SERIN & Flow Control Strategery
PJAllen
BannedPosts: 5,065
· I want to receive data via SERIN and stash those bytes into an I2C EEPROM, on-the-fly.· I·can accomplish·this in hunk-and-peck·fashion, with or without flow control,·but I fall short sending several bytes in succession.· I'm trying to send this data out as a text file.·
· The I2C writing operation takes·'a lot' of·time and·to accomodate that I need to signal DTE that the SX is busy.· So, I thought that I'd try setting a pin as "CTS" output.· This got me a little further than having no flow control at all, but not by much.
· I'm only going 1200bps, so I figured that with 28_000_000Hz I could zip off the CTS in time, but it looks like it gets tangled up anyway.· I suspect that "CTS" doesn't drop fast enough.
· Is there an internal counter that could be monitored and used to de-activate "CTS" when the Stop Bit starts?· ·
· [noparse][[/noparse] I'm using a MAX232, my states are correct. ]
· The I2C writing operation takes·'a lot' of·time and·to accomodate that I need to signal DTE that the SX is busy.· So, I thought that I'd try setting a pin as "CTS" output.· This got me a little further than having no flow control at all, but not by much.
· I'm only going 1200bps, so I figured that with 28_000_000Hz I could zip off the CTS in time, but it looks like it gets tangled up anyway.· I suspect that "CTS" doesn't drop fast enough.
· Is there an internal counter that could be monitored and used to de-activate "CTS" when the Stop Bit starts?· ·
· [noparse][[/noparse] I'm using a MAX232, my states are correct. ]
FOR addr = 1 TO 72
CTS = 0 ' CTS active
SERIN serial, T1200, outByte
CTS = 1 ' CTS inactive
EEWrite addr_LSB, addr_MSB, outByte
NEXT
Comments
-Phil
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!
I've been looking at the Assembly generated by SX/B:
I see that CLRB RC.4 is ASM for RC.4 = 0 and SETB RC.4 is ASM for RC.4 = 1.· Where would be the best places for them if they set/drop the CTS output?· Which ASM line detects the start bit·and which assumes the stop bit?
JB RC.7, @$
... is what is waiting on the start bit; the stop bit is ignored and the program moves the work value in __PARAM1 to the output during the stop bit timing (I think the purpose of stop bits is to give the receiver time to do something with the byte it just received).
I'll·see how·making CTS active immediately before that goes, keeping the CTS drop-out where I placed it in my preceding.
Maybe I'm asking too much of RTS/CTS by trying to get DTE to stop&go like this (haltingly.)
Can I have the program·go with·the data-input/UART as an INTERRUPT-based routine at the beginning and then drop into a non-INTERRUPT operation?
I'm a little surprised that it's being so finicky at 1200bps.· It fills up VAR arrays·like crazy, but introducing the flow control as a break in order to transfer the RX'd data into an I2C gums up the works.· SimpleTerm may be real tight on stop-bit time/s (no "stop-bit and a half.")
What would happen if PJ were to use his original code, with or without a small buffer, but use an SX 48 running at 75 mHz - without Interrupts? Would the handshaking perform better / correctly?
PJ, doing interrupts is pretty easy actually (until you have to deal with timing issues :-| There's plenty of post and sample code. Check out some of the NV articles... it's well worth learning 'cause you can do such cool stuff with it. Actually, the background buffering of serial data streams was one of the earliest uses of interupts!
cheers,
Howard in Florida
PS. the level of mutual support here is incredible! Thanks all very much...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Buzz Whirr Click Repeat
· I don't need an SX48 to run a faster crystal/osc.· I have a 50Mc osc which I may pull out.· Maybe I could hit it with more speed.· I don't know.· Maybe I could modify the ASM for 300bps.
· Anyway, I think I'm going at a pretty good clip right now, but I'm missing every other character with my present program.·
· (My terminal program has a modified setting, "Data Send Block Size 1 -- Best response to flow control, slow data transfer rate")·
· At 1200bps, a bit is 833usec.· However long it takes for CLRB to execute (my big assumption) can't be a meaningful fraction of that 833, it should nail it.· Maybe I can place my drop CTS earlier.
· I only have a Blitz and, so, I program the SX, reconfigure and send data, pull the eeprom and read that with a eprom programmer.
· Fun, fun, fun.
CRP,
· Hey, your avatar -- that's the apparatus Spock·was making·on "The City on the Edge of Forever."
There's more to this than the serial-EEPROM trip, so there would be timing considerations later, I'm sure.
In the end, I don't think you're going to get what you want attempting to crowbar CTS control into an inline SERIN.
PJ, it's not impossible to do what you want without interrupts; but, trust me, it's a whole lot easier if you use them. Just think of it as part of your growth as a programmer.
-Phil
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!
You can have a 16 byte (or even 32 byte = 1 page on an EEPROM ) buffer if you take the plunge into interrupts (w SX/B and the many examples of serial ISR ASM routines at the forums it's not too hard). The high-level SX/B commands that are affected if you use an ISR are the internally time-dependent code structures, e.g. PAUSE, PAUSEUS, SERIN/SEROUT, PULSIN/PULSOUT, FREQOUT. Any commands that are clocked are unaffected (even if in an interrupt happens in the middle of a SHIFTIN or I2CRECV, it just means the clock pulse might be slightly longer/shorter).
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST
1uffakind.com/robots/povBitMapBuilder.php
1uffakind.com/robots/resistorLadder.php
I'm not attempting to crowbar...·I'm attempting to shoe-horn... it's different.·
Too much fun.
As an alternative, could you come up with some sort of foreground protocol where the PC will only send a character at a time as the SX system asks for it via a serial command? I've used such arrangements with BASIC Stamps, which don't have background capability, and they work reliably. You could still use CTS if you want, except that the PC program would read it as a bit input and send one character, wait for CTS to go low, then high, send another character, etc.
-Phil
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!
If anybody could help explain this a bit,
Given:
ASM MOV FSR,#__TRISC MODE $0F SETB IND.7 MOV !RC,IND BANK $00 JNB RC.7,@$ JB RC.7,@$ MOV __PARAM2,#8 MOV __PARAM3,#2 MOV __PARAM4,#76 DJNZ __PARAM4,@$ DJNZ __PARAM3,@$-3 NOP NOP MOV __PARAM3,#3 MOV __PARAM4,#151 DJNZ __PARAM4,@$ DJNZ __PARAM3,@$-3 JMP @$+2 RR __PARAM1 MOVB __PARAM1.7,RC.7 [color=red] DJNZ __PARAM2,@$-17[/color] MOV outbyte,__PARAM1 ENDASMDoes the -17 in DJNZ __PARAM2,@$-17 mean that it jumps back 17 program lines?· Where is it jumping to, BANK $00?
·
Here's the interpretation:
ASM
MOV FSR,#__TRISC ;point to the (shadow) memory location that holds the direction register for port C... located in bank F on an SX28
MODE $0F ;set the MODE to accept changes to the SX direction reggisters
SETB IND.7 ;prep bit 7 of the shadow register to input direction
MOV !RC,IND ;now transfer the shadow to the actual direction register inside the SX
BANK $00 ;return to the default bank
JNB RC.7,@$ ;loop here, continually testing the serial line for a a low, indicating no transmitter is connected
JB RC.7,@$ ;loop here continually testing for a high, indicating a transmitter is connected, and in the serial STOP condition
MOV __PARAM2,#8 ;a low now indicates a falling start bit edge has been detected, set the bit counter to 8 bits
MOV __PARAM3,#2 ;now load the some parameters to do some time loops (duration depending on baud rates)
MOV __PARAM4,#76 ; to sample the second bit (first data bit after the START bit) at its mid point..... one and one half bit lengths
DJNZ __PARAM4,@$ ;and do the timing here
DJNZ __PARAM3,@$-3 ; with two nested loops.... outer loop is 2 times the inner loop of 76, each of 5 instruction cycles (jump instruction itself is 3 cycles plus the @ page insruction plus the dec-skip-if zero)
NOP ;some fine tuning of the timing
NOP ;to get the sampling interval just right
MOV __PARAM3,#3 ;now load the parameters for timing duration to sample all following bits in their middle; relative to the middle of the first data bit ...... one bit length
MOV __PARAM4,#151 ;outer loop is 3 times the inner loop of 151 loops of 5 instruction cycles each
DJNZ __PARAM4,@$ ;and execute the delay here with
DJNZ __PARAM3,@$-3 ; again two nested loops
JMP @$+2 ;this is a little weird but is just a 4 instruction time delay... effectively not really a jump in program sequence
RR __PARAM1 ;move the shifter along one bit and make room for the new bit coming in
MOVB __PARAM1.7,RC.7 ;merge the sampled bit into position 7 of the shifter.... the previous value is overwitten
DJNZ __PARAM2,@$-17 ;decrement the bit counter, and if it is not zero (indicating it has done 8 bits) jump back 17 to load up the second set of delay loops again to sample in the middle of the next bit
MOV outbyte,__PARAM1 ;copy the received result to the destination variable
ENDASM
PJ... the $ indicates the present value of the program counter, and the $+ or $- values are relative to that point. The $-17 you referred to is actually 17 instructions... that is program LOCATIONS. The assembled code DJNZ are compound instructions consuming 2 bytes each, and then there is the @ page instruction on top of that. So, no, it does not point back to the BANK which is 17 LINES back, but instead it points 17 SINGLE INSTRUCTIONS back to the load of the timing parameters.
Hope that helps.
Cheers,
Peter (pjv)
PJ -- if you view the "list" output of a compiled SX/B or src program, you will see the hex addresses for actual line instructions on the left. This can be useful if you are trying to deconstruct the assembly of high level commands, esp. in the case of "compound" assembly commands like DJNZ, which expand to much more than a single instruction line. Also, the SX-Key reference has a nice little summary of assembly commands organized by single-word and multiple-word instructions (a "word" in this case being a single instruction).
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST
1uffakind.com/robots/povBitMapBuilder.php
1uffakind.com/robots/resistorLadder.php
> CRP,
> Hey, your avatar -- that's the apparatus Spock·was making·on "The City on the Edge of Forever."
PJ, yes - that's my favorite scene of all Star Trek... oh, it's probably copyrighted material, but for educational purposes, it's ok to use. Educational as in "here's what not to do when controlling vaccuum tube circuits with sophisticated microcontrollers."
; --- end disclaimer ---
IRET
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Buzz Whirr Click Repeat