Post by thread:Urgent: Random No Generator Needed

Tim Kerby <spam_OUTPICLISTTakeThisOuTMITVMA.MIT.EDU> wrote: > I am about to reach a deadline on a project at school (for the > engineering scheme scotland) and my random no generation isnt > working. It must take the tmr0 or any internal part of the chip as > i dont have lines to spare and give out a number between 1 and 12 > (inclusive). Tim: To select a TMR0-based number between 1 and 12, inclusive, do something like this: LOOP: MOVF TMR0,W ANDLW 00001111B ADDLW -12 BC LOOP ADDLW 13 After executing the fragment above, W will contain a number in the range [1-12]. This is just about the worst possible way to get random numbers -- the distribution is VERY skewed toward the low end of the range, and the whole thing's based on a deterministic timer -- but it's quick (so long as you haven't enabled a large TMR0 prescaler), easy, and doesn't require any additional file registers. -Andy === Andrew Warren - .....fastfwdKILLspam@spam@ix.netcom.com === === Fast Forward Engineering - Vista, California === === === === Did the information in this post help you? Consider === === contributing to the PICLIST Fund. Details are at: === === http://www.geocities.com/SiliconValley/2499/fund.html ===

The prescaler is at maximum but the clock is 20MHz. I will include a loop to give a new number if the number chosen was the same as the last which should help as only an even distribution of nos is needed and not a true random. The code looks good. Mine got the timer value and counted subtractions of a number and the code was huge. Is it the timer which is skewed or the way the code is implemented? Thanks Tim {Quote hidden}>Tim: > >To select a TMR0-based number between 1 and 12, inclusive, do >something like this: > > LOOP: MOVF TMR0,W > ANDLW 00001111B > ADDLW -12 > BC LOOP > ADDLW 13 > >After executing the fragment above, W will contain a number in the >range [1-12]. This is just about the worst possible way to get >random numbers -- the distribution is VERY skewed toward the low end >of the range, and the whole thing's based on a deterministic timer -- >but it's quick (so long as you haven't enabled a large TMR0 >prescaler), easy, and doesn't require any additional file registers. > >-Andy > >=== Andrew Warren - fastfwdKILLspamix.netcom.com === >=== Fast Forward Engineering - Vista, California === >=== === >=== Did the information in this post help you? Consider === >=== contributing to the PICLIST Fund. Details are at: === >=== http://www.geocities.com/SiliconValley/2499/fund.html === >

Tim Kerby <.....PICLISTKILLspam.....MITVMA.MIT.EDU> wrote: > The prescaler is at maximum but the clock is 20MHz. I will include > a loop to give a new number if the number chosen was the same as the > last which should help as only an even distribution of nos is needed > and not a true random. The code looks good. Mine got the timer > value and counted subtractions of a number and the code was huge. > Is it the timer which is skewed or the way the code is implemented? Tim: It's the way that I implemented the code... Guess I should have included some comments. Here's a documented version: LOOP: MOVF TMR0,W ;Grab the timer value (0-255). ANDLW 00001111B ;Mask off all but the lower nibble, ;so W is now in the range [0-15]. ADDLW -12 ;Is W > 11? BC LOOP ;If so, loop back and try again; ;TMR0 will soon increment to the ;point where its low nibble will be ;equal to 0 again. ADDLW 13 ;Otherwise, W is in the range ;[-12 to -1]; add 13 to bring it ;into the range [1-12]. As you can see, the returned value has an even distribution from 2 to 12, but it'll contain "1" 5 times as often as any other number. If you can spare a register, you can get a more-even distribution by using a linear-feedback shift-register as your basis, rather than TMR0. Here's a method that uses Marv Isaacman's LFSR code from my company's web page: RANDOM EQU [some file register] .... MOVLW 1 ;Seed the random-number generator MOVWF RANDOM ;(your program only needs to do this ;once). .... LOOP: MOVLW 01DH ;RANDOM holds the next number in the CLRC ;pseudo-random sequence. RLF RANDOM ; SKPNC ; XORWF RANDOM ; MOVF RANDOM,W ;Grab the random number (0-255). ANDLW 00001111B ;Mask off all but the lower nibble, ;so W is now in the range [0-15]. ADDLW -12 ;Is W > 11? BC LOOP ;If so, loop back and try again with ;the next number in the sequence. ADDLW 13 ;Otherwise, W is in the range ;[-12 to -1]; add 13 to bring it ;into the range [1-12]. ; At this point, W is in the range [1-12]. Hope this helps... -Andy === Andrew Warren - EraseMEfastfwdspam_OUTTakeThisOuTix.netcom.com === === Fast Forward Engineering - Vista, California === === === === Did the information in this post help you? Consider === === contributing to the PICLIST Fund. Details are at: === === http://www.geocities.com/SiliconValley/2499/fund.html ===