Post by thread:[PIC] seperate num into tens+units

hi, i need a way to seperate a number, say 54. and put 5 in a register called tens, and 4 in another called units. is there some binary operation to do this, or will it have to be a custom subroutine? i need this kinda fast, can someone help? -- View this message in context: www.nabble.com/seperate-num-into-tens%2Bunits-tf3356654.html#a9335697 Sent from the PIC - [PIC] mailing list archive at Nabble.com.

See binary to bcd at http://www.piclist.org/techref/microchip/math/radix/index.htm Harold {Quote hidden}> > hi, i need a way to seperate a number, say 54. and put 5 in a register > called > tens, and 4 in another called units. > > is there some binary operation to do this, or will it have to be a custom > subroutine? > > i need this kinda fast, can someone help? > -- > View this message in context: > www.nabble.com/seperate-num-into-tens%2Bunits-tf3356654.html#a9335697 > Sent from the PIC - [PIC] mailing list archive at Nabble.com. > > -

noob, I think your best bet is to write a loop that subtracts ten and increments the "tens" register until the Carry flag is set, then decrement the tens register, add 10 to your number, and move it to the units register. Basically, you're doing division by hand. - Marcel On 3/6/07, picnoob <spam_OUTleeguy92TakeThisOuTgooglemail.com> wrote: {Quote hidden}> > hi, i need a way to seperate a number, say 54. and put 5 in a register called > tens, and 4 in another called units. > > is there some binary operation to do this, or will it have to be a custom > subroutine? > > i need this kinda fast, can someone help? > -- > View this message in context: www.nabble.com/seperate-num-into-tens%2Bunits-tf3356654.html#a9335697 > Sent from the PIC - [PIC] mailing list archive at Nabble.com. > > -

I had this issue short time ago, I found that a separate routine to handle units, tens, etc is more efficient and consumes less code, than splitting, of course depends on the final application, I used this routine for a timer struct number { unsigned char u; unsigned char d; unsigned char c; }; // Increase (like ++) void inc(struct number *value) { value->u++; if (value->u == 10) { value->u = 0; value->d++; } if (value->d == 10) { value->d = 0; value->c++; } if (value->c == 10) { value->c = 0; } } // Decrease (like --) void decr(struct number *value) { value->u--; if (value->u == 255) { value->u = 9; value->d--; } if (value->d == 255) { value->d = 9; value->c--; } if (value->c == 255) { value->c = 9; } } // Number = 0 void clr(struct number *value) { value->u = 0; value->d = 0; value->c = 0; } // Set value to struct void set(struct number *value, unsigned char c, unsigned char d, unsigned char u) { value->u = u; value->d = d; value->c = c; } // Test if zero unsigned char zero(struct number value) { if ((value.u == 0) && (value.d == 0) && (value.c == 0)) return 1; else return 0; } // Equal two structures void equal (struct number src, struct number *dst) { dst->u = src.u; dst->d = src.d; dst->c = src.c; } Cristo Alanis Design Engineer Tyco Electronics 2500 Courage Blvd St A Brownsville Tx 78521 > {Original Message removed}

On Tue, 2007-03-06 at 12:18 -0600, PAUL James wrote: > Sir, > > I would "AND" the number '54' with 0x0F. This will give you '4'. Store > it in it's destination register, whatever tht may be. Then do a SWPAF > and "AND" it again with 0x0F. This will give you the '5'. Store it > In it's destination register. Now you're done. That's only if you want hex 0x54 printed as 5 and 4. There's nothing wrong with that, but in my mind it's likely the op was speaking of decimal 54, which in hex is 0x36. TTYL

On Tue, Mar 06, 2007 at 12:18:39PM -0600, PAUL James wrote: > > Sir, > > I would "AND" the number '54' with 0x0F. This will give you '4'. Store > it in it's destination register, whatever tht may be. That's only if the number is 54 hex. 54 decimal is 36 hex. So anding with 0x0F would actually give you a 6. You'd need to convert hex to BCD, then perform the operation in question. > Then do a SWPAF > and "AND" it again with 0x0F. This will give you the '5'. Store it > In it's destination register. Now you're done. Same problem as before. For my sunrise/sunset outdoor light controller I wrote a hex to BCD routine. It's a bit involved. The basic idea was to convert the high nex nybble to BCD via a table, then add the low BCD digit. For example your 54 decimal (0x36) would take the 0x30 and convert it to 0x48, which is the BCD equivalent to 0x30. It then adds the 6 back giving 0x4E. You then decimal adjust by adding 6 more giving 0x54, which is the BCD equivalent to the original 0x36. Here are my comments from the routine which can be found here: http://www.finitesite.com/d3jsys/clock.asm ; Add the low nybbles together. Compensate if the sum is greater than 9 ; Then add the result to the high nybble of the BCD of the high nybble. ; The low digit add plus the compensation may have pushed the low digit back ; into the hex range. Need to check and recompensate. For example 1F maps to ; (16 + 0F). The low digit sum is 15 hex and with the compensation is bumped ; to 1B. It needs to be compensated again by adding 6 more to get to 21 BCD. ; ... ; This is where recompensate the low digit sum if necessary. Note that ; This one only checks the low digit of the sum since presumably there will ; be some high digit carry when adding 8+9+6 for example. The simplest way is to do the repeated subtraction as suggested in another post. My conversion does simplify tracking the high BCD digit, but the low digit recompensation is a bit of a pain. BAJ

picnoob wrote: > hi, i need a way to seperate a number, say 54. and put 5 in a > register called tens, and 4 in another called units. Here. This routine separates the number into three digits; if you know your original number won't ever be larger than 99 decimal, you can remove all the lines marked with a "*****" comment and it'll just separate the number into two digits: LIST R=DEC UNITS EQU .... ;A FILE REGISTER. TENS EQU .... ;ANOTHER FILE REGISTER. HUNDREDS EQU .... ;ANOTHER FILE REGISTER. ***** ; ENTER WITH ORIGINAL 8-BIT VALUE IN "UNITS". EXITS WITH ONES' ; DIGIT IN "UNITS", TENS' DIGIT IN "TENS", AND HUNDREDS' DIGIT ; IN "HUNDREDS". CONVERT CLRF TENS CLRF HUNDREDS ; ***** DO100S MOVLW 100 ; ***** SUBWF UNITS,W ; ***** BNC DO10S ; ***** MOVWF UNITS ; ***** INCF HUNDREDS ; ***** GOTO DO100S ; ***** DO10S MOVLW 10 SUBWF UNITS,W BNC ADJUST MOVWF UNITS INCF TENS GOTO DO10S DONE -Andy === Andrew Warren - fastfwdKILLspamix.netcom.com

;number 0-255 to extract is in temp ;result in d100:d010:d011, add 0x30 to each to make ASCII bin2bcd clrf d100 swapf temp,w addwf temp,w andlw b'00001111' skpndc addlw 0x16 skpndc addlw 0x06 addlw 0x06 skpdc addlw -0x06 btfsc temp,4 addlw 0x16 - 1 + 0x6 skpdc addlw -0x06 btfsc temp,5 addlw 0x30 btfsc temp,6 addlw 0x60 btfsc temp,7 addlw 0x20 addlw 0x60 rlf d100,f btfss d100,w addlw -0x60 movwf d011 btfsc temp,7 incf d100,f movf d100,w movwf d100 swapf d011,w ;separate 10s andlw 0x0f movwf d010 movlw 0x0f ;separate 1s andwf d011,w movwf d011

Jinx's post is a completion (separating the 10s from the 1s) of http://www.piclist.com/techref/microchip/math/radix/b2bhp-8b3d.htm by Scott Dattalo which is probably the fastest single byte binary to decimal conversion for the PIC. It doesn't appear, at a glance, to be much larger than the loop methods and it does it in one pass; no loops. BAM! Done! Anyone who can understand how it works is smarter than I. Scotts notes: http://www.piclist.com/techref/method/math/radixs.htm --- James. > {Original Message removed}

> Jinx's post is a completion (separating the 10s from the 1s) of > www.piclist.com/techref/microchip/math/radix/b2bhp-8b3d.htm > by Scott Dattalo I should have attributed that The 16-bit convertor is pretty wizard too http://www.piclist.com/techref/microchip/math/radix/b2a-16b5a-rl.htm Both the 8-bit and 16-bit are used extensively for LCD, no problems (note gotos if porting to 18F) ;================================================ ; Convert 16-bit data to ASCII ;================================================ ;Input 16-bit number in hi:lo ;Output 5-byte ASCII array ascii16 nop radix dec ;base 10 movf fsr,w ;save fsr movwf f_temp clrf known_zero clrf temp movlw dg1 movwf fsr goto $+2 sub10k incf temp movlw 10000 & 255 subwf lo IFNDEF known_zero movlw 10000>>8 skpc movlw (10000>>8)+1 subwf hi ELSE rlf known_zero,w sublw (10000>>8)+1 subwf hi ENDIF bc sub10k call out_temp movlw 10 movwf temp add1K decf temp movlw 1000 & 255 addwf lo IFNDEF known_zero movlw 1000 >> 8 skpnc movlw (1000>>8)+1 addwf hi ELSE rlf known_zero,w addlw 1000>>8 addwf hi ENDIF bnc add1k call out_temp clrf temp movlw 100 goto $+2 sub100 incf temp subwf lo skpnc goto sub100 decf hi btfss hi,7 goto sub100 call out_temp movlw 10 movwf temp add10 decf temp addwf lo bnc add10 call out_temp ;convert and store call out_lo ;convert and store movf f_temp,w ;restore fsr movwf fsr radix hex return ;convert to ASCII and store out_temp movf temp,w addlw 0x30 movwf indf incf fsr return out_lo movf lo,w addlw 0x30 movwf indf return

On Wed, Mar 07, 2007 at 09:05:25AM +1100, John La Rooy wrote: > You could also use a lookup table if you have plenty of free space. It's also real easy with a couple of compressed tables too. A table to convert the upper hex nybble to its corresponding BCD and a decimal add adjust table to compensate for hex digits with no corresponding BCD digit should be enough. No more than 32 bytes. BAJ