Calculating a known and deadly accurate interrupt frequency rate (pic multicalc etc)

[HankMcSpank]

Ok, so I'm back to PICs after an enforced time away...I'm wanting to get stuck into DDS, but straight out the blocks, I'm confused (it doesn't take much!)

Right, I want a steady and accurate 20khz interrupt rate - being of the lazy ilk, my first port of call was mister_e's mutlicalc ('Timer helper' tab V1.3.1 dated 2006 - is this the latest version?)

I'm using timer2 (8 bit timer)
Oscillator is running @ 16Mhz
The calculator said I should use a preloader of 63

So I knocked up this little program...

@ __CONFIG _CONFIG1, _FCMEN_OFF & _FOSC_INTOSC & _WDTE_OFF & _MCLRE_OFF & _CP_OFF & _IESO_OFF & _BOREN_OFF & _PWRTE_OFF & _LVP_OFF
@ __CONFIG _CONFIG2, _LVP_OFF
INCLUDE "DT_INTS-14.bas" ' Base Interrupt System
Osccon = %01111010   'sets the internal oscillator to 16Mhz
DEFINE  OSC 16
TrisA.5 = 0     'Pin2 an output.     
CM1CON0 = 0   ' COMPARATORS OFF
CM2CON0 = 0   ' COMPARATORS OFF
ASM
INT_LIST  macro ; IntSource,    Label,         Type, ResetFlag?
    INT_Handler  TMR2_INT,  _DDS,  asm,  YES
    endm
    INT_CREATE       ; Creates the interrupt processor
ENDASM
T2CON.2 = 1   ' Timer2 on
PR2 =  63     ' this should yield an interrupt rate of 20khz at 16Mhz.
 
@ INT_ENABLE TMR2_INT
Main:
    pause 100
    goto main   
END
'+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
DDS:
toggle PORTA.5
@ INT_RETURN

But when I actually scope pin 2 - I see a frequency of 31.2Khz - now bearing in mind because I'm toggling, the scoped frequency will be half of the actual interrupt frequency, therefore there must be an interrupt frequency of 62.4khz going on with the PR2 value?!!.

So what am not doing right?!! (it's sure to be embarrasingly simple!)

(oh, btw whats's the purpose of the 'reload (instr cycle)' slider in mister_e's calculator??)

So after dicking about for a while, I decided that better get my backside in gear & knock up a spreadsheet to work this all out...

Oscillator Freq         16000000Hz
Instruction clock        4000000Hz
Time per instru       0.00000025 Seconds
PR2 set to                   200 counts
Interrupt takes          0.00005 Seconds
Interrupt Frequency        20000Hz
Scoped  Toggle rate        10000Hz

Now when I scope pin2, I see something more like the expected result - 9.93khz ....but where did the other .07khz go?!!

So what am I missing to get an accurate interrupt toggle frequency reading of 10khz? (which would be a 20khz interrupt rate)

[cncmachineguy]

Hank, with regular timers, you need to account for the time it takes to reload the timer after it times out. that what the slider is for. But when using PR, it is my understanding the timer just restarts, so no re-load time, no accounting for interrupt time or ISR entry and exit. It should be right on as I understand it. Are you quite sure your scope is true? can you check the OSC with it? maybe the 16M isn't quite. (Maybe you can put a clock ot on one of the OSC pins if using internal clock)

[Darrel Taylor]

PicMultiCalc doesn't do TMR2,4,6 unless PR2=255 and you reload on every interrupt.

TMR2 counts up to the PR2 value + 1, then resets and interrupts.
So you were only getting 64 counts. (62.5Khz)

For 200 counts, set PR2 to 199.

[HankMcSpank]

Thanks guys.... PR2 of 199 does indeed get me a toggle frequency of 10khz (actually it's showing bouncing between 9.99khz & 10.0khz ....but that's likely to be a rounding/beating error to do with my scope).

Darrel thanks for solving the multicalc puzzle to ...thought I was going insane (I probably am anyway "Nurse, more medication quick").

[cncmachineguy]

You know I knew that, and as soon as I opened this this morning I thought "PR needs to be reload value -1", I bet thats what DT has posted. It still doesn't explain 9.93K, cuz the math if using 201 for the value (pr=200) works out to 9.95K. Thats where I was stumped.