DM365 EMIF Interface tCYC

KatieRoberts-Hoffman

Hello 36x gurus,

I have a question about the EMIF interface clock cycle as two documents seem to be in conflict:

1) EMIF users guide SPRUFI1C – page 46 – says tCYC is the period at which the EMIF operates. This value is used in all the setup, strobe and hold calculations in that document.

2) DM365 datasheet SPRS457D – page 101 – says E=(0.5)*PLLC1 SYSCLK4 period in ns. This value is used in all the setup, strobe and hold calculations in that document.

Our SYSCLK4 is 121.5MHz (8.23ns)

E = 0.5 * 8.23ns = 4.12ns

I believe that tCYC and E are the same but this is where I need confirmation. Is tCYC 8.23ns or is it 4.12ns?

Thanks,

Katie

over 14 years ago

0 Tai Nguyen over 14 years ago

TI__Expert 5955 points

Hello Katie,

Sorry for the confusion. If the SYSCLK4 is 121.5Mhz then the tCYC is 2E = 8.23ns in this case.

Thanks,

Tai Nguyen

0 Tai Nguyen over 14 years ago in reply to Tai Nguyen

TI__Expert 5955 points

Hello Katie,

I think we have an error in the DM36x data sheet.

E should be half of the AEMIF clock. So in this case of 121.5Mhz E is 60.75Mhz so tCYC is 16.46ns.

I will correct the DM36x datasheets.

Thanks,

Tai

0 Ron Scott over 14 years ago in reply to Tai Nguyen

Prodigy 40 points

Hi Tai

Katie entered this question on my behalf so I will jump in here.

If what you are saying is true then you need to update both documents because tCYC is defined in SPRUFI1 as the period of SYSCLK4 which in my case is 121.5MHz or 8.23ns and you are saying tCYC is 16.46ns. Please confirm

Thanks

Ron

0 Tai Nguyen over 14 years ago in reply to Ron Scott

TI__Expert 5955 points

Hi Ron,

Yes, I will update the AEMIF UG (SPRUFI1c) also.

The AEMIF clock is running a half rate of the periphral clock and that is why you see a revision with a "2E" sometime but we did not do it right. Our AEMIF peripheral can run at full or half rate but on this device we hard-wired it to a half rate mode.

Thanks,

Tai

0 Ron Scott over 14 years ago in reply to Tai Nguyen

Prodigy 40 points

Hi Tai

I think DM365 datasheet SPRS457D Table 6-22 spec number 10 is incorrect also. Shouldn't the read strobe be (RST+1)*E.

Also spec number 1, shouldn't the turn around time be (TA+1)*E

Please confirm both

Thanks

Ron

0 Tai Nguyen over 14 years ago in reply to Ron Scott

TI__Expert 5955 points

Hi Ron,

As we correct the note for E=2*PLLC1 SYSCLK4 period in ns that should take care of the specs you are questioning.

Thanks,

Tai

0 Ron Scott over 14 years ago in reply to Tai Nguyen

Prodigy 40 points

Just changing the definition of E will not help. You need the +1 to compensate for the -1 in the register field value.

0 Tai Nguyen over 14 years ago in reply to Ron Scott

TI__Expert 5955 points

Hi Ron,

These vaues are obtained from the test programs on the tester and since they are typical values, I think, adding 1 to the equation does not change anything.

Thanks,

Tai

0 Ron Scott over 14 years ago in reply to Tai Nguyen

Prodigy 40 points

Hi Tai

RST is a register value representing a predetermined number of clock cycles minus 1 not a measured parameter. Also in table 6-22 spec 3 a 1 is added to each of the three register values (RS+RST+RH)+3 to compensate for the minus 1 in the register settings.

Thanks

Ron

0 Tai Nguyen over 14 years ago in reply to Ron Scott

TI__Expert 5955 points

Hi Ron,

I see. You are correct. I will update spec 1 and spec 10.

Thanks for letting us know.

-Tai

Processors

Processors forum

DM365 EMIF Interface tCYC