UART not functioning

The first thing I would try would be to run your modified UART1 application on the EVM. Use the oscilloscope to check if UART1_TXD is toggling. If not then we can check your software setup more carefully. If so then there is some board dependency and we can look closely at the schematics.

The pinmuxing looks correct. In CCS check to make sure the pinmux settings actually took effect by reading address 0x01C1414C, which should be 0x11111122.

Jeff

HI Jeff,

Thank you so much for your advice.

Instead of trying my modified UART1 application on the EVM, I tested my application on UART0 and UART2 on the custom board (C6747 DSP). Strangely, only UART2 is working (I see toggling TxD). Personally I don't think the hardware for UART0 and UART1 are faulty.

Jeff, is there something I missed totally? Why UART2 is working (coincidentally the same UART I tested my application on EVM and worked) and UART0 and UART1 are not? I checked my pinmux again and PSC, and don't see any thing suspicious. I can't test on the EVM now because it is occupied by another engineer.

I really need your professional help. I am really lost here. I believe it is something to do with the software.

Thanks.

Chee-Beng

Try printing out the UART1 registers at 0x01D0C000. Are they all configured properly?

When you changed your code I assume you did a find/replace on UART2 to UART1. Is it possible you are forgetting to change a few of the settings over when you do this? I can't say without seeing your code.

If possible you can send the code that works on the EVM and I can try modifying it for UART1 and seeing how it goes.

Jeff

Thanks, Jeff.

I will try to read the UART1 registers. Before that, I would like to know am I right to say that the only things to change when changing UART are:

1. The Device ID in the UART device driver in the TCF.

2. The PINMUX.

3. The PSC

Thanks.

Chee-Beng

That seems to be it.

Again, if you can send over the code for uart2 that works on the EVM I can try it out over here.

Jeff

Hi Jeff,

I attached herewith the workings codes for Uart2. I forgot to inform you that the EVM I used was C6748 DSP, but my custom board uses C6747. Hence, the codes I attached are for the EVM C6748 DSP.

Thanks for your help. Looking forward to receiving your good news.

Chee-Beng

Unfortunately not all the files are there in the .zip. It says I'm missing \src\persistable\SpiFlash.cpp

In the meantime, I would suggest you verify your board by setting the UART1 pins as GPIO outputs and seeing if you can toggle them manually using the memory window of CCS. It might be an issue with the traces on the board.

Also the C6747 and C6748 have different memory maps and pinmux settings. Have you verified that you ported them all over correctly?

The best test is still to try and run the UART1 code on the EVM. That will eliminate many variables, so see if you can get the board back to try that test.

Jeff

Hi Jeff,

Sorry for missing out some files. I resend all the files one more time here. Hope you can help me locate where are the problems.

For you information, I had tested the UART1 pin as GPIO and it works. So I ruled out physical link problem.

To answer your question on porting over from C6748 to C6747, I am aware of the differences. Hence, I started off using the C6747 EVM GEL and platform to work with instead of the C6748EVM's. By the way, I had also sent you the pinmux previous for comment. You, too, think that the pinmux configuration is correct.

Regarding about memory maps, does memory allocation such as SDRAM, IRAM affect UART1 functionality in this case? I know I still haven't sorted out my SDRAM problem yet. I am forcing everything to use IRAM now just to test out my custom board. I will come back to understand more about memory later. Can I leave them later?

Thanks.

Chee-Beng

Hi Jeff,

I had printed out UART2 and UART1 memory for comparison Hopefully can spot any clues to my problem to UART1. The memory is printed just before the first GIO_write() statement, which somehow goes into a infinite loop when I use UART1 but not UART2. This GIO_write() is the first UART statement that reset the GPS receiver.

The memory location highlighted in red are the differences between UART1 and UART2. However, I cannot understand the significance. Can you advice?

For Uart 2 (working uart) at 0x01D0D000:

Offset

00h 0x00000000 0x00000004 0x000000C1 0x00000003

10h 0x00000000 0x00000060 0x00000000 0x00000000

20h 0x000000D0 0x00000003 0x44141102 0x00000000

30h 0x0000E003 0x00000000 0x00000000 0x00000000

For Uart 1 (not working) at 0x01D0C000:

Offset

In the UART 1 case, it looks like you have received some garbage on the RX line prior to this point. Your printout shows LSR=0xF9, which indicates unread data in FIFO and Framing error. Use the scope to verify there is no activity on the line before this.

The code you sent is still not complete. It is missing serialisable.h. You can verify you sent everything by unzipping the files to a new directory and trying to compile yourself.

Jeff

Hi Jeff,

I am terribly sorry for sending incomplete files. This time, I also include the TI CSL and PSP files. These files are located in directory external/ti/.....

As for the LSR=0xF9, I also noticed it. Will this cause the GIO_write() to loop endlessly? If yes, why? In practice, there is no guarantee that the client to this UART1, which is the GPS receiver in this case, will not send data out before I GIO_write() the "reset" command to it.

Thanks.

Chee-Beng

Still not working. Please extract all the files to a new directory on your hard drive and try to compile again so you can see what needs to be fixed. You should send 1 zip file that has all the files and directory structure intact already (not more zip files) so no modifications have to be made.

 undefined    first referenced                                                                                    
  symbol          in file                                                                                         
 ---------    ----------------                                                                                    
 _GPSUartInit ..\\e12\\newdir\\build\\gps\\Debug\\gpscfg.obj

You should step into the GIO_write fuction in CCS. Eventually you will end up in the UART driver code if you keep stepping down, which will give you information on where it is stuck polling.

Jeff

Hi Jeff,

I thought I got all inside and tested. I can compile and build successfully. Anyway, I re-do all again. Hope there is no further problem at your side.

Thanks for your help.

Chee-Beng

Thanks I finally got it compiled and running.

Can you check if you have a pullup on the TX and RX lines for UART1? These are necessary for the drivers to work.

Jeff

Hi Jeff,

Nice to learn that the codes can be compiled. I sincerely hope you can help me with this UART1 obstacle I am facing now. My boss is breathing down my neck. We are rushing to get our new product prototype working by end of this month, and I have a lot more work to do.

Today I had ensured the UART1 of our custom board are pulled-up using 2.2K ohms resistors for both RX and TX. However, it is still not working. It still looped endlessly in the first GIO_write() statement.

Thanks, Jeff. Thank you for everything.

Chee-Beng

Ok we're still looking into this.

Theres also the option of just using the register level CSL to implement the UART functions without going through BIOS or the UART driver. Try writing some code to send the reset signal  to the GPS this way in the meantime.

Jeff

Thanks, Jeff.

I will implement it using the register level CSL as mentioned. However, I would like to see the BIOS and UART driver work as it is more elegant, especially handling the interrupt.

Looking forward to hearing from you soon.

Thanks.

Chee-Beng

Hi,

1. Can you please let me know the version of the BIOS PSP drivers that you are using.

2. please let me know the mode of operation (polled/interrupt/edma)

3. can you check if the your code works in the polled mode of operation(with different instances)

4. can you check if the interrupt handler is invoked when the code is executed in interrupt mode?

5. Can you also check that your UART event number is mapped correctly to the proper HWI number

regards,

imtiaz

Hi Imtiaz,

1. The version of BIOS PSP driver is 01.30.01

2. The mode of operation is interrupt.

3. For interrupt mode, only Uart 2 is working. For poll mode, all are working.

4. I don't understand how to check whether the interrupt handler is invoked? Can you show me to way, please?

5. May I know how to check this? I only know my hwiNumber is set to 9, and HWI number are:

/* ECM configuration */

bios.HWI.instance("HWI_INT7").interruptSelectNumber = 0;

bios.HWI.instance("HWI_INT8").interruptSelectNumber = 1;

bios.HWI.instance("HWI_INT9").interruptSelectNumber = 2;

bios.HWI.instance("HWI_INT10").interruptSelectNumber = 3;

Thanks.

Chee-Beng