Stange behaviour of OUTA....

Joerg · 2008-01-11 21:04

After some starting difficulties and some very helpful tips from forum members, here something i am wondering about:

If i put data on outa an am using a variable direct the result makes sense to me; if i use the variable via pointer the bits are reversed:

i.e. variable N := 13 Result direct: Bit0=1, Bit1=0, Bit2=1, Bit3=1, Bit4..7=0
Result via pointer Bit0..3=0, Bit4=1, Bit5=1, Bit6=0, bit7=1

Here a program with this behaviour:

'test2.spin
'

VAR byte N
    long Stack[noparse][[/noparse]10]


PUB test2
  dira[noparse][[/noparse]0..7]~~
  N := 0
  cognew(outx(@N), @Stack)

  repeat
    outa[noparse][[/noparse]7..0] := N
    waitcnt(clkfreq / 1_000_000 * 10_000 + cnt)
    N := ina[noparse][[/noparse]11..8]

 
PUB outx(AdrInVar)
  dira[noparse][[/noparse]23..16]~~

  repeat
    outa[noparse][[/noparse]23..16] := long[noparse][[/noparse]AdrInVar]

Does anybody know why this happens?

Saluti Joerg

Phil Pilgrim (PhiPi) · 2008-01-11 21:21

It shouldn't matter, but since N is a byte variable, have you tried it with byte[noparse][[/noparse]AdrInVar]?

-Phil

Joerg · 2008-01-11 21:43

Yes i did!
I have tried to change form byte to long, thats why long[noparse][[/noparse]AdrInVar] shows up in the code.

Thank you Phil.

Buona notte Joerg

deSilva · 2008-01-11 21:55

Ths would not change anything as the Propeller is a little endian machine...
Joerg, can it be that you are just LOOKING upside down? Pins 16..24 are at the upper side of the Prop...

Beau Schwabe · 2008-01-11 22:03

Saluti Joerg,

is it because the direction you are assigning to your LED's reads...

[noparse][[/noparse]23..16]

...instead of...

[noparse][[/noparse]16..23]

... ?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

deSilva · 2008-01-11 22:08

No, as he used the inverted form for
OUTA[noparse][[/noparse]7..0] := N
as well..

Phil Pilgrim (PhiPi) · 2008-01-12 05:17

I tried the following, slightly modified version of your program (skipped reading the pins) and didn't see any bit inversion happening:

[b]CON[/b]

  [b]_clkmode[/b] = [b]xtal[/b]1 + [b]pll[/b]16x
  [b]_xinfreq[/b] = 5_000_000

'test2.spin
'

[b]VAR[/b] [b]byte[/b] N
    [b]long[/b] Stack[noparse][[/noparse]10&#093;


[b]PUB[/b] [b]test[/b]2
  [b]dira[/b][noparse][[/noparse]0..7&#093;~~
  N := 0
  [b]cognew[/b](outx(@N), @Stack)

  [b]repeat[/b]
    [b]outa[/b][noparse][[/noparse]7..0&#093; := N
    [b]waitcnt[/b](clkfreq / 1_000_000 * 10_000 + [b]cnt[/b])
    N := 13

 
[b]PUB[/b] outx(AdrInVar)
  [b]dira[/b][noparse][[/noparse]23..16&#093;~~

  [b]repeat[/b]
    [b]outa[/b][noparse][[/noparse]23..16&#093; := [b]long[/b][noparse][[/noparse]AdrInVar&#093;

-Phil

Peter Jakacki · 2008-01-12 06:40

Joerg,

I haven't really read this thread properly but isn't it a problem if you declare a byte variable before a long as the longs need to be word aligned. What would happen I think is that the stack would use the same address as the byte. Declare longs, then words, then bytes and you won't get any alignment problems.

*Peter*

stevenmess2004 · 2008-01-12 10:19

Shouldn't be. The spin compiler automatically aligns variables correctly. And long variables need to be long aligned not word aligned.

Steven

deSilva · 2008-01-12 12:14

Peter is fundamentally right; but teh compiler sorts this out outomatically in exactly that way. And it should not be a problem in the case of the posted code with just one byte only in it.

VAR byte N
    long Stack[noparse][[/noparse]10]

This here will give you more issues:

VAR byte X, Y, N
    long Stack[noparse][[/noparse]10]

I still suspect - as yesterday - Joerg was just looking upside down

Post Edited (deSilva) : 1/12/2008 12:19:23 PM GMT

Joerg · 2008-01-12 13:41

And deSilva is right!

I simply messed up P16..P23 -> PC7..PC0 and P24..P31 -> PD7..PD0

Thanks anyway for all the tips!

I will modify my MCU module for UMDL right now!

Saluti Joerg

Joerg · 2008-01-13 15:02

Here the new Propeller module for UMDL.

The doc can be found here: www.systech-gmbh.ch -> UMDL -> Parallax -> P8X32A-D40

Saluti Joerg

deSilva · 2008-01-13 17:16

Joerg, do I see it right that the main difference between this board and the Parallax Proto Board is a DIN 41622 connector?

Joerg · 2008-01-13 18:07

Dear deSilva

No! The UMDL system (formerly called HC(S)08 system) uses a simple 2x36pin header (72pol, 0.1"). This is enough for 8 8bit ports (as most other MCU's are using a port nomenclature). The MCU modules are fitting on base plate (with 10 pin headers and additional stuff). The original idea was to create a system for different MCU's (different families and different manufacturers).
So i can change the MCU if for some reason the chosen one does not fit anymore. The system is totally modular and based on the 10 pin header philosophy. So i have developed RS232 interfaces as well as stepper motor drivers and a lot more modules (and also a lot more in mind). But this is only one part of the story: Each module (if once tested!!!!!!!!!) has the schematics ready for being used in the final design. And this is true for the software also: My EBS08 (Einfaches Betriebs System 08) is also modular, consisting on a TIC generator for controlling simple tasks and a lot of modules around. They are (at the time being) written in ASM or C for the Freescale HC(S)08 MCU's, but step by step i will write these modules for the Propeller chip.
In fact i have realized a lot of my projects using UMDL and EBS08 for hardware and software development and i have used the basic modules to make the final design. This is a fast and secure approach and i just LOVE it.

Saluti Joerg

PS. If somebody is interested have a look here: www.systech-gmbh.ch -> UMDL

Stange behaviour of OUTA....

Comments