FSMC and bus cycle time, EVAL3210-E

[jpeacock2]

Posted on March 13, 2009 at 16:01

FSMC and bus cycle time, EVAL3210-E

[jpeacock2]

Posted on May 17, 2011 at 13:06

I have an LCD peripheral connected to the FSMC SRAM bus (EVAL3210E board, ISI9320 LCD controller). The FSMC parameters handle address and data setup times, but there doesn't seem to be anything for bus cycle times (100ns write or 300ns read in this case). It seems that if I DMA to this peripheral the cycle times will be violated. Is there any way to handle cycle time with the FSMC controller? Or do I have to manually insure that reads and writes are far enough apart to not violate the cycle time, which would preclude using DMA?

The example software for the EVAL3210E board doesn't seem to allow for this either, in particular where the init routine does a long string of writes to the ISI9320. Does it just happen to work out that there's enough overhead in the C routines to get around the 100ns/300ns cycle time restriction?

[jj]

Posted on May 17, 2011 at 13:06

Cannot yet completely/directly answer your question - perhaps some of our findings will prove useful.

Please clarify - is the ''Lcd peripheral'' you reference the 2.4'' TFT which comes ''standard'' with ST's ''E'' Eval board? A newer version of the Lcd Controller exists (ISI9325) and our initial testing shows that it is more ''forgiving.'' (faster) There is a significant weakness -imho - with this class Lcd Controller - you will ''not'' find these on larger, more popular (3.5'' & up) TFTs. Thus much of your development effort will require modification when your clients seek a larger display. (and they will) Does your use of DMA transfer 16 bit data from ''on board'' SRAM directly to the LCD bus? Again - our initial test results seem to indicate some variation in transfer rates - based upon DMA source/destination. It should not be too difficult for you to ''scope'' the ''nor'' of Lcd_WR & Lcd_CS signals - to determine the exact bus cycle ''write'' time. If you do this - and post results - I will replicate on our end... [ This message was edited by: jj.sprague on 13-03-2009 21:16 ]