NetNews Offline 2

home *** CD-ROM | disk | FTP | other *** search

/ NetNews Offline 2 / NetNews Offline Volume 2.iso / news / comp / dcom / modems-part1 / 9068 < prev next >

Wrap

Internet Message Format | 1996-08-05 | 19.8 KB

Path: EU.net!sun4nl!stout!gerard From: gerard@sbu.stork.nl (Gerard van Draanen) Newsgroups: comp.sys.palmtops,alt.comp.hardware.homebuilt,comp.dcom.modems,comp.os.msdos.programmer,comp.sys.handhelds,comp.sys.ibm.pc.hardware.comm,sci.electronics Subject: RFC -- Input IR-RC on a (Sharp PC300) PC ; remote control Keywords: Sharp, PC3k, IRDA, IR, RC, Assembly, Req for comment, LED, Sample, Bit Message-ID: <3021@stout.UUCP> Date: 25 Mar 96 12:44:51 GMT Followup-To: comp.sys.palmtops Organization: Stork Bepak B.V., Utrecht, NL Hello Netters, I'm currently working on getting my Sharp PC3000 Handheld computer to be able to send and receive via an IR interface. TX/RX for (regular) DataComm.... CTS/RI for Bitstream information. I could use a little help on a couple of subjects..... - How to properly setup some electronics around the IR-receiver, for it to be able to capture ~40 kHz IR signals into binary (+or -10V ) signals suitable for detection on a PC's rs232 port. - How to optimize the (sample) assembly language program I made, so it becomes more useful... And maybe.... one of you people has done all this before... If so, I'd be glad to hear about the availability of such 'sample' software. The idea, is that I'm doing a 'raw' sampling and storing, with oversampling to be able to repeat the IR signal through output across a IR-Led in parallel to the PC's speaker. Here's the information, and analysis so-far. Feel free to add to this, or to give me pointers on possible assy language improvements..... Let's get this stuff going... my preliminary test are very hopeful.. Thanks in advance for your further input, Gerard. =================== _____ Mngr.Tech.Autom. / ============================== Gerard van Draanen. / _ _ _, __ _, __/ How come, after all books read gerard@sbu.stork.nl / / /-' / /_/ / /_/ You still can't go by the book =================== ~~~~ `~ ~ ~ ~ ~ ============================== -- Gerard van Draanen -- IR.TXT -- -1- '960308 '960325 Wireless communication between (PC) equipment is being used more and more. A number of new PC's are supplied standard with IRDA interface. There's a possibility, with relavively simple H/W adjustment to create such a communication link without cable. Esp. in an industrial environment, an IR communication can be an alternative to reduce on cabling (and installation) cost in a couple of situations. Simply put, the wireless comm. comes down to an IR-led & IR-receiver. In some cases, 1-way traffic will suffice. (Remote-Control of auxillary equipment, displays, Barcode-lezers, TVs etc. ) A sender/receiver thus has 2 specific types of input/output: - Bitstream ... Data to be sampled without further interpretation . User assigns 'logical meaning' to a sample .. Sample can be repeated as output -- on a different location -- on a different time to get a reaction from equipment, equal to a situation of sending the original sample directly. .. Multiple samples can be combinded, or timed, to make more more complicatied input/output situations. - Datacomm ... (ASCII) data is transferred, WITH its usual timing / meaning. There are a couple of ways to go about sending and receiving the signals... - Dedicated external Harware, capable of 'understanding' bitstream samples - Non-dedicated hardware, but extensive s/w to 'understand' the sample - A 'stupid' simple device, giving bitstream to be repeated I'll go for the 'stupid' device... I don't want to add external h/w to achieve my intended purpose here... Basic possibilities are IR-Led in parallel to speaker or on RTS / DTR output of Serieal port and/or TX output of Serial port or on one of the 'OUT' bits on the Parallel port IR-receiver or RI/ DCD/ CTS/ DSR from Serieel poort and/or on RX input of Serial port. or on one of the 'IN' bits on the Parallel port. A basic need, is to create a fixed sample rate, not related to PC speed. The baudrate generator in a PC can function perfectly for this.. But... the sample rate needs to be in the 80--120kHz range (oversampling), and that needs a bit of extra thought on the PC's end. For Datatcomm, using TX/RX on the serial port should suffice, without further software adjustments. This will enable operation similar to a 3-wire connection. Note:One (1) sender can reach multiple receivers The TX/RX comm can not be used for a Bitstream, since the 'framing' information will interfere with the signal to be sent. The original equipment will not have facilities to strip or add these bits. In the 'BitStream' case, signals from RTS (out) and RI (in) pins are preferred. /O. Gerard van Draanen -- IR.TXT -- -2- As background information I've studied a number of magazines and books. I don't have (any) understanding of the IR-RC basics, other than what I've read from these articles... References: Ct magazine Nr7 + Nr8 '91 Elektuur 4-'94 en 11-'94 Subj. IR-RC op PC, en Philips RC5 codes And for some detail info on the PC's serial port and h/w configuration: Crafting C tools for the IBM PCs ISBN: 0-13-188418-2 025 (Joe Campbell, Prentice hall) The IBM PC & PS/2 ISBN: 1-55615-131-4 (Peter Norton, Microsoft Press) Note:Rohm Electronics GmbH; Karl-Arnold-Str 15;47877 Willich-Munchheide; Duitsland... Tel +49-2154-9210 (Fax : +49-21544-9214-00) Has an IR (IrDA1.0) compact IC for Send/Rcv+Modulatie: RPM 800CB 2k4 t/m 115k2 Transfer rate in mini housing..(12*5.5*4.5 mm) An IR-RC signal is generally made-up out of a modulated base frequency in 31, 38, or 40 kHz range. '0' & '1's are then 'Gated' on the base frequency. The setup of the Data-signal is... ---init----[pause1]___data_____ or ---init----[pause1]...header...[pause2]___data___[pause3] the possibly repeating part is : .................. So, there is a difference in 'one-time' signals and signals being repeated until the 'button' is released. The signal content is Factory specific. Numerous codes, combinations, timings etc. exist for all the various manufacturers. ========== Some background info on the Alarm timer, Serial port and Sound/Speaker: ========== Alarm Interrupt (vector 4A) is at 0:128 in byte-reversed format v.b. PC3k: 02 2f C4 25 So, its a reference to an alarm-handling function at 25C4:2F02 Such a routine can be used to -timed- send a specific IR-RC code. Further interaction with this service requires some further study. A possible base for time can be the built-in 8253 programmable interval timer: Ch-0 = 18.2/sec PC clock timer output Ch-1 = Ram-refresh Ch-2 = Speaker frequency <== Basic freq via programmable devider. Ch-3 = ?, often via NMI Ch-2 can be used for Sound-generation (and -thus- for IR led control) 8253-5 Input signal is 1.19xxx MHz, output is via Devider Programing the Devider: OUT 43, B6 <- Prepare 8253 to accept devider OUT 42, 0A <- Low order OUT 42, 00 <- High order Will giva ca. 100kHz signal to Sound-Amp That signal .... we should be able to pick up OUT43, OUT42 8253-5 --> Ch-2 ----> --> Speaker OUT 61, xx --> Bit 0= pulses on Ch-2 to speaker Bit 1= 1of0 setting of Speaker bit INP 61 AND FC (As to avoid damaging other INP61 bits) ... IF it would be possible to READ the current state of the Ch-2 status, THEN that state could function as a timing signal.... However... I have found no possibility to READ the signal state. Gerard van Draanen -- IR.TXT -- -3- ========== As an alternate timing source, we can use the PC's baudrate-generator in the PC RS232 poort. PC adress 0:400-401 ==> COM 1 adress (Normally $3F8) ==> UART-BASE 0:402-403 ==> COM 2 adress UART-BASE +0 = TX/RX byte buffer en 0 = BAUD_MSByte... Before adressing this... first 1 = BAUD_LSByte UART-BASE+3 -> MSbit=1 Devid.0==> 115k2 baud en (So ca. 10k CPS) 7654 3 2 1 0 1 = INTR-Enable 232 Stat TX TX Empty Ready (With this, an INT routing can be calle at each Byte TX/RX ) 76543 21 0 2 = INT-ID xxxxx 00 = 232 Input INT-pending 01 = TX 10 = RX 11 = Serialization error (Crafting C Tools... pag 379 for INT => Char->Buffer ) 3 = DATA-FORMAT 7 6 5 4 3 2 1 0 Baud Set Stick Even Parity Stop Word Latch Break Parity (0=Odd) Enable bits Length 5-8 765 4 3 2 1 0 4 = CONTROL Loop !IRQ User RTS DTR Test 7 6 5 4 3 2 1 0 5 = SIO-STAT x XMIT XMIT RX'd Frame Par. Over Data SHreg Hold a Error Error Run Ready Empty Empty BREAK (Ready to send when SIO-STAT = "0010 0000" ) 7 6 5 4 3 2 1 0 6 = STAT-232 DCD RI DSR CTS DELTA DELTA DELTA DELTA DCD RI DSR CTS (Reading STAT-232 resets the DELTA.. [latch sig. change] ) Now, by using SIO-STAT while sending and receiving , a fixed time-base for the bitstream sample can be achieved. Per dataBIT on the TX line, a corresponding BIT on the CONTROL line (CTS) can be sent. So the BIT rate on the Serial connector is needed. _____|~|XXXXXXXXp|~|__|~|xxxxxxxxp|~|____ Data Stop Data Stop Start Par Start Par ____ = Time available for setup of next byte When a fixed 01010101010101 pattern can be achieved, than, that signal can be fed-back to one of the pins, and the next BIT on the IR-led can be set up. Sending "01010101" with EVEN parity (results in 0) & 1 stop-bit will give.. 10101010101_10101010101_10101010101_ on the TX line.... SO (1) (1) (1)on the XMIT Hold empty reg XMIT Hold empty can function as a trigger for the (same) next Byte for the TX line.... On each -1-0-1- transition of the TX line, a corresponding bit from the bitstream can be handled. Gerard van Draanen -- IR.TXT -- -4- In principle, storing the sample Bitstream data can be done with simple Assembler SHL, RLC, STC, CLC, SAL & JC [decr_ptr] assignments.. By first shifting a '1', then... when this first bit rolls out into the 'Carry' bit, this can function as a trigger to decrement the byte-pointer.. So... not much overhead is needed to store the binary data [ ] [ ]1[10011001] [10000111] 0 <==== First set of 8 samples Determ. Fill ^ Second set of 8 samples Max.len dire- ^ Carry bit ==> Decrement byte-pointer Sample ction Current position byte-pointer The very first bit in a sample is always a '1', and its presece activates the further sampling. At 100kHz we have ca 10 microseconds for data handling...so... not much. After a sample has been taken, the first valid (#'0') postion will determine the length of the sample to be stored. Possible H/W electronic.. Z?SState-relais .. or manual =-----3CD RTS Datacomm. RTS v: TX-> IR-led IR-Led @Y DTR->IR-Rcv->RX <==Rcv op RX =|<--o\ o DTR (equal to 3-wire) \o TX IR-Rcv Bitstream RTS ^: TX-> RI (Timing) +->|-o\ o CTS DTR-> IR-led | \o RX RTS->IR-Rcv->CTS <==Rcv op CTS | +----o\ o RTS Note: The IR receiver needs a bit of electronics to \o DTR amplify/digitize the incoming signal. Timing +-ZDDD?-- TX RI+ SIO_STAT( Xmit-empty) gives timing for Send/Recv | @DDDY by setting up a fixed 10101010 pattern on TX at 115k2 +->|----- RI Because the RI (ring indicator) is seldomly used, this part of the h/w can be put INSIDE of the PC. This allows the detection of a timing signal without external hardware. When the IR-led is in parallel to the speaker, than that can be used for Sending information. With all this background information... I've set up an assembly language test program, and tried that in debug. (my irgt.com uuencoded-included). I connected the TX line to the RI line, and alternated the CTS with that connection. It seems to work quite nicely, I get anticipated results till 57k6, so I guess by assembly routines could still be a bit quicker or the way I had set up for testing was not optimal for the full 115k2 sampling rate. So... I'm possibly in need for someone to optimize the following ir.asm assembly programming... I may have been quite 'clumsy' with my selection of instructions. This is due to lack of experience in the assy-programming field. ;Gerard van Draanen -- IR.TXT -- -5- ;Logical setup / flow: $3F8 = SIO_BASE SIO_TX_RX =0 SIO_BD_MSB=0 SIO_BD_LSB=1 SIO_DA_FMT=3 SIO_CTRL =4 SIO_STAT =5 SIO_STAT_I=6 RX_MASK =1 TX_MASK =$20 RI_MASK =$40 CTS_MASK =$10 ;CLEAR A 16k+$FF BUFFER AT _OFFSET ;Debug instruction: FDS:2000L5000 00 ;Variables at fixed (offset) adress $2000 [SI] _OFFSET : MOV SI,_OFFSET DW 0000 ; SAFETY=0 at SI+0 and SI+1 DB 00 ; OLDRI=0 at SI+2 DB 00 ; 1ST1=0 at SI+3 DW 0000 ; SAMPSIZE=0 at SI+4 and SI+5 MOV BX,4FFF ; Max buffer length (at SI + $FF ) MOV CX,0 ; Timeout for 1st '1' sample (Loop to RD_RI ) ;==== ;Setup 115k2 rate , 1 stopbit, 7 databits, Even parity for 10101010101 stream ;==== ;Read status, set Bit7, load Baudrate, clr Bit7 mov dx,3fb in al,dx ; ; Check if you like ; Latch bit on mov al,9a out dx,al mov dx,3f8 ; LSB Baud ; xx update for testing 60=1200bd, 0c=9600bd, 06=19k2 01=115k2 mov al,01 out dx,al inc dx ; MSB Baud mov al,00 out dx,al mov dx,03fb in al,dx ; Latch bit off and al,7f out dx,al ;==== RTS_nDTR = 0A ; No interrupts, RTS^ DTRv ;==== SET_SIO:MOV DX,SIOBASE+SIO_CTRL MOV AL,RTS_nDTR OUT DX,AL ; ;Gerard van Draanen -- IR.TXT -- -6- ; TX_RDY: MOV DX,SIOBASE+SIO_STAT ; Test --ready for xmit IN AL,DX AND AL,TX_MASK CMP AL,TX_MASK JNZ RD_RI TX_55: MOV AL,$55 ; Xmit $55, creating 101010.. stream MOV DX,SIOBASE+SIO_TX_RX OUT DX,AL RI_0: MOV AL,0 MOV [SI+2],AL ; 1st RI state will be '0' wait for '1' PUSH AX SAFETY: POP AX MOV AL,1 SUB [SI+0],AL ; but don't wait forever JNZ RD_RI MOV AL,1 SUB [SI+1],AL JZ EXIT_RI RD_RI: MOV DX,SIOBASE+SIO_STAT_I IN AL,DX PUSH AX AND AL,RI_MASK CMP [SI+2],AL ; Compare to OLD RI state JE SAFETY MOV AL,0 MOV [SI+0],AL ; Restore safety MOV [SI+1],AL RD_CTS: POP AX ; Get it again PUSH AX AND AL,RI_MASK MOV [SI+2],AL ; Save Current RI state POP AX TST_CTS:AND AL,CTS_MASK ; Test CTS CMP AL,CTS_MASK JZ STORE_1 ; 1 MOV AL,0 ; Test if any '1' already seen CMP [SI+3],AL JNZ STORE_0 ; 0 LOOP RD_RI JMP EXIT_TIMEOUT ; Still no '1' seen after $FFFF samples STORE_1:MOV AL,1 MOV [SI+3],AL STC ; 'Carry' STORE_0:MOV AL,[SI+100+BX] ; Get buffer contents RCL AL,1 ; Roll into Buffer at [BX] PUSHF MOV [SI+100+BX],AL POPF JNC TX_RDY ; High bit rolled out ? (1st '1' sample) MOV AL,1 ADD [SI+4],AL ; Another Byte-full of samples taken JNZ NEXTB ADD [SI+5],AL NEXTB: DEC BX ; Update pointer to current sample JNZ STORE_1 S_DONE: NOP ; Sample completed NOP ; Normal completion ... Sample in $2000+100 -- +100+4FFF EXIT_RI: ; Had to wait too long for RI to change state => No Freq. avail EXIT_TIMEOUT: ; Had RI, but still no '1' sample after $FFFF tries => Empty ---- I typed this in as assembly into a debug session, and tested that.. with 'proper' results on the sample till some 50k sample rate.... So it needs a bit of optimization to be able to do 115k2 bit sampling.. ---- Here's the uuencoded irgt.com --- begin 644 irgt.com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end ------ Hoping to get some respons on this..... Thanks in advance, Gerard.