Post by thread:[PIC]: Max bit bang serial on 4Mhz 16F876

Folks, Before I go off and waste a bunch of time trying to do the impossible (but hey it may be fun), does anyone have any advice on what is the fastest half duplex rs323 comms I can expect to get using software only on a 4MHz 16f876 (using a resonator not a crystal)? I'm currently using the code created by Andy N1YEW and posted on piclist.com and have this working fine at 19200 baud, but if possible I would like to run a little faster. I know I should probably be using the built in hardware but the board layout (not mine) will not allow it. I could also switch to a higher speed but again I would like to avoid changes to the board. The way things are I can probably dedicate the entire cpu to the serial code while transmitting/receiving with the exception of a TMR0 roll over interrupt that has been coded with minimal register save and just a simple increment of a counter (approx 5 instructions long), the interrupt occurs every 65535 instructions (TMR0 is using divide by 256 pre-scaler). Thanks for any help Andy -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

It depends a lot on what constraints there are from other requirements of your system - rates as high as 115K2 are possible at 4MHz if you don't mind it using up almost all your CPU time doing serial comms. On Tue, 25 Mar 2003 23:40:18 -0000, you wrote: {Quote hidden}>Folks, >Before I go off and waste a bunch of time trying to do the impossible (but >hey it may be fun), does anyone have any advice on what is the fastest half >duplex rs323 comms I can expect to get using software only on a 4MHz 16f876 >(using a resonator not a crystal)? > >I'm currently using the code created by Andy N1YEW and posted on piclist.com >and have this working fine at 19200 baud, but if possible I would like to >run a little faster. I know I should probably be using the built in hardware >but the board layout (not mine) will not allow it. I could also switch to a >higher speed but again I would like to avoid changes to the board. The way >things are I can probably dedicate the entire cpu to the serial code while >transmitting/receiving with the exception of a TMR0 roll over interrupt that >has been coded with minimal register save and just a simple increment of a >counter (approx 5 instructions long), the interrupt occurs every 65535 >instructions (TMR0 is using divide by 256 pre-scaler). > >Thanks for any help > >Andy -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

On Tue, 25 Mar 2003, Andy Shaw wrote: > Folks, > Before I go off and waste a bunch of time trying to do the impossible (but > hey it may be fun), does anyone have any advice on what is the fastest half > duplex rs323 comms I can expect to get using software only on a 4MHz 16f876 > (using a resonator not a crystal)? > > I'm currently using the code created by Andy N1YEW and posted on piclist.com > and have this working fine at 19200 baud, but if possible I would like to > run a little faster. I know I should probably be using the built in hardware > but the board layout (not mine) will not allow it. I could also switch to a > higher speed but again I would like to avoid changes to the board. The way > things are I can probably dedicate the entire cpu to the serial code while > transmitting/receiving with the exception of a TMR0 roll over interrupt that > has been coded with minimal register save and just a simple increment of a > counter (approx 5 instructions long), the interrupt occurs every 65535 > instructions (TMR0 is using divide by 256 pre-scaler). Receiving will be harder, but transmitting is easily doable upto 115.2kbaud. I'd do something like this: wait_for_tmr0: incfsz TMR0,W goto ready_to_send goto wait_for_tmr0 ready_to_send: movf port_shadow,W ;Start bit andlw ~(1<<bTX_PIN) movwf IOPORT CALL delay_4 iorlw (1<<bTX_PIN) btfsc tx_byte,0 ; bit 0 andlw ~(1<<bTX_PIN) movwf IOPORT CALL delay_5 iorlw (1<<bTX_PIN) btfsc tx_byte,1 ; bit 1 andlw ~(1<<bTX_PIN) movwf IOPORT ... ; stop bit CALL delay_7 iorlw (1<<bTX_PIN) movwf IOPORT return delay_7 nop delay_6 nop delay_5 nop delay_4 return 115.2 kBaud ~ 8.7uS per bit. If you dither the bit times with 8 and 9us you can average about 8.5uS per bit and be okay. If you situate your RX pin on either bit 7 or bit 0 of your IOPORT, then consider this: Search_for_start_bit btfsc IOPORT,7 goto Search_for_start_bit RLF IOPORT,W RLF rx_sample1,F RLF IOPORT,W RLF rx_sample1,F RLF IOPORT,W RLF rx_sample1,F RLF IOPORT,W RLF rx_sample1,F RLF IOPORT,W RLF rx_sample1,F RLF IOPORT,W RLF rx_sample1,F RLF IOPORT,W RLF rx_sample1,F RLF IOPORT,W RLF rx_sample1,W That's good for about 2 bits. You'd need to repeat this code (for rx_sample2,...) to get the other 8 bits. You'll average about 5 samples per bit and need to then perform some clever bit decoding. It's not a trivial piece of code (and I ain't got the time to write it right now!). However, this snippet might be useful incf x,W ; clear consecutive 1's andwf x.W ; e.g. if x == 0xf7, then after these two instructions x will equal 0xf0. Scott -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

> Before I go off and waste a bunch of time trying to do the impossible > (but hey it may be fun), does anyone have any advice on what is the > fastest half duplex rs323 comms I can expect to get using software only > on a 4MHz 16f876 (using a resonator not a crystal)? I once did a 9600 baud software UART on a 16C62A running at 160KHz. That was pushing the limit because it worked for transmitting, but receiving was unreliable because of the measurement jitter on the leading edge of the start bit. Scaling that up to 4MHz oscillator comes out to 240Kbaud. That means 115.2Kbaud should be quite doable assuming your oscillator is accurate enough. My code required constants for the oscillator frequency and the desired baud rate, and a bunch of assembly time computations automatically produced the code. ***************************************************************** Embed Inc, embedded system specialists in Littleton Massachusetts (978) 742-9014, http://www.embedinc.com -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email spam_OUTlistservTakeThisOuTmitvma.mit.edu with SET PICList DIGEST in the body

> Oh and if anyone is interested I'm using this to > control a home built cnc machine If this is a one off, use a PIC with two UARTs or an external UART, or even something like a 16F628 talking to the main PIC and acting like an external UART. > ... Take a look at http://www.gloomy-place.com/cnc.htm Pretty cool! ***************************************************************** Embed Inc, embedded system specialists in Littleton Massachusetts (978) 742-9014, http://www.embedinc.com -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email .....listservKILLspam@spam@mitvma.mit.edu with SET PICList DIGEST in the body