Resource clash PWN and s/w uart on a 328l

[FFMan]

Now that I have the NPT code working I wanted to retro fit it to a previous clock project whereby i had to manually set the time, which is awkward as the display is just a single swept needle.

the needle is driven by PWM, but when i try and add the NPT routines on a s/w uart the rx of the uart doesn't appear to work, using pins 12 & 13. If i comment out the pwm init it's ok.

Is there a clash and is there a solution. I could close the pwm when i want to use the uart as it will only be used every 24 hours for a few seconds.

[FFMan]

I have checked for timer clashes using s/w uart and pwm16 and there doesn't appear to be one.

PWM16 uses timer1 and s/w uart uses timer2.

i'll do more testing, but the behaviour i see is that if i comment out hte openpwm statement the serial IO works, if i leave it in the i fail to receive a response from the wifi module.

[dkinzer]

[T]he behaviour i see is that if i comment out hte openpwm statement the serial IO works, if i leave it in the i fail to receive a response from the wifi module.

I'd like to try to reproduce this. Could you make a simple test case that has just the PWM and SW UART code in it?

[FFMan]

Try this - if you comment out the open pwn i get a reply, if its active i dont'

Hopefully not some muppetry on my part !

Code: Select all

option com1speed 115200

	public Const	snFast			as single=50.0
	public Const	snSlow			as single=10.0
	PUBLIC CONST	byChan			as byte=1

Sub Main()


	dim by_CommPort 					as byte	=7
	dim GPS_RX							as byte = 0
	dim GPS_TX							as byte = 0
	dim				by_Mask 			as byte
	dim 			GPSOComS(1 To 30) 	As Byte				'GPS comms
	dim	 			GPSIComS(1 To 90) 	As Byte				'GPS comms was 200
	dim				SDIn				as byte
	
	by_Mask=bx00001000			'non inverted, 8 N 1
	Call OpenQueue(GPSOComS, SizeOf(GPSOComS))
	Call OpenQueue(GPSIComS, SizeOf(GPSIComS))

	's/w uart testing
	Call definecom(3,12,13,by_mask)
	Call Opencom(3,9600,GPSIcoms,GPSOcoms)

	'Open the PWM
    'Call OpenPWM(byChan, snFast, zxFastPWM) 

	'Call DefineCom(by_CommPort,GPS_RX,GPS_TX,by_Mask)			h/w test
	'Call OpenCom(by_CommPort, 115200, GPSIComS, GPSOComS)		h/w test
	
	debug.print "Start"
	sleep (5.0)
	
	Call PutQueueStr(GPSOComs,"AT" & CHR(13) & CHR(10))

	debug.print "Read"
	while -1
		if getqueuecount(GPSIComs)>0 then
			Call GetQueue(GPSIcoms,SDIn,1)
			debug.print chr(SDIn);
			'call putqueuestr(GPSocoms,"!")
		end if

	wend
	
End Sub

[dkinzer]

Try this - if you comment out the open pwn i get a reply, if its active i dont'

I used the program below, modified slightly since I don't have the GPS unit. For this test, I use a terminal program on the PC to send/receive on the SW UART channel. The test program simply sends back whatever it receives and also displays the hex code of the received character.

I tried it initially on a ZX-328n and when I observed that it worked I switched to a ZX-328L where it seemed to work as well.

Code: Select all

option com1speed 115200
   public Const   snFast         as single=50.0
   public Const   snSlow         as single=10.0
   PUBLIC CONST   byChan         as byte=1

Sub Main()
   dim by_CommPort                as byte   =7
   dim GPS_RX                     as byte = 0
   dim GPS_TX                     as byte = 0
   dim by_Mask                    as byte
   dim GPSOComS(1 To 30)          As Byte            'GPS comms
   dim GPSIComS(1 To 90)          As Byte            'GPS comms was 200
   dim SDIn                       as byte
   
   by_Mask=bx00001000         'non inverted, 8 N 1
   Call OpenQueue(GPSOComS, SizeOf(GPSOComS))
   Call OpenQueue(GPSIComS, SizeOf(GPSIComS))

   's/w uart testing
   Call definecom(3,12,13,by_mask)
   Call Opencom(3,9600,GPSIcoms,GPSOcoms)

   'Open the PWM
   Call OpenPWM(byChan, snFast, zxFastPWM)

   'Call DefineCom(by_CommPort,GPS_RX,GPS_TX,by_Mask)         h/w test
   'Call OpenCom(by_CommPort, 115200, GPSIComS, GPSOComS)      h/w test
   
   debug.print "Start"
'   sleep (5.0)
   
'   Call PutQueueStr(GPSOComs,"AT" & CHR(13) & CHR(10))

   debug.print "Read"
   Do
      if getqueuecount(GPSIComs)>0 then
         Call GetQueue(GPSIcoms,SDIn,1)
         Call PutQueueByte(GPSOcoms, SDIn)
         debug.print "got character 0x"; CStrHex(SDIn)
'         call putqueuestr(GPSocoms,"!")
      end if
   Loop
   
End Sub 

[FFMan]

ok thats interesting - let me do some more elimination tests and see what i find.

thanks

[FFMan]

Don,

try as i might i can't make this work and i've spent like 4 hours on it. I stripped the processor board out of the project and went through checking all the connections and try PWN chan 2, com port different pins etc.

The I built a test 328n on the breadboard with just the minimal pins, XTAL & capacitors, VCC, gnd x 2 and the comms to the PC pins 1-3 and gnd.

I looped 12 & 13 together so that when its working i get the echo back.

I've tried this with 2 different 328s and a different psu, but my results are always the same.

If i open the pwm, i don't get the echo, if i comment it out i do.

What else could be going on. I'm using the latest download package.

[dkinzer]

What else could be going on.

I'd be happy to try your new minimal test case here with loopback if you'd provide it. I want to be sure I'm using exactly the same code that you are.

[FFMan]

It's the same test case as before, i just can't make it work on 2 test setups. stripped back to nothing just the pins mentioned. Don't need an external device, just a loop back and i don't get the echo, or the characters never get sent. I guess i could pin point which it is.

I can't understand what it could be, i've tried swapping everything even the psu.

[dkinzer]

It's the same test case as before, i just can't make it work on 2 test setups.

With the first test case you posted, I get "Start", "Read" and "AT" displaying in the debug window with pins 12 and 13 connected. However, the "AT" doesn't show up for five seconds due to the Sleep() call at line 33. At first, I thought that I was observing a replication of your issue but I eventually saw the "AT" confirming that it is working as expected. If I comment out the Sleep() call, the "AT" appears immediately.

[dkinzer]

If I comment out the Sleep() call, the "AT" appears immediately.

Oops. I didn't have the OpenPWM() commented out. I now see the behavior that you reported but I don't know why.

Curiously, if I move the OpenPWM() call to after the PutQueueStr() call it works as expected. Further, with the OpenPWM() call in the original position and the loop modified as shown below it also works.

Code: Select all

     While -1
         if getqueuecount(GPSIComs)>0 then
             Call GetQueue(GPSIcoms,SDIn,1)
             debug.print chr(SDIn);
             'call putqueuestr(GPSocoms,"!")
         else
             Call PutQueueStr(GPSOComs,"AT" & CHR(13) & CHR(10))
         end if
     Wend

[FFMan]

Interesting

The sleep (5.0) allows the wifi module to connect to the wifi and acquire an ip address, so in the main project the first retrieved message is a 'GOT IP' message.

In the main project I'd like the PWM to run all the time but the NPT to run only every 24 hours. I'm sure I could think of a work around based on your findings or is it your intention to investigate further ?

thanks

[dkinzer]

I'm sure I could think of a work around based on your findings or is it your intention to investigate further ?

The Sleep() call isn't the problem. Rather, it is some interaction between the PWM and the software UART or possibly the queue.

I do intend to investigate further with the goal of finding the cause of the problem and, if possible, a solution.

[FFMan]

it seems to be the transmit that is broken by the open pwm.

might be the rx as well but so far as i can tell AT never makes it out the tx pin

[FFMan]

if i play about with the order of the functions, open queue, open com, open pwm

then i get open in order:-

queue,com,pwm - fail

pwm,queue,com - ok

queue,pwm,com - ok