Lookdown and Strange Return Value

TJHJ · 2008-05-30 21:14

Ok So I am trying to get this thermocouple thing to work and my most recent thought is to use a lookdown table useing the known voltages.
But it does odd things. To create a search method of sorts it adds until it finds a match, not the best, but I am ok with a premant rounding up solution.
The table is very large it has approx 1500 data points.

PUB Look(Vi) | Table  
 
 Table := 0
 Table := Lookdown(Vi:0,0039,0079,0119,0158,0198,0238,0277,0317,0357,0397,0397,0437,0477......
 Repeat 
    Vi := Vi+1
    Table := Lookdown(Vi:0,0039,0079,0119,0158,0198,0238,0277,0317,0357,0397,0397,0437,0477.....
    If Table > 0
      Quit     
Return Table-1

It works if I call Look(0037) or Look(37) It returns 1, as it should,or Look(0075) Gives 2.
But if I put in Look(0030) it returns 857. Or Look(0031) gives 82.
And Ideas what is going on here?

Thanks as always
TJ

Mike Green · 2008-05-30 22:15

This is a terrible way to search a table, especially one as large as this. I suggest you define your table using a DAT section. Assuming you've got values less than 32767, you could use one word for each entry and fill the last entry with $7FFF. You'd search it with

PUB look(v) : i ' Note: i is always initialized to zero
   repeat while v > myTable[noparse][[/noparse]i++]

The table starts with the first value greater or equal to v.

Phil Pilgrim (PhiPi) · 2008-05-30 22:40

TJ,

Where does the data in your table come from? If it's mathematically-defined, perhaps a formula (either continuous or piecewise continuous) would provide a more compact way to arrive at the desired result.

-Phil

Rayman · 2008-05-31 00:49

First, I don't think your first Lookdown statement does anything and can be removed.

I bet it's finding your values 30, 31 in the table somewhere... Try examining V1 during the iterations to see what's happening...

Who knows, you may have found a bug... Or, maybe theres a mistake in your table.

Or, maybe the memory space for the table is being corrupted somehow...

You might try a test look for the last entry in the table and see if that works right...

TJHJ · 2008-05-31 00:56

Here is the Calc program I came up with which I think is even worse.

CON 
A0 = -0.0176004136860
A1 =  0.0389212049750
A2 = 0.0000185587700320
A3 = -0.000000994575928740
A4 = 0.000000000318409457190
A5 = -0.000000000000560728448890
A6 = 0.000000000000000560750590590
A7 = -0.000000000000000000320207200030
A8 = 0.0000000000000000000000971511471520
A9 = -0.0000000000000000000000000121047212750
OBJ
F : "Float32"
ADC : "CD_LTC1298"
PUB Init'(AD1)
  f.Start
  'ADC.Start(AD1)
PUB DEGC(V) |  VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9, V1, V2, V3, V4, V5, V6, V7, V8, V9, AI1, AI2, AI3, AI4, AI5, AI6, AI7, AI8, AI9
  'V := F.FFloat(Vin)   ' Convert V to Float
  VP2 := F.Pow(V, F.FFloat(2))
  VP3 := F.Pow(V, F.FFloat(3))
  VP4 := F.Pow(V, F.FFloat(4))
  VP5 := F.Pow(V, F.FFloat(5))
  VP6 := F.Pow(V, F.FFloat(6))
  VP7 := F.Pow(V, F.FFloat(7))
  VP8 := F.Pow(V, F.FFloat(8))
  VP9 := F.Pow(V, F.FFloat(9))
 
  
  V1 := F.FMul(A1, V)
  V2 := F.Fmul(A2, VP2)
  V3 := F.Fmul(A3, VP3)
  V4 := F.Fmul(A4, VP4)
  V5 := F.Fmul(A5, VP5)
  V6 := F.Fmul(A6, VP6)
  V7 := F.Fmul(A7, VP7)
  V8 := F.Fmul(A8, VP8)
  V9 := F.Fmul(A8, VP9)
  
  AI1 := F.FAdd(A0, V1)
  AI2 := F.Fadd(AI1, V2)
  AI3 := F.Fadd(AI2, V3)
  AI4 := F.Fadd(AI3, V4)
  AI5 := F.FAdd(AI4, V5)
  AI6 := F.Fadd(AI5, V6)
  AI7 := F.FAdd(AI6, V7)
  AI8 := F.Fadd(AI7, V8)
  AI9 := F.Fadd(AI8, V9)
  Return F.FRound(AI9)

What Do you guys think?

So I ended up creating the table to 1500 degrees, even though they limit out at accuracy at somewhere around 1350.

Post Edited (TJHJ) : 5/31/2008 1:21:14 AM GMT

TJHJ · 2008-05-31 01:17

Rayman said...

bet it's finding your values 30, 31 in the table somewhere... Try examining V1 during the iterations to see what's happening...

It seems to me that it is finding values which contain 30, 31 ect. I counted out the 83 number and its 1231. Is this is how this should be working.

Rayman said...
First, I don't think your first Lookdown statement does anything and can be removed.

Yeah sorry I have been expermenting and was dumping the 1st lookup value out to a screen just to see if it was finding something before the actuall seach began. Its not needed but I was useing it for examination. Thanks for the help.

I hope it is not a bug, but on the upside I would know I am not insane.

Thanks for all the help as always.

TJ

Lookdown and Strange Return Value

Comments