Posted on April 24, 2013 at 02:46Thanks for the data - helpful. Three 320x480 TFTs are listed - hope that yours is the one w/suffix ST28 - others have much less contrast, narrower viewing angles and one has an earlier version of TFT Control IC. Can you confirm that yours is ST28-F01?
That Control IC is most impressive - always better if it can be built into the display (either cog or cof). Much detail in their thick manual - chip enables great flexibility.
To your issue: silent to this point are the over-all architecture of your display system. We know of MCU and TFT - but memory must be provided to store the large image & other data. With 153,600 pixels - it is critical that the fetches from this mass memory store be as efficient as possible.
What is the size of your memory-MCU data path? (expect 16 bits - but 32 would be better) Data should be stored in sequentially increasing memory addresses so that your inner transfer loop can be optimized. Does both your fetch from memory - and then output from MCU to display - occur at your best calculated rate? Have you checked to see if any bottlenecks occur (undesired slowing of this transfer process - possibly due to ''in the loop'' counter comparisons.) In past designs - we've used a free GPIO pin to count the total number of transfers - and used this as the means to ''halt'' the transfer process. (and this TFT controller provides no penalties for any ''extra'' writes - if I read the data sheet correctly)
Years back our group did some set-top box design - you may benefit from studying the methods video guys use to analyze any change in image - in the hope that such change (and the overhead it demands) can be lessened.
Have you experimented to see if any speed benefit results from row vs. column update? Data sheet fails to detail - we've noted some variation in past chips - may prove worthwhile for you to run some experiments - and see.
Note futher that other firms manufacture such 320x480 - and use a different TFT control IC. I'd expect that your bottleneck is more data fetch (between memory and MCU) and then transfer (MCU to TFT) - than that caused by the TFT Controller. Is this your experience? While you've given us the big picture ''55mS'' for a screen rendering - have you monitored and logged the time required by fetch, and then transfer to TFT. What are these times?
Hope that something here proves of some value. Suspect w/growth of smartphone market - newest version of this style/size TFT is a very good choice...
