handling Large Numbers

[Peter1960]

OK, Brain Freeze !! (At the moment I cannot even get a slope formula right !)

This is what I am trying to do

x var word [2]


ADCIN 6,x

x = x*2923-1343923

x1 = div32 1000

assembly says value truncated

it only allows to max ~63000

help !

Pete

[skimask]

OK, Brain Freeze !! (At the moment I cannot even get a slope formula right !)
This is what I am trying to do
x var word [2]
ADCIN 6,x
x = x*2923-1343923
x1 = div32 1000
assembly says value truncated
it only allows to max ~63000
help !
Pete

Page 31 of the manual, 4th line down...

[Peter1960]

Page 31 4th line down in mine says

** Top 16 Bits of Multiplication

I also found the DIV32, however ?

May I inquire what you were saying, please

Pete

[skimask]

Ok, maybe we've got different versions of the manual, different sections on different pages...I was referring to Section 4.17, but I'm pulling ALL of this straight from the manual:

x var word [2] ---- space between 'word' and '[2]' doesn't work.

ADCIN 6,x ---- this might work, if you set the configuration and defines correctly, which I don't see.

x = x*2923-1343923
---this won't work at all. For one thing, 'x' can't be any larger than 22 at the start of the operation because it'll be over the 16 bit limit. 1,343,923 is actually a 21 bit number. PBP only operates on 16 bit numbers. You might be wanting to do a 16x16=32bit number multiply operation, but you aren't going to do it with a '*'. And even if you did use it correctly, you couldn't subtract 1,343,923 from the result because, again, 1,343,923 is a number that can't be held in 16 bits.

x1 = div32 1000 --- inconsistent usage with the correct way of using it

assembly says value truncated
it only allows to max ~63000 ----- because PBP doesn't deal with numbers larger than 16 bits (65535)...

[Darrel Taylor]

Peter,

It looks like the original formula would have been

y = 0.2923x - 134.39

With a few extra decimal places.

But since you will end up with an integer in the end, you can just round off to -134 for the b parameter (y=mX + b).

For the m parameter, 0.2923 is 2923 ten thousanths. Which is an easy way to remember that you need to divide by 10,000 afterwards.

So you multiply the A/D reading times 2923.
Then divide it by 10,000
Like this...

X      VAR WORD
Y      VAR WORD

ADCIN 6,X

Y = X * 2923
Y = DIV32 10000
Y = Y - 134

I think that's about as close as you can get with 16/31 bit math.

HTH,

[Peter1960]

Hi Daryl,
Must have been your telepathy, as I did precicly that, ie

div32 first THEN subtract

ADCIN 7,y

Pause 100 ' adc settling

x1 = y

y = y*385

y = div32 1000

y = y - 5088

y1=y1+y ' Loop to average results

If x = 9 then goto outtwo

x=x+1

Pause 100

Goto Ambient

Outtwo: Debug "A raw= ",#x1,$0D,$0A

y=y1/1000

Y1= y1//1000

Y1 = Y1/100

Debug "A calc= ",#y,".",#y1,$0D,$0A

x=0
y=0
y1=0

now results are JUST Fine !!!!
thanks Steve and Daryll

kindregards

pete

sorry Steve is meant to say Skimask

[mister_e]

sorry Steve is meant to say Skimask

Ouf! Thank you very much. I have some reputation to preserve

thanks Steve and Daryll

Who's Daryll now? i know the OLD movie, a magician and an OLD guy, but the OLD guy name is DarrEl

Just kidding.

[Darrel Taylor]

Who's Daryll now?

That's my brother Daryll.
Then there's my other brother Darryl.
<br>