Question about FullDuplexSerial (from obex by Chip Gracey, Jeff Martin)

marzec309

Hi guys,

At the risk of looking stupid . I'm have a little problem with a chunk of code.

repeat 6
      outa[16]~~
      outa[16]~
      repeat until tmp1 >= $00
        outa[3]~~
        outa[3]~
        repeat 
          waitcnt(cnt + clkfreq / 10000)
        until ser.rxcount  > 0
        tmp1 := ser.rx
      Buff[tmp0] := tmp1
      tmp0++

I can't seem to hold the program in the inner most loop " repeat until ser.rxcount > 0" It doesn't seem to matter how I code it. the program will jump the loop after 1mS. Even with out the waitcnt it still jumps ship :frown:. I'm trying to wait for a 6 byte serial stream, that is on a 50mS interval. If I could just sync to this serial packet, I would have a bunch of time for other fun stuff. Any ideas what I'm doing wrong?

kuroneko

What is this rxcount method you speak of? Said object doesn't have one of those ...

Also, you're using tmp1 >= $00 which isn't a comparison (greater or equal). In SPIN that's an assignment (like +=). You want tmp1 => $00.

marzec309

As pulled from my version of FULL DUPLEX SERIAL, I don't ever remember adding it.

PUB rxcount : count

'' Check for bytes in receivebuffer (never waits)
'' returns the number of bytes 

  count := (rx_head - rx_tail) & 15
  if count == 0
    count := -1

kuroneko

I'm still a bit fuzzy as to what the exact problem is. Did you fix the operator issue? Also, not having this method available I used something like this:

CON
  _clkmode = XTAL1|PLL16X
  _xinfreq = 5_000_000

OBJ
  serial: "FullDuplexSerial"

VAR
  byte  Buff[32]
  
PUB null | tmp0, tmp1

  serial.start(31, 30, %0000, 115200)
  
  dira[16]~~
  dira[17]~~

  tmp0~
  
  repeat 6
    outa[16]~~
    outa[16]~
    repeat
      outa[17]~~
      outa[17]~
      waitcnt(cnt + clkfreq / 10000)
    until (tmp1 := serial.rxcheck) <> -1
    Buff[tmp0++] := tmp1

  repeat tmp0 from 0 to 5
    serial.tx(Buff[tmp0])
    
DAT

This demo stays in the inner loop until it gets a character then breaks out, stores it and repeats. After having collected 6 characters it sends them back through the serial link. HTH

marzec309

I give your use of rxcheck a try and thanks for spotting my operator typo. My problem is that my code as posted earlier, wouldn't wait in the loop with rxcount. it seem to time out @ 1mS, I'm debugging with a logic analyzer, thus explaining the outa[] in each repeat cycle. It would enter the inner most repeat and stay put for ~1mS then drop out before it receive a byte.

Thanks

kuroneko

I admit the behaviour is puzzling. Do you have a way of tracing rxcount (to serial) when it apparently forces the loop to exit?

For the sake of it I added an rxcount method and ran this test

repeat
    !outa[16]
    waitcnt(cnt + clkfreq / 10)
  until serial.rxcount > 0

Which stays in that loop until I press a key.

Does your serial link work otherwise, i.e. do you get what you expect from normal interactions with a terminal?

marzec309

I found the gremlin causing my troubles. After trying your code, the loop was still falling through prematurely. After, I started to dig around else where, found the problem in the call to serial.start(rx,tx,mode,baud), i was setting mode with 1, not realizing it wanted four bits, like you set it on your call. No more falling through the loop YAY.

thanks Kuroneko