Searching \ for 'Look up table size' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page:
Search entire site for: 'Look up table size'.

Truncated match.
PICList Thread
'Look up table size'
1995\03\15@091355 by ail:walkerr

Andrew Warren <spam_OUTfastfwdTakeThisOuTspamIX.NETCOM.COM> wrote:

>At the risk of offending the Keeper of the List, the lookup table in
>question is not "huge".  At worst, it'd consume about 32 bytes, with
>(maybe) another 8 used for interpolation between entries in the table.


This one caught my attention. How do you interpolate between two table
points with 8 bytes of code?? Am I missing something here?


1995\03\15@221351 by Andrew Warren

Roy Walker ( wrote:

> This one caught my attention. How do you interpolate between two table
>points with 8 bytes of code?? Am I missing something here?


Yeah, probably.  Off the top of my head, I'd do this thermistor
thing by drawing the voltage vs. temperature curve, splitting it
into some small ( < 10, probably... 8 would be a good choice) number of
sections, each of which would be nearly linear, then just putting the
slopes of the linear approximations of those sections in the lookup
table.  If I needed speed, I'd put the endpoints of the lines in the
table, too.

Finding a point somewhere on a line, given another point on the line and
a slope, may take even less than 8 instructions.


Andrew Warren -
Fast Forward Engineering, Vista, California

1995\03\16@220125 by Henri

Hi there,
because the discussion about solving the Steinhart equation and a
lookup table style solution isn't stopped yet I like to add
another point of view.
The thing I like to point out is, that the ADC AD7714 is a
_24_ bit ADC. And so far I know are some people using the whole
resolution they can get from the part.
To say it first: I haven't a solution, but I think to use an 8bit
LUT won't do any good when using a 24bit seed in the Steinhart
equation. In addition to that one would have a huge need for
LUT places in order not to get lost the accuracy.
Having a LUT place for every ADC value would cause a need for
16.777.216 (2^24) values. Even the idea that looks up every 10th
place and then interpolates the values in between is not much better
(for this special case).
So, wait I'm trying to do is not to anoy the "get you a LUT" folks,
but I wonder if anybody has another idea.
I'm really interessted in this 'cause I'm using the AD7714, too,
but max. 16bits (data rate vs. resolution).
Right know I start to think about the floating point math lib
of AN526. Does anybody of you have any good/bad expirience with
using floating point on a 16C5X?
Nowadays I'm starting to use a Pt100 thermistor. In a small
section this thing is quasi-linear, so the only thing I would
have to do is a scalation (calibration) from the ADC value (16bit)
to the corresponding temperature value (1 or 2 numbers behind the
decimal point). The math I need for that is pretty easy, but it has to
be either floating or fixed point. In some very far future I
might start to add a thermo coupple to the ADC+PIC. Then
I would have to perform the could junction compensation in the
PIC, too. That means, I must deal with some values raised to some
powers (2,3,4 ..).
Does anybody have a general idea/solution to solve those kind
of equations with the nessecary accuracy with the help of
a PIC 16C5X || 16CXX ?
I would be really glad to hear about that.

tnx in advance

=============> The sanest place is still behind the trigger <===============
||         Henri Schultze        || Magdeburg D-39122 Alt-Fermersleben 88 ||
||     ||          The Cracker Company          ||
=================> in a world of compromise, some don't <===================

More... (looser matching)
- Last day of these posts
- In 1995 , 1996 only
- Today
- New search...