Disable transmit on serial port

[marcick]

Hi all,
on my PIC18f67K22 I use the serial port2 as a debug line to output comments during program execution.
So here and there I have an "Hserout2 ["I'm doing this",13,10]
Also I receive data from a terminal connected to serial port2 (I use Instant_Interrupt by the great Darrel R.I.P.).

Now I need that when some condictions occours, all the Hserout2 line are skipped and no strings come out from serial port2, while receiving continues to work. Lets' say this is a "ghost mode".
Then, I need that when I receive a particular string in the serial port2, the Hserout2 debug line is re-enabled and works again.

Look like it is simple by disabling the TX on the serial port2 -->> txsta2.5=0 but it doesnt't work and the program enter in a strange condition that I still have to understand. Looks like it freezes somewhere.

Do you have another smart idea on how to stop Hserout2 ?
p.s. I can't put a "If then" in every parto f the code where I use Hserout2, too much job.

[tumbleweed]

Look like it is simple by disabling the TX on the serial port2 -->> txsta2.5=0 but it doesnt't work and the program enter in a strange condition that I still have to understand. Looks like it freezes somewhere.

When you set TXEN = 0 (TXSTA2.5) that clears the TXIF status flag, and the first thing the library does is loop waiting for TXIF to be set so it can load the TXREG2, creating an infinite loop:

From PBPPIC18.LIB and PBPPI18L.LIB files:

HSEROUT2 movlb    15        ; Set bank select to 15 to pick up any SFRs not in Access bank
hserout2loop CLRWDT?        ; Keep Watchdog clear
        btfss   PIR3, TX2IF    ; Wait till ready
        bra     hserout2loop

You could try setting the TX pin to input mode using the TRISG.1 = 1, but all devices don't work the same so having SPEN = 1 may over-ride this. If you do that you would want to have a pullup on the TXD output pin.


You could always modify the library file(s) to check for TXEN:

HSEROUT2 movlb    15        ; Set bank select to 15 to pick up any SFRs not in Access bank
        btfss   TXSTA2, TXEN   ; **ADDED - check for TXEN set... return if clear
        goto    DUNN                 ; exit
hserout2loop CLRWDT?        ; Keep Watchdog clear
        btfss   PIR3, TX2IF    ; Wait till ready
        bra     hserout2loop

[marcick]

Wow, thanks for a so detailed explanation, now it's clear what happens.
I am not thrilled to modify the library. Maybe I could replace all the Hserout2s with a macro that tests a variable and decides whether to run Hserout2 or not. But maybe it's too complicated.....
Any other ideas?

[tumbleweed]

Did you try setting/clearing TRISG.1 when you want to disable/enable the port?
As I said, it might not work with this device... didn't try it, but it's worth a shot.

If you have the pins available you could also run the TXD output into one of the comparator inputs and then use the CMxOUT as the TXD pin.
Bit of a kludge since you'd have to setup and enable/disable the comparator output.

If it were me, I'd modify the library. You'd have to keep a note if you ever update in the future, though.

[marcick]

Did you try setting/clearing TRISG.1 when you want to disable/enable the port?

I didn't try yet. I will do all the experiments in the next days and be back here.
Thank you very much for now :-)

[tumbleweed]

Another option (if you don't mind changing the hdw and have a spare IO pin) is to add a resistor inline with the TXD pin (several K ohms), and connect the spare IO pin to the "output" side of the resistor.

To enable TX set the spare IO pin as an input (TRISx = 1), and to disable the TX output set the IO pin high (LAT = 1, TRIS = 0).

[marcick]

Ok I found the time to make some tests.

Setting Tx pin as an input does not work. Hserout2 continues to send data.

Modifying the library half works: setting Txsta2.5=0 stop the output and the program run normally, but then Txsta2.5=1 does not restart the output.

[tumbleweed]

I don't have a 67K22 handy, so I tried this with an 18F26K22.

It's dangerous to compare different devices, but one thing I noticed is that after re-enabling TXEN (setting TXSTA2.5=1), it can take a bit of time before the TXIF becomes set again. From what I can tell the time appears to vary with the baudrate setting, so maybe it's related somehow.

Try adding a several msec delay after setting TXEN and see if that fixes it, perhaps start with 10ms and work down from there (9600 baud is approx 1msec/char)

[marcick]

mmmmhhhh ... looks like it doesn't solve.
If TXIF weren't set, the program should be here forever

hserout2loop CLRWDT? ; Keep Watchdog clear
btfss PIR3, TX2IF ; Wait till ready
bra hserout2loop

and instead the program run normally after setting again TXSTA2.5=1 (I'm in a loop where I flash a led and I see it) but Hserout2 (in the same loop) no longer works

I show you my full code to do this test. I flash a led and wait for a char on the serial port to disable Hserout2. Then, after 5 seconds the Hserout2 should restart to work but not.

y=0
Main_Loop:
    Hserin2 50, ML00, [x]
    if x="z" then
        Hserout2[13,10,"zzz ... detected",13,10]
        txsta2.5=0
        y=1
    endif

    if y>0 then y=y+1

    if y>5 then
        Txsta2.5=1
        y=0
        pause 100
    endif

ML00:
    pause 1000
    toggle led
    Hserout2["loop ",#y,13,10]
    goto main_loop

[tumbleweed]

Try this and see if it has any effect...

; enable clear input overrun error
define HSER2_CLROERR 1

y=0
Main_Loop:
    Hserin2 50, ML00, [x]
    if x="z" then
        Hserout2[13,10,"zzz ... detected",13,10]
        ; wait for last char to finish and disable TX
        while TXSTA2.1 = 0 : wend       ; wait TRMT
        TXSTA2.5 = 0    ; clear TXEN
        y=1
    endif

    if y>0 then y=y+1

    if y>5 then
        TXSTA2.5 = 1    ; set TXEN
        ; wait for TXIF to re-assert
        while PIR3.4 = 0 : wend     ; wait TX2IF
        y=0
        pause 100
    endif

ML00:
    pause 1000
    toggle led
    Hserout2["loop ",#y,13,10]
    goto main_loop

[marcick]

Thanks a lot for your support but ...... it doesn't work ....

[tumbleweed]

That figures. Now I'm really curious as to why it doesn't work.

I thought I had a board with a 67K22 on it, but it seems not..
I do have some 87K22 hdw around somewhere... at least that one shares the same datasheet as the 67.
I'll have to do some digging...

[tumbleweed]

I finally got around to trying this on an 18F87K22, and the following code works.
The trick seems to be to wait for any chars in the TX buffer to be sent before enabling/disabling TXEN.
I did this as an output-only test so that no other code interferes with/clouds how it works.

; 18F87K22
; test enabling and disabling uart2 TXEN
; this assumes that HSEROUT2 routine in the PBPPIC18.LIB/PBPPI18L.LIB library
; has been change to add a check for TXEN:
;   HSEROUT2 movlb    15        ; Set bank select to 15 to pick up any SFRs not in Access bank
;        btfss   TXSTA2, TXEN   ; **ADDED - check for TXEN... return if not set
;        goto    DUNN           ; **ADDED
;   hserout2loop CLRWDT?        ; Keep Watchdog clear

define OSC 64
#define FOSC 64     ; set FOSC=OSC... used by set_intosc

; set baudrate
define HSER2_BAUD 115200
; enable clear input overrun error
define HSER2_CLROERR 1

; setup K22 intosc
call set_intosc

; loop forever... should see "12345678" over and over
; should never see any alpha chars (ie "ABCD")
call TX2_enable
while (1)
    hserout2["1234"]
    call TX2_disable
    hserout2["ABCD"]    ; tx disabled... should not see this
    call TX2_Enable
    hserout2["5678"]
wend

TX2_Disable:
    ; wait for any current TX to finish
    while TXSTA2.1 = 0 : wend       ; wait TRMT
    ; disable TX
    TXSTA2.5 = 0                    ; clear TXEN
    return
    
TX2_Enable:
    ; wait for any current TX to finish
    while TXSTA2.1 = 0 : wend       ; wait TRMT
    ; enable TX
    TXSTA2.5 = 1                    ; set TXEN
    ; wait for TXIF to re-assert
    while PIR3.4 = 0 : wend         ; wait TX2IF
    return

set_intosc:
    OSCTUNE = 0
    #if FOSC = 64
      OSCCON = $70      ; 16 + 4xPLL
      OSCTUNE.6 = 1     ; enable pll
    #endif
    #if FOSC = 32
      OSCCON = $60      ; 8 + 4xPLL
      OSCTUNE.6 = 1     ; enable pll
    #endif
    #if FOSC = 16
      OSCCON = $70
    #endif
    #if FOSC = 8
      OSCCON = $60
    #endif
    return

#CONFIG
 config RETEN = ON         ; Enabled
 config INTOSCSEL = HIGH   ; LF-INTOSC in High-power mode during Sleep
 config SOSCSEL = HIGH     ; High Power SOSC circuit selected
 config XINST = OFF        ; Disabled
 config FOSC = INTIO2      ; Internal RC oscillator
 config PLLCFG = OFF       ; Disabled
 config FCMEN = OFF        ; Disabled
 config IESO = OFF         ; Disabled
 config PWRTEN = OFF
 config BOREN = SBORDIS    ; Enabled in hardware SBOREN disabled
 config BORV = 3           ; 1.8V
 config BORPWR = ZPBORMV   ; ZPBORMV instead of BORMV is selected
 config WDTEN = OFF        ; WDT disabled in hardware; SWDTEN bit disabled
 config WDTPS = 1048576    ; 1:1048576
 config RTCOSC = SOSCREF   ; RTCC uses SOSC
 config EASHFT = ON        ; Address Shifting enabled
 config ABW = MM           ; 8-bit address bus
 config BW = 16            ; 16-bit external bus mode
 config WAIT = OFF         ; Disabled
 config CCP2MX = PORTC     ; RC1
 config ECCPMX = PORTE     ; Enhanced CCP1/3 [P1B/P1C/P3B/P3C] muxed with RE6/RE5/RE4/RE3
 config MSSPMSK = MSK7     ; 7 Bit address masking mode
 config MCLRE = ON         ; MCLR Enabled RG5 Disabled
 config STVREN = ON        ; Enabled
 config BBSIZ = BB2K       ; 2K word Boot Block size
 config DEBUG = OFF        ; Disabled
#ENDCONFIG

[marcick]

Wow, amazing, it works perfectly !
Thank you very much guy, would like me to offer you a beer ? (paypal)

[tumbleweed]

Great. I was hoping it would work on the 67K22 since they're basically the same. Should probably work on most others too as long as you account for where the PIRx TXIF is located. Plus, some of the other 18F's use different library files.

A virtual beer is fine...