Need help with Flexis

HumanOzz

Hey guys,

I need some good suggestions on how to solve the things i'm doing in my project.
Well basically, i want to use more than 8 Flexi sensors in 1 propeller chip and the problems with this is.
Flexi sensors use RCTIME object right, and when used in the background mode, its making use of a new cog with it.
What i mean is, 1 flexi sensor uses 1 cog. So i'm wasting a cog.
But if we use the RCTIME in the forground mode, the main program waits for the change in the pin state because of the waitpeq command and from what i read, the propeller goes into a low-powered mode.
So, from what i understand, if the main program waits for the change in the state of this one pin, then it wont be able to detect any changes in the other pins that i have connected with flexi sensors right.

Is there some ways to do this?
I have thought of using more than 1 more propeller chips, where the other chip handles the flexi sensors in the background mode and send the data to the main chip. But then, its kind of a waste of chips right.
I need a good idea on how to efficiently use the flexi sensors without using more than 1 propeller chip.
Can you help guys?
Thanks

Ariba

You can measure all the Flex Sensors one after the other.
If you don't want that the main cog needs to wait until all the sensors are measured then do this in a separate cog.
The main cog can then read the values from the FlexValue array.

You can spare a lot of components with the attached circuit.
The code to measure all the sensors looks like that:

CON
 MsrPin  = 0     'Pin settings
 Flex1   = 1
 Flex2   = 2
 ...
VAR
 long FlexValue[noparse][[/noparse]8]
OBJ
 rc : "rctime"

PUB
 ...
 dira[noparse][[/noparse]Flex1] := 1
 rc.rctime(MsrPin,1,@FlexValue[noparse][[/noparse]0])
 dira[noparse][[/noparse]Flex1] := 0
 dira[noparse][[/noparse]Flex2] := 1
 rc.rctime(MsrPin,1,@FlexValue[noparse][[/noparse] 1])
 dira[noparse][[/noparse]Flex2] := 0
 dira[noparse][[/noparse]Flex3] := 1
 rc.rctime(MsrPin,1,@FlexValue[noparse][[/noparse] 2])
 dira[noparse][[/noparse]Flex3] := 0
 ...

This writes the results in the FlexValue array.

Andy

Post Edited (Ariba) : 8/1/2010 11:43:52 AM GMT

HumanOzz

Thanks for reply ariba,
hmm, this looks good, but dont you think the main program stops the assigned cog too fast for the readings to get to its real value?
i mean, if you can see from the program,
The gap between rctime.start and another rctime.start is kindof short.

Ariba

It may be you have seen my post to early, I have it edited after some minutes.
If you use rctime in Foreground mode then it waits until one measurement is done.

The voltage at the Capacitor looks like that:


And here the variation to do all measuring in background with a second cog:

CON
 MsrPin  = 0     'Pin settings
 Flex1   = 1
 Flex2   = 2
 ...
VAR
 long FlexValue[noparse][[/noparse]8]
 long stack[noparse][[/noparse]20] 
OBJ
 rc : "rctime"

PUB main
 cognew(flexMsr,@stack)
 ...
 val = FlexValue[noparse][[/noparse] 1]   'Read a value
 ..

PUB flexMsr
 repeat
   dira[noparse][[/noparse]Flex1] := 1
   rc.rctime(MsrPin,1,@FlexValue[noparse][[/noparse]0])
   dira[noparse][[/noparse]Flex1] := 0
   dira[noparse][[/noparse]Flex2] := 1
   rc.rctime(MsrPin,1,@FlexValue[noparse][[/noparse] 1])
   dira[noparse][[/noparse]Flex2] := 0
   dira[noparse][[/noparse]Flex3] := 1
   rc.rctime(MsrPin,1,@FlexValue[noparse][[/noparse] 2])
   dira[noparse][[/noparse]Flex3] := 0
   ...

Andy

HumanOzz

yeah, its true that if we use rc.rctime, we'll wait until the measurement is done.
But then, if for example, we are at this point "rc.rctime(MsrPin,1,@FlexValue[noparse][[/noparse]0])",
and lets say, that flexi sensor involved doesnt detect any changes, the program will wait at this point until theres a change in state right? (which happen if we place some pressure at that flexi sensor). Because of the waitpeq.
What i mean is, if the program is at that point, waiting and waiting for the change of state, it wont be able to detect any changes that happens at the other flexis right?
What do you think?

mpark

The waitpeq just waits until the capacitor is discharged. It is not waiting on a change of state in the sensor. Perhaps you are confused by the comments about "state" in rctime.spin. Those refer to the capacitor's state (i.e., charged or discharged).

Ariba

OK, I see your problem. The Flexi Sensor has ~5 MOhm when not pressed and ~20 kOhm when pressed.
So you need a rctime with timeout to not slow it down to much.

Here a possible code:

CON
 MsrPin  = 0     'Pin settings
 Flex1   = 1
 Flex2   = 2
 ...
 TMOUT   = ?     '~10us resolution
VAR
 long FlexValue[noparse][[/noparse]8]
 long stack[noparse][[/noparse]20] 

PUB main
 cognew(flexMsr,@stack)
 ...
 val = FlexValue[noparse][[/noparse] 1]   'Read a value
 ..

PUB flexMsr
 repeat
   dira[noparse][[/noparse]Flex1] := 1
   FlexValue[noparse][[/noparse]0] := rctime(MsrPin,1,TMOUT)
   dira[noparse][[/noparse]Flex1] := 0
   dira[noparse][[/noparse]Flex2] := 1
   FlexValue[noparse][[/noparse] 1] := rctime(MsrPin,1,TMOUT)
   dira[noparse][[/noparse]Flex2] := 0
   dira[noparse][[/noparse]Flex3] := 1
   FlexValue[noparse][[/noparse] 2] := rctime(MsrPin,1,TMOUT)
   dira[noparse][[/noparse]Flex3] := 0
   ...

PUB rctime(pin,state,timeout) : time
  outa[noparse][[/noparse]pin] := state                   'make I/O an output in the state
  dira[noparse][[/noparse]pin] := 1                               
  waitcnt(clkfreq/1000 + cnt)          'pause for 1mS to charge cap
  dira[noparse][[/noparse]pin] := 0                       'make I/O an input
  time := cnt                          'grab clock tick counter value
  repeat until ina[noparse][[/noparse]pin]<>state or timeout==0   'wait until pin goes into opposite state
    timout--                           ' or timout
  if timeout==0
    return -1                          '
  time := cnt - time                   'clock cycles passed state changed
  time := (time-1600) >> 4             'offset adjustment and scale

This is not tested! It's a bit dangerous to post such a code without testing, but I have not the time now.
Perhaps you need to adjust the offset adjustment. The timout is in ~10 microseconds steps (the repeat loop time).

BTW: I think there is a bug in the original rctime object at the watpeq. With state = 0 and Pin > 0 it will not work.
WAITPEQ((1-State)<<Pin, |<Pin, 0) would be better.

Andy

HumanOzz

Yeah thats right, but for this case, dont you think that the capacitor wont be discharged if the flexi doesnt detect anything at all. So its like, its waiting for some pressure to be applied then it will proceed to the next flexi sensor.

HumanOzz

Thanks ariba, i think thats the thing i'm looking for. [noparse]:)[/noparse]