Retrieving 32bit Multiply Result

[kasapo]

Hi darel,

i have a motor control application. in my application pic 16f877 is reading 2 position and then take the ratio of these values and multiply with the constant velocity. for example
x=5 y=10 then Vy = Vc (constan velocity) , Vx = Vc*x/y
so two motor begin and stop at the same time. i wrote some code by using Div32 but sometimes i cannot get the values thet it must be.

// for Trial i give posy greater than posx
POSX VAR WORD
POSX1 VAR POSX.BYTE1
POSX0 VAR POSX.BYTE0

POSY VAR WORD
POSY1 VAR POSY.BYTE1
POSY0 VAR POSY.BYTE0

V VAR WORD
V0 VAR V.BYTE0
V1 VAR V.BYTE1

HV VAR WORD
HV0 VAR HV.BYTE0
HV1 VAR HV.BYTE1

A VAR WORD
A0 VAR A.BYTE0
A1 VAR A.BYTE1


C VAR WORD
C1 VAR C.BYTE1
C0 VAR C.BYTE0

DUMMY VAR WORD


Asm
PutMulResult macro Const32
MOVE?CB Low Const32, R2
MOVE?CB Low (Const32 >> 8), R2 + 1
MOVE?CB Low (Const32 >> 16), R0
MOVE?CB Low (Const32 >> 24), R0 + 1
endm
EndAsm

GoSub USBREAD
POSX1=HB
POSX0=LB
GoSub USBREAD
POSY1=HB
POSY0=LB

IF POSY > 6554 Then
DUMMY = POSY * 10 // i scale posy by 10
C= Div32 POSX
@ PutMulResult 2236920 // i also enter here the scaled c constant velocity by 10
V= Div32 C
HV = 34 / C
Else
DUMMY = POSY * 10
C = DUMMY / POSX
@ PutMulResult 2236920
V= Div32 C // when this is greter than 16 bits it returns ffff
HV = 34 / C
EndIF

how can i get the V when it is greater than 16 bit ?
thanks

[Darrel Taylor]

There's no way to do that with DIV32. The Quotient must be less than 65535.

You would need a TRUE 32 bit divide routine.

Or you could use the Floating Point Math routines from melabs.
http://www.melabs.com/resources/fp.htm

Darrel

[kasapo]

hi darrel ,
do you have some example about the usage of floating point numbers.

after i do my multiplication and division how can i use aint variable?
i have to send the aint to my motor controller byte by byte.

thanks

[mytekcontrols]

Hi Darrel,

I like your ideas about getting at DIV32 system variables, but I have a slightly different situation. How do I use a variable for the divisor that wont violate the 15 bit rule?

I tried the following example, but it don't work (I get: Error Bad expression):

divisor var word
dummy var word

dummy = DIV32 divisor

I really need the ability to alter the divisor via program control, and at times it will need to be bigger then an 8-Bit value. Any ideas?

Thanks,

Edit: Opps! my mistake. It does work properly with a word variable as the divisor, just so long as you stay with the 15 bit rule (32767). This is good news.

[Darrel Taylor]

hehe,

DT

[Josuetas]

I just saw this link. it is of great interest to me right now...

I have this question... how do you add two of these Dwords?

how do i know if the add of two words overflows?
i.e. $1FFFE + $2E

my guess is to add the lower words $FFFE+$2E. But how do i know if there is an overflow? so that i can add one to the higher WORD? even worse..

i.e. $EFFF8 + $6FFE4

My guess again: add two lower words... if over flow add two higher words +1, else add two higher words....

I think i can manage myself once i am sure abour the overflow

Thanks,

[Darrel Taylor]

For words and bytes, you can detect an overflow by testing if the result is less than what you added.

Due to trucation of a 16-bit word ...
$FF00 + $0180 = $0080

So the result will always be less than either of the values added if an overflow occurs.

I think i can manage myself once i am sure abour the overflow

I'm sure you could. But that's just not my way ...

;Initialize your hardware first

A         VAR WORD[2]
B         VAR WORD[2]
Result    VAR WORD[2]
OVRFLOW32 VAR BIT

ASM
; ===== Load 32bit Constant into DWORD =============================
MOVE?CD macro C32in, Dout                      ; Max= 4,294,967,296
    MOVE?CB   low C32in, Dout
    MOVE?CB   low (C32in >> 8), Dout + 1
    MOVE?CB   low (C32in >> 16), Dout + 2
    MOVE?CB   low (C32in >> 24), Dout + 3
  endm
ENDASM

@  MOVE?CD  0xEFFF8, _A  ; Load 983,032 into A
@  MOVE?CD  0x6FFE4, _B  ; Load 458,724 into B

GOSUB Add32

LCDOUT  $FE,1,  "A=   ",HEX4 A[1],":",HEX4 A[0]
LCDOUT  $FE,$C0,"B=   ",HEX4 B[1],":",HEX4 B[0]
LCDOUT  $FE,$90,"Res= ",HEX4 Result[1],":",HEX4 Result[0]
LCDOUT  $FE,$D0,"OVRFLOW32 = ", BIN1 OVRFLOW32
stop

; ===== Add 2 32bit variables ======================================
Add32:
    OVRFLOW32 = 0
    Result[0] = A[0] + B[0]
    IF Result[0] < B[0] then Result[1] = 1
    Result[1] = Result[1] + A[1]
    IF Result[1] < A[1] then OVRFLOW32 = 1
    Result[1] = Result[1] + B[1]
    IF Result[1] < B[1] then OVRFLOW32 = 1
RETURN

Which displays ...

A=   000E:FFF8
B=   0006:FFE4
Res= 0015:FFDC
OVRFLOW32 = 0

HTH,