AnsweredAssumed Answered

BF548 DMA SPORT RX behavior

Question asked by rschoop on Aug 8, 2013
Latest reply on Sep 26, 2013 by Ronak.Shah

Hello all,

 

I have a custom made board based on the BF548 processor and have a question regarding the SPORT interface.

 

I have the following situation :  

 

I am using SPORT1 as RX  with  external  clock  and  also external  RFS.   The  clock and the  RFS  are generated by  GP Timers from the same BF548 processor.  These  GP Timer generated clock and frame  are also applied to a external  chip  for  clocking in  serial data into the BF548 through SPORT1.   This external chip expects a Chip Select  and a  clock as input.

 

I have troubles with the  DMA  unframed mode ,  because  the number of  DMA interrupts in a given period  is  not equal to  number of times  I  triggered the  clock and the  frame .  So  I have a  sync problem here  I believe.

 

When  using the DMA  framed mode  I don't have the above mentioned problem ,  but   still I received   junk data.   In the framed mode  ,  a frame sync is  required for  each word transfer, but   actually  I am using  only  one  frame sync  (late frame)  for  64 clocks  period.  I am not using 4  frame syncs  for  each  16 bit word.    Maybe it is worth mentioning that  I am using the  Primary and Secondary for  receiving data.  So  I am receiving in total  8 x 16 words in the  FIFO .  This data is supposed to be DMA'ed to my  memory location.

 

I think it is strange that  the DMA is working properly  but  the  data is junk.   The reason why  I am not using   4  frame syncs is because the  external chip  expect  one  frame sync of  64 clocks.  I  think  that  I  can bypass this  by   using an additional timer to emulate the required signals but  still  ,  I  am wondering why  the  one frame sync  is  not working .

 

Is the reason for receiving  junk data is because  I  am not using  a  frame sync for each word ?? 

Why then   is the  DMA  gives an interrupt after the  8x 16 words  properly ??  I assume  because there is no  4 frame syncs ,  it  should not fire an interrupt.

 

Looking forward for a reaction ..

 

Thanks,

 

rschoop

Outcomes