USBSERVICE + serout2 problem

[aberco]

Last edited by Darrel Taylor; - 10th October 2010 at 16:56. Reason: Incorrect Mod

Well ok, but what does makes it incorrect?

[aberco]

Ok, there's indeed something that relates to endpoint. I understand that you use two bytes located within the USB RAM space to get the TX and RX reports.

BD1STATOUT = _USBMEMORYADDRESS + 8 ; 408h or 208h
BD1STATIN = _USBMEMORYADDRESS + 0Ch ; 40Ch or 20Ch

Now I'm clueless about what this refers to... tried to look at the hid_desc.pbp/asm code and compare it with cdc_desc.pbp/asm, and also to look on the net for RAM endpoint allocation at these two address but no luck so far

[aberco]

Follow up on the monologue, and after a day of digging through literature, I now have a slightly deeper understanding of the USB protocol. The more helpful was Microchip datasheet, within the USB section related to BDT RAM.



200h is start of USB RAM for PIC18F1xK50, 400h for 18Fx550/x455/x450.

I moved the allocations to Endpoint 3 for CDC usage:

RXOWNED VAR BD3STATOUT.7 ; Out report Owned (OUT is PC to PIC)
TXOWNED VAR BD3STATIN.7 ; IN report Owned (IN is PIC to PC)
-------------------------------------------------------------------------------------------------------
BD3STATOUT = _USBMEMORYADDRESS + 18h ; 418h or 218h
BD3STATIN = _USBMEMORYADDRESS + 1Ch ; 41Ch or 21Ch

Bit 7 of BDxSTAT states which device (SIE or CPU) owns the right to write or read into the dual ported USB RAM space, so this is what is used for the TX and RX ready flag.

For some reason the compiler doesn't like the 3 in "BD3STATOUT and BD3STATIN" (Bad data type error). CONFIGURED_STATE should not differs between HID or CDC as it is not endpoint dependent.

[Darrel Taylor]

I moved the allocations to Endpoint 3 for CDC usage:
...
For some reason the compiler doesn't like the 3 in "BD3STATOUT and BD3STATIN"

You missed the EXT vars ...

BD3STATOUT         VAR BYTE EXT      ; OUT report status byte
BD3STATIN          VAR BYTE EXT      ; IN report status byte

[aberco]

Ah thanks! that was just above, an easy fix.

RX_Ready seems to work! but TX_Ready is always true... maybe because the time when the OUT endpoint is owned by the SIE is so short that I can't see it? However if I do a loop that sends data as soon as the SIE is ready I have huge data loss.
Right now I have to add a 200ms delay between each 32Bytes packets I send in order to avoid data transmission loss. It is rather slow... reading a 24LC512 at this speed would take a bit less than 4h.

Am I missing something here?

[Darrel Taylor]

My shipping container just arrived yesterday with all my stuff from California, still unloading it.
Hopefully I can get some computers setup soon to be able to work on this stuff again.

Can't sit in the forum at work, just too busy during the day.

If you haven't figured it out by next week, I'll be back on it.
Sorry, I haven't been much help.

[aberco]

No Darrel, you have been very helpful so far, at least to set me on the right course to fix issues.

I'm much more of an hardware person, getting better and better at programming but some time simple software tricks are not obvious to me... However when it comes to reading datasheets and understanding hardware peripherals, registers and configuration bits I feel at home.

I will carry on progress one step at a time

[aberco]

I think that most of my USB problems are solved now!

I rewrote the USB Out routine as such:

'*******************************************************
'Writing data to USB port
'*******************************************************
DoUSBOut:

USBBufferSizeTX = 0
FOR Counter = 1 TO USBBufferSizeMax 'Calculate number of valid bytes within the string
IF USBBuffer [Counter - 1] > 0 THEN USBBufferSizeTX = USBBufferSizeTX + 1
NEXT Counter

IF TX_READY THEN 'USB data can be transmitted
    TX_READY = 0 'Clear USB data send flag
    USBOUT 3, USBBuffer, USBBufferSizeTX, DoUSBOut
ENDIF

FOR Counter = 1 TO USBBufferSizeMax 'Clear the USB buffer after data has been sent
USBBuffer [Counter - 1] = 0
NEXT Counter

RETURN

- Fist step is to calculate how many bytes are used in the string (not = 0)
- Second, sending data
- third, blanking the data buffer for next operation

I have no more garbage data in my strings, and I was able to remove the Pause without any data loss. Transmission is very fast now, with over 10kB/s (I have large conversion routines between my DoUSBOut, could even be faster without these).

Now onto other problems, especially getting all the interrupts to work together,and implementing ADC and ultra-low power routine.

What about publishing the generic code for CDC? I'll let you decide since it is actually minor modification to code you have written.

[aberco]

And everything collapsed!

The CDC USB routine was working well until now. It is working very well on my early prototypes, I made 3 more identical one and the PIC just stop running randomly after a few seconds up to several minutes. USB communication stops and LEDs turns off. It still respond to the INT0 interrupt though, and resume normal operation after reset... until it crashes again randomly.

I am puzzled as the hardware is strictly identical, the only difference I could spot is in the silicon revision of the PIC18LF14K50, says 0000006 on the working batch and 0000005 on the batch that crashes. I have inserted a simple blinking LED loop routine at the very beginning of my code, and it still crashes and stop blinking. Removing part of the program related to USB solve the crashing problem so I would think that there is something wrong in the USB CDC code I have modified.

Problem is that in the revision errata there is nothing linked to the USB SIE, and I am totally clueless about where to start to debug this issue. What could be wrong to have the PIC crash at a random time interval? some kind of buffer overflow?

[aberco]

A bit more debugging, it seems that only the USBIN side does crash.

'*******************************************************
'Reading incoming data from USB port
'*******************************************************
DoUSBIn:

FOR counter = 1 TO USBBufferSizeMax 'Clear the USB buffer before data is received
    USBBuffer [counter - 1] = 0
NEXT COUNTER

WHILE !RX_READY                                                                   
    GOSUB UpdateLEDStatus
WEND
                                   
IF RX_READY THEN
    RX_READY = 0 'Clear USB data received flag
    USBBufferCount = USBBufferSizeRX
    USBIn 3, USBBuffer, USBBufferCount, DoUSBIn    
ENDIF

RETURN

At random the UpdateLEDStatus subroutine stop to be serviced, and the USBIN does not receive anymore data. When it crashes, RX_READY = 0, then if I send data RX_READY = 1 and just stay there, without being cleared. the USBOUT side toes work well though, even when the IN side has crashed.

Might have something to do with USB_ASM_Service.pbp not working properly?

-> A bit more debug introducing some LED blink to see if the USB_ASM_Service is being called: it stops being called when it crashes...

[aberco]

Well.. finally was able to track the problem down to a conflict between the use of SPWM_INT.bas and USB_ASM_Service.pbp. If I disable the Timer1 interrupt used for sPWM then the USB works well. Strange thing is that the code works well on 3 prototypes with the older PIC silicon revision, and does not work with 3 prototypes with the newer silicon revision...

Anyway, here is the debug program I have that run the sPWM and at the same time read a string from USB and send it back (using CDC serial port emulation).

INCLUDE "cdc_desc.bas"


INCLUDE "DT_INTS-18.bas"     ; Base Interrupt System
INCLUDE "ReEnterPBP-18.bas"  ; Include if using PBP interrupts


INCLUDE "SPWM_INT.bas"


INCLUDE "USB_ASM_Service.pbp"


' USB status variable handled by USBASMService
' "PLUGGED" indicates when the device is fully connected to the PC
' "RX_READY" indicates when a report has been received
' "TX_READY" indicates when it's OK to Send. The device is Plugged and the SIE is ready to accept data


'*******************************************************
'Init the software multi_PWM system
'*******************************************************


DEFINE SPWM_FREQ  200 'SPWM Frequency
DEFINE SPWM_RES   256 'SPWM Resolution


DutyCycles VAR BYTE [3] 'DutyCycle Variables
RedDutyCycle VAR DutyCycles [0]          
GreenDutyCycle VAR DutyCycles [1]          
BlueDutyCycle VAR DutyCycles [2]


ASM
SPWM_LIST  macro ;Define Pin's to use for SPWM and the associated DutyCycle variables
     SPWM_PIN  PORTC, 3, _RedDutyCycle
     SPWM_PIN  PORTC, 6, _GreenDutyCycle  
     SPWM_PIN  PORTC, 4, _BlueDutyCycle
  endm
  SPWM_INIT  SPWM_LIST ;Initialize the Pins
ENDASM


'*******************************************************
'Init the interrupt handling system
'*******************************************************


ASM
;----[High Priority Interrupts]-------------------------
INT_LIST  macro     ; IntSource,          Label,  Type, ResetFlag?
        INT_Handler     USB_INT,  _DoUSBSERVICE,   ASM,  yes
        INT_Handler   TMR1_INT,  SPWMhandler,  ASM,  yes
    endm
    INT_CREATE ; Creates the High Priority interrupt processor
ENDASM


'*******************************************************
'BlueTooth module initialisation
'*******************************************************


@ INT_ENABLE  USB_INT ; Enable USB Servicing interrupt
@ INT_ENABLE  TMR1_INT ; Enable PWM interrupt


'*******************************************************
'Main program loop
'*******************************************************
MainLoop:


GOSUB DoUSBIn


GOTO MainLoop


'*******************************************************
'Reading incoming data from USB port
'*******************************************************
DoUSBIn:


FOR counter = 1 TO USBBufferSizeMax 'Clear the USB buffer before data is received
    USBBuffer [counter - 1] = 0
NEXT COUNTER


WHILE !RX_READY                                                                   
    PAUSEUS 5
WEND


IF RX_READY THEN
    RX_READY = 0 'Clear USB data received flag
    USBBufferCount = USBBufferSizeRX
    USBIn 3, USBBuffer, USBBufferCount, DoUSBIn    
ENDIF


GOSUB DoUSBOUT


RETURN


'*******************************************************
'Writing data to USB port
'*******************************************************
DoUSBOut:


USBBufferSizeTX = 0
FOR counter = 1 TO USBBufferSizeMax 'Calculate number of valid bytes within the string
    IF USBBuffer [counter - 1] > 0 THEN USBBufferSizeTX = USBBufferSizeTX + 1
NEXT COUNTER


IF TX_READY THEN 'USB data can be transmitted
    TX_READY = 0 'Clear USB data send flag
    USBOUT 3, USBBuffer, USBBufferSizeTX, DoUSBOut
ENDIF


HSEROUT [STR USBBuffer \USBBufferSizeTX]                                                      


ClearUSBBuffer:


FOR counter = 1 TO USBBufferSizeMax 'Clear the USB buffer after data has been sent
    USBBuffer [counter - 1] = 0
NEXT COUNTER


RETURN


END

[aberco]

After debugging everything pinpoint an issue coming from the different revision of the PIC. The CDC routine works with silicon revision 0000006 of the PIC18LF14K50, but not 0000005.
Really getting stuck there as I have no clue of how to solve the problem, and I cannot rely on ordering a specific revision of the microcontroller...