit examples also use the same values for these two boards (see https://github.com/spth/OpenRabbit/blob/trunk/examples/targetconfigurations.h):

#define SERIAL_DIVIDER_38400 18
#define CLOCK_DOUBLER 0x07 // Clock doubler for 11.0592 MHz base
#define MB0CR_VALUE 0xc8   // Flash - 0 wait states (for 45 ns Flash @ 22.12 MHz) with write-protection
#define MB2CR_VALUE 0xc5   // RAM - 0 wait states (for 55 ns RAM @ 22.12 MHz)

AFAIK, the examples do not enable the spreader, and do not set MC1CR nor MC3CR (even when the latter are defined in targetconfigurations.h for boards that have more than 256K of RAM or Flash).

[spth]

Currently, the simplest example program in OpenRabbit is the "Hello, world!" that outputs that line repeatedly on serial port A @ 38400 baud. The code is quite simple (https://github.com/spth/OpenRabbit/blob/trunk/examples/hello/hello.c), but of course compiling it will pull in some startup and library code from SDCC.

P.S.: It would make sense to add an even simpler LED blink example toOpenrabbit. Is there some common LED pin location among RCM?
Writing it in C or assembler both would have different advantages: written in C it would show that all the startup stuff works, so if it works, but not "Hello, world!", we'd know that the problem is exactly about the serial transmission. Written in assembler it could help tell us that OpenRabbit is working for flashing and starting the program, even when there's an issue with the SDCC startup code.