How to reset/prescale the timer?

Hello David,

On disabling the timer, try writing to the GPTMTAV register.

Regards

Amit

I am not trying to disable it, but to reset the count to zero

Also, I need to control the prescaler.

Thanks!

Isn't that what TimerLoadSet does if you set it to zero?

Robert

I thought so too, but it didn't work

Interesting, I haven't as of yet needed to do this.  Someone must have though.  If that indeed doen't work the Tivaware documentation is in need of an update.

Hmm, looks like the load register is actually the reload register.  You have to write to the value register from the looks of the chip documentation. And it looks like the TIVAWare API doesn't provide an interface for that although the chip documentation does say that it is a writeable register.

Maybe someone else has a better idea?

Robert

Hello Robert,

The Load Register is actually the register for the limits of the up count or down count. The Value Register is what affects the run time changes to the counter

Regards

Amit

That's how I thought it must be behaving after David's report and re-reading the TIVAWare documentation.  Which then lead me to check the chip documentation.

The TIVAWare documentation can, unfortunately, be read as updating the value register since it mentions that it also/still updates the register if the timer is running, a qualification that is normally only needed for the value register.  Reload registers are usually updateable whether the timer is running or not.

Robert

Based on Amit's comment, HWREG(TIMER1_BASE+0x50)=0; will do the trick to reset the counter (if you know the API in TiVaware, please let me know)

but HWREG(TIMER1_BASE+0x38)=25; still does nothing to affect the prescaler

Any suggestion on the prescaler (setting and resetting)?

I would expect to have to stop the timer before changing the prescaler but admit I haven't checked.

Robert

Hello David,

I would think it to be disable, update the prescalar and then enable. But to confirm it would need to run some tests as well.

Regards

Amit

I have tried that already and the prescaler simply has no effect no matter what I assigned to it

TimerDisable(TIMER1_BASE, TIMER_A);

TimerPrescaleSet(TIMER1_BASE, TIMER_A, 255); //No effect
HWREG(TIMER1_BASE+0x50)=0; //this indeed resets the counter
//HWREG(TIMER1_BASE+0x38)=255; //no effect

TimerEnable(TIMER1_BASE, TIMER_A);

Are you sure the prescaler is available in your timer configuration?  I know the chip I'm using documents restrictions based on timer width.

Robert

Amit Ashara said:

The Load Register is actually the register for the limits of the up count or down count. The Value Register is what affects the run time changes to the counter

Interestingly, my chip documentation matches the TIVAware documentation but only for the count down case.  For count up writing to the value register is required.

That's a subtle point that should probably be repeated in the TIVAware manual.

Robert

Might the answer be that this MCU - in this particular timing mode - employs the prescaler as a, "Timer Extension" - which is outside of your (and most here's) expectation.

If true - the prescaler will be "hit" only upon your timer's overflow - thus will not cause the effect I suspect that you seek...

I also considered such case since it was mentioned in some modes

but the prescaler register's value never change when counter overflows, it retains whatever I assigned to it, and it has no effect on anything that I can see

I am using edge count on 1294, and Table 13-3 t suggests this mode has an 8-bit prescaler + 16-bit counter

As you have awareness of that (unwanted) timer extension behavior might the order of prescaler & timer set-up, config. & enabling be critical?

Like Robert - I've not used in awhile - yet recall some such sensitivity - reported here - long past...

In addition - we note that you, "count up" and our MCU manual warns that this limits prescale use:

"Note that when counting down in one-shot or periodic modes, the prescaler acts as a true prescaler and contains the least-significant bits of the count. When counting up in one-shot or periodic modes, the prescaler acts as a timer extension and holds the most-significant bits of the count. In input edge count, input edge time and PWM mode, the prescaler always acts as a timer extension, regardless of the count direction."

Hello All,

Well there is one issue with the code though. The documentation mentions that the Timer Capture Count Mode is a 16-bit mode but the code is configuring it for 32-bit mode.

Secondly, the Prescale in this mode (up count) specifically is a 8-bit extender and it seems that it cannot be loaded since the count from 0 to the 24-bit value is what is being expected. Yes, it an be argued that it is a bug/document does not state it as the base register can be loaded and not the extender register. This will take some digging to find out.

Regards

Amit

Amit, could you tell me how to config it to 16-bit mode? thanks

I tried to force the counter to 16-bit based on my reading of page 976, but the result is the same, no prescaler, neither real prescaler nor extension when overflow

Output:

>>>>>>>>>>>

int main(void) {
unsigned int val, val2;

g_ui32SysClock = MAP_SysCtlClockFreqSet((SYSCTL_XTAL_25MHZ |SYSCTL_OSC_MAIN |SYSCTL_USE_PLL |SYSCTL_CFG_VCO_480), 120000000);

MAP_SysCtlPeripheralEnable(SYSCTL_PERIPH_TIMER1);
MAP_SysCtlPeripheralEnable(SYSCTL_PERIPH_GPIOA);

MAP_GPIOPinConfigure(GPIO_PA2_T1CCP0);
MAP_GPIOPinTypeTimer(GPIO_PORTA_BASE, GPIO_PIN_2);

ConfigureUART();

TimerConfigure(TIMER1_BASE, TIMER_CFG_A_CAP_COUNT_UP);
HWREG(TIMER1_BASE)=4;

TimerControlEvent(TIMER1_BASE, TIMER_A, TIMER_EVENT_POS_EDGE);

TimerDisable(TIMER1_BASE, TIMER_A);

TimerPrescaleSet(TIMER1_BASE, TIMER_A, 255);
HWREG(TIMER1_BASE+0x50)=0;
TimerEnable(TIMER1_BASE, TIMER_A);

while (1){

val2 =HWREG(TIMER1_BASE+0x38);
val = TimerValueGet(TIMER1_BASE, TIMER_A);
UARTprintf("Timer1A: (%u):(%u) \n", val2&0xff, val&0xffff);
}

}

Hello David

For 16-bit mode it is

TimerConfigure(TIMER1_BASE, (TIMER_CFG_SPLIT_PAIR|TIMER_CFG_A_CAP_COUNT_UP));

instead of

TimerConfigure(TIMER1_BASE, TIMER_CFG_A_CAP_COUNT_UP);

Regards

Amit

Tried it, the result is the same.

Prescaler has no effect, and it doesn't serve as extension either (I read val2 =HWREG(TIMER1_BASE+0x38);)

TIMER_CFG_SPLIT_PAIR = 4, so I got a feeling it is the same as my previous test with HWREG(TIMER1_BASE)=4

While struggling with prescaler/extension, this line also caught my attention:

On Page 962, 13.3.3.3, of 1294 data sheet:

....Based on this criteria, the maximum input frequency for edge detection is 1/4 of the frequency.

Since the "the" was not clear here, I would like to have your clarification. In my codes, I used

g_ui32SysClock = MAP_SysCtlClockFreqSet((SYSCTL_XTAL_25MHZ |SYSCTL_OSC_MAIN |SYSCTL_USE_PLL |SYSCTL_CFG_VCO_480), 120000000);

So the maximum input frequency is 1/4 of which clock, 120Mhz? 25Mhz? or what?

Hello David,

I made some code changes and now it works. In the data sheet the conditions for the timer in up count is mentioned as GPTMTnILR > GPTMTnMATCHR and GPTMTnPR > GPTMTnPMR.

Also the read has to be done in the same register as TimerValueGet namely GPTMTnR

    TimerConfigure(TIMER1_BASE, (TIMER_CFG_SPLIT_PAIR|TIMER_CFG_A_CAP_COUNT_UP));

    TimerControlEvent(TIMER1_BASE, TIMER_A, TIMER_EVENT_POS_EDGE);
    TimerLoadSet(TIMER1_BASE, TIMER_A, 15);
    TimerMatchSet(TIMER1_BASE, TIMER_A, 1);
    TimerPrescaleSet(TIMER1_BASE, TIMER_A, 15);
    TimerPrescaleMatchSet(TIMER1_BASE, TIMER_A, 4);
    TimerEnable(TIMER1_BASE, TIMER_A);

    while (1){
        val = TimerValueGet(TIMER1_BASE, TIMER_A);
        UARTprintf("Timer1A: (%u):(%u) \n", ((val&0xff0000) >> 16), val&0xffff);
    }

Regards

Amit

Thanks for your reply!

[Post removed, please see my following post for more info]

Hi Amit,

It seems the meaning of TimerLoadSet, MatchSet, PrescaleSet and PrescaleMatchSet are all redefined here for the function. Could you explain the function of each one in your codes?

Thanks!

Hello David,

I think we need to find a way to get this into the data sheet.

Regards

Amit

Hi David,

I have found TM4C1294 for low edge counts PreScaler only needs to be set if the timer is used in 24 bit up count mode. Otherwise it remains in 16 bit up count mode with no wrapping of count high order bits 23:16.

Also TnILR must to be set above TnMatchCHR or the timer will stop and not reload once it reaches MatchCHR. To state TnILR does not have any control sorry to disagree with that part and find it must be set above MatchCHR for the total number of expected edges to capture. Where as in up count edge time mode TnMatchCHR has no meaning and can be omitted since the timer keeps counting past TnILR after the capture interrupt triggers.

You might try using hex value 0x00FF for the value in PreScaler versus 255 actually 511 not 254 for the full bit field. Then you get 00FF.XXXX.XXXX for the total edge count value returned in both TnV + TnR registers. Otherwise in 16 bit edge count mode TnV holds the match count value.

11.10.15:

511 was not correct when starting from bit 15 bit position 16 counting from 1 but 0x00FF gives the full extra 8 bits for capturing the maximum of 140ms edge times.

Thanks!

BP101 said:

My dynamic floating edge count 4 edge gap band

Might this be made (more) understandable & inviting through the use of "bullet points" with each item (somewhat) detailed & justified.  (i.e. as to role & intent)

Colorful language has its place - but (perhaps) not here - especially when seeking complex (& fringe) tech support...

After digesting that 8 word cascade - responders must ponder "where to begin!"    KISS rules - especially so when seeking diagnostic support...

Hi Amit,

Trying exactly that process in the time capture interrupt handler. Except the reload from the previous captured edge time becomes a difference between 2 edge times. Previous saved edge time data is read from TnR and or either compared to the present and actual TnR. or TnV value. Either way the conversion data is the same in both timer modes almost to the number. Multiplying the edge time or edge counts conversion result does not increase nor decrease the reported RPM it remains under 253 no matter the edge time or count.

Seems as if GPTM#13 is always present even when there is no alternate clock source. When it has been set either way the TnR data does not  change values being returned. In edge time mode the TnV, TnR registers are counting into the high  0x00XX.XXXX but the HWREG read back is very low values and or more often above uint32_t so the ui32 value must be wrapping backwards when that happens hard to tell in single stepping if that is true.

Last step prior to exiting interrupt:

/* Reset the reload time TnILR value. */
ROM_TimerLoadSet(TIMER4_BASE, TIMER_A, ulNewEdgeTime);
HWREG(TIMER4_BASE + TIMER_O_TAMR) |= !TIMER_TAMR_TAILD;

Hi Amit,

>If the clock source is not alternate, then there is none

Agree there shouldn't be #13 yet there appears to be similar behavior in the production run TM4C1294NCPDTi3 for timer 4 and 5.

Otherwise possibly HWREG macro is not actually returning the register results from GPTM_TnR & TnV.

Can you confirm this is the case or not?

BP101 said:

Not able to know for sure when the count is changing relative to the edges

Does not KISS simply, easily & quickly "spring" to the rescue?   (again)

Surely this will reveal:

• Feed the CCPx pin with a "known good" input signal
• Toggle that input signal
• Simply loop on the counter's value - stopping upon change.

It should be possible to "time stamp" the input signal's toggle.

It should be possible to "time stamp" the counter value's change.   Simple subtraction and accommodation of MCU's clock speed should pinpoint.

One would think (and hope.)