Home > Developement Guide > FAQ |
FAQ |
HMS30C7202 : ADC LCD PMU PLL RESET SDRAM SMI MMC UART CAN SPI USB TIMER PWM PS2 KED GPIO APPLICATION |
HMS30C7201 : ADC PMU PLL ETC APPLICATION |
 |
Q1. SCLK is FCLK (70MHz)?
|
|
|
|
|
 |
Q2. If auto-precharge is disabled (SCDON A=0), does the SDRAM controller then generate precharge commands as necessary?
Is there any reason why I would want to disable auto-precharge? It looks like it is better to use auto-precharge, right?
|
|
|
|
|
 |
Q3. There is almost no timing parameter of the SDRAM controller in the datasheet, and only very little adjustment is possible.
Things like read to write, consecutive read or write bursts etc. are not specified. So it is impossible for me to check if the CPU
will meet all the SDRAMs specifications, and if there are problems, it will not be possible to correct them in software with adding
additional cycles. Can you guarantee that the CPU will work with the above mentioned SDRAMs?
|
|
|
|
|
 |
Q4. Burst length: The CPU datasheet says on ch.6: "Video accesses to the SDRAM occur in fixed-burst lengths of 16 words.
ARM and DMA controller accesses occur in a fixed-burst length of four words." Word is 32 bit. This means that video accesses
are 32 x 16 bit, and ARM+DMA 8 x 16 bit. So there are 4 consecutive read bursts for each video access? Or does the controller
use full page bursts?
|
|
|
|
|
 |
Q5. See chapter 6: "Supports 2/4 banks with page lengths of 256 or 512 half words". If so, how does it know the SDRAM's page
size?
|
|
|
|
|
 |
Q6. SDRAM mode register (in the SDRAM, not SDCON) How this register must be initialized?
|
|
|
|
|
 |
Q7. Base address of SDRAM 0 Memory Map Table 4-1: base address = 1024 MByte = 0x4000.0000 ch.6.2.1 SDCON register
Bit 3: "1 = a device is present at address range 0-32MByte" Shouldn't it be 1024-1056MByte?
|
|
|
|
|
 |
Q8. Referring to the figures in "12.5.2 SDRAM Interface"
|
|
|
 |
Q8-1. RAS/CAS Timing:The 42ns are 3xSCLK ? Is 3x fixed or does it depend on any other parameter?
|
|
|
|
|
 |
Q8-2. All timing diagrams:Shouldn't RAS and CAS go high again after 1 SCLK (-> NOP command)? RAS+CAS low is auto/self
refresh command, and with WE low it's load mode register command, or am I wrong?
|
|
|
|
|
 |
Q8-3. Single/Burst Read Operation:The 42-46ns is 3xSCLK. Is this with programmed CAS latency=3 ? Yes The SDRAM supports
CAS latency=2 at max 100MHz. So it should be no problem to use CAS latency=2 (SDCON C[1:0]=10) with 70MHz.Do you expect
any problems?
|
|
|
|
|
 |
Q8-4. Single Read/Write: Does the SDRAM controller perform single accesses at all?
|
|
|
|
|
 |
Q9. How to connect the SDRAM's address lines(64MBit, 16M x 16) to the CPU?
|
|
|
|
|
 |
Q10. If the SDRAM controller handles all this, I will of course use the faster mode and disable auto precharge.So the SDRAM
controller issues a PRECHARGE command automatically whenever it is necessary?It outputs the auto precharge control bit on
SA10?
|
|
|
|
|
 |
Q11. In "12.5.2 SDRAM Interface", there are figures for "Single Read Operation" and "Single Write Operation". Because the SDRAM
controller makes only 8-halfword burst accesses, I wonder why there are these figures. When does the CPU make this "Single
Read/Write Operation"?
|
|
|
|
|
 |
Q12. Read access: when does the SMI latch the input data?
|
|
|
|
|
 |
Q13. When does CS goes LOW? When the address is put on the bus (figures in 12.5.1) or one BCLK after the address (figures
in 7.4)? This makes one BCLK difference for the waitstates.
|
|
|
|
|
 |
Q14. What is the difference between "7.4.2 read normal wait (sequential)" and "7.4.3 read burst wait (sequential)" from the
memory's view? (Q14-1) If the figures are correct, the memory sees no difference between normal sequential and burst
sequential, except that for burst the first cycle can have different timing. This means that sequential accesses are always burst,
even if burst is disabled (see 7.4.2). Is this true? (Q14-2) I think I can use burst read wait=3 and normal access wait=4, exactly
like in 7.4.3:for the 1st access, nRCS active to data latch is 4xBCLK, and for the following accesses RA change to data latch is
also 4xBCLK. Is this correct? (Q14-3)
|
|
|
|
|
 |
Q15. The problem is about the MMC Response Buffer Register (0x8001.5048). I couldn't write or erase it. Also a reset didn't
change the value.It only changes the value after a power down (then it assumes another- random -value). Can it be modified?
How?
|
|
|
|
|