Post by thread:16F84 Multiply 13bit by 0.0625

Grant Forest wrote: > Goodday, > Finally got my LM76 Sign Digital i2c sensor working > but now have to multiply the 13-bit two complement word by > 0.0625 (1 bit indicated 0.0625 degrees C) > Had a look in the archive but nothing applicable. > Anyone using a Lm76, and got around this multiply, > or where I can look for some fraction code. > Tks > Grant Forest Well, .0625 is 1/16 a right shift is divide by 2, 4 right shifts are a divide by 16. -- * | __O Thomas C. Sefranek spam_OUTtcsTakeThisOuTcmcorp.com |_-\<,_ Amateur Radio Operator: WA1RHP (*)/ (*) Bicycle mobile on 145.41, 448.625 MHz hamradio.cmcorp.com/inventory/Inventory.html http://www.harvardrepeater.org

> Finally got my LM76 Sign Digital i2c sensor working > but now have to multiply the 13-bit two complement word by > 0.0625 (1 bit indicated 0.0625 degrees C) Well, .0625 is 1/16 a right shift is divide by 2, 4 right shifts are a divide by 16. Aren't you glad you spent all that time getting a nice clean analog section of circuitry, worthy of a 13bit A-D converter, only to throw away four of the bits? Consider multiplying by 100 16ths, (100/16) to get the voltages as an integer value equal to the number of hundredths of degrees. 100/16s is a bit less than 8, so the result should still fit in 16 bits. One of the math wizards might suggest an algorithm (to start with, 100/16 = 25/4) that will let you escape going to a 24 bit int. (I wouldn't mind seeing a simultaneous multiply/divide algorithm for handling fractions like this in general! (ie where you're sure the final result will be less than N bytes, but where the intermediate values can exceed N bytes.)) BillW

At 08:53 AM 3/26/00 +1000, you wrote: >Goodday, >Finally got my LM76 Sign Digital i2c sensor working >but now have to multiply the 13-bit two complement word by >0.0625 (1 bit indicated 0.0625 degrees C) >Had a look in the archive but nothing applicable. >Anyone using a Lm76, and got around this multiply, >or where I can look for some fraction code. >Tks >Grant Forest > Looks like /16 here, but if you're interested in general in fractional arithmetic methods, my old TI TMS320 guides describe something called "Q-format". Extremely clever stuff. Can probably find it somewhere on the TI site. - Dan Michaels Oricom Technologies http://www.sni.net/~oricom ==========================

Grant Forest <.....PICLISTKILLspam@spam@MITVMA.MIT.EDU> wrote: > have to multiply the 13-bit two complement word by 0.062 and Thomas C. Sefranek <PICLISTKILLspamMITVMA.MIT.EDU> replied: > Well, .0625 is 1/16 > a right shift is divide by 2, > 4 right shifts are a divide by 16. Grant: If you just do 4 right-shifts, you'll end up with a 9-bit two's- complement number... Which is ok, I guess, but it'd probably be easier to work with the value if it were represented as a straight 8- bit number in the range [0-256], with a SEPARATE sign bit. To get THAT, do this: RRF HI RRF LO RRF HI RRF LO RRF HI RRF LO RRF HI RRF LO,W BTFSC HI,0 SUBLW 0 MOVWF LO After executing that code, LO will contain a number between 0 and 256 (representing the absolute value of the temperature, in degrees C), and bit 0 of HI will hold the sign of the temperature (1 = negative, 0 = positive). Bits 1-7 of HI will contain garbage; if that bothers you, you can mask them by adding: MOVLW 00000001B ANDWF HI -Andy === Andrew Warren - .....fastfwdKILLspam.....ix.netcom.com === Fast Forward Engineering - San Diego, California === http://www.geocities.com/SiliconValley/2499

Grant, In general, you may want to use a code generator at http://www.piclist.com/codegen/constdivmul to multiply an integer variable by a floating point constant(0.0625, 6.25, or whatever). It will work also in case you have a variable with both integer and fractional part(Q-format). In your case, results are (0 - is the least significant byte): ; Acc = Acc * 0.062500 ; ; ALGORITHM: ; Clear accumulator ; Add input / 16 to accumulator ; Move accumulator to result ; ; Error in constant approximation : 0.000000, % ; Input: Acc0 .. Acc1 (13 bits) ; Output: Acc0 .. Acc1 (9 bits) cblock Acc0 Acc1 endc clrc rrf Acc1, f rrf Acc0, f clrc rrf Acc1, f rrf Acc0, f clrc rrf Acc1, f rrf Acc0, f clrc rrf Acc1, f rrf Acc0, f ;---------------------------------------- and ; Acc = Acc * 6.250000 ; ; ALGORITHM: ; Clear accumulator ; Add input * 4 to accumulator ; Add input * 2 to accumulator ; Add input / 4 to accumulator ; Move accumulator to result ; ; Error in constant approximation : 0.000000, % ; Input: Acc0 .. Acc1 (13 bits) ; Output: Acc0 .. Acc1 (16 bits) cblock Acc0 Acc1 Temp0 Temp1 endc movf Acc0, w movwf Temp0 movf Acc1, w movwf Temp1 clrc rrf Acc1, f rrf Acc0, f clrc rrf Acc1, f rrf Acc0, f clrc rlf Temp0, f rlf Temp1, f movf Temp0, w addwf Acc0, f movf Temp1, w skpnc incfsz Temp1, w addwf Acc1, f clrc rlf Temp0, f rlf Temp1, f movf Temp0, w addwf Acc0, f movf Temp1, w skpnc incfsz Temp1, w addwf Acc1, f ;---------------------------------------- Nikolai On Sunday, March 26, 2000 Grant Forest wrote: > Goodday, > Finally got my LM76 Sign Digital i2c sensor working > but now have to multiply the 13-bit two complement word by > 0.0625 (1 bit indicated 0.0625 degrees C) > Had a look in the archive but nothing applicable. > Anyone using a Lm76, and got around this multiply, > or where I can look for some fraction code. > Tks > Grant Forest